Elbieta Adamczyk, Adam Mickiewicz University, Pozna

Arjen Versloot, University of Amsterdam

A MULTIVARIATE ANALYSIS OF MORPHOLOGICAL VARIATION IN PLURAL INFLECTION IN
NORTH SEA GERMANIC LANGUAGES
The focus of the present study is on the diachronic development of irregular plural formations
in two representatives of the North Sea Germanic branch, i.e. English and Frisian.Old
Germanic languages exhibit a wide range of plural formation patterns, inherited from IndoEuropean and subsequently reshaped in Proto-Germanic, e.g., Old English:dg dagas
days, cealf cealfru calves, ft ftfeet.A salient feature of the declensional system of
Old Germanic languages was an increasing syncretism of the inflectional classes, resulting
from an interplay of phonological and morphological developments (phonological reduction
and analogical levelling). These tendenciesinduced a reduction in the diversity of the inherited
inflectional exponents, including those marking specifically the plural. In effect, the
distribution of the plural markers, which remained largely lexical,was partly reordered by
phonological andsemantic criteria (cf. Dammel, Krschner & Nbling 2010).
The mechanism of the remodelling of the Germanic plural paradigms entailed a
number of interacting factors which contributed to the shape of the nominal inflection as it is
known in Modern English and Frisian.The aim of the present study is to identify and weigh
the significance of the factors which contributed to the preservation of irregular plural patterns
in these languages. Theinvestigationinvolved two steps: (a) identifying the persistence of
individual irregular plural forms and (b) applying a multivariate analysis to define the weight
of the different factors contributing to the preservation of irregular plurals.The North Sea
Germanic languages were chosen as the testingground for the analysis, since they depart from
a relatively late common starting point, being nearly identical forthe reconstructed ProtoEnglish and Proto-Frisian.
The factors which will be considered in order to account for the attested plurality
patterns include the following:
(a) frequency of occurrence
(b) morphological salience of plural marking
(c) interaction with default plural endings
(d) prosody
(e) morpho-phonological complexity of the plural marking
(f) semantics
(g) language contact
The interplay of factors can potentially lead to three outcomes for the development of
irregular plural forms, namelya word may (1) preserve the historically developed plural form,
(2) become regular or (3) develop a new irregular form.The present study presents the
outcome of a multivariateanalysis, wherethe factors are quantified and their correlation with
the dependent variable of (type of) irregularity is computed (cf. Sutter 2009).
References
Dammel, Antje, Sebastian Krschner & Damaris Nbling. 2010. Pluralallomorphie in zehn germanischen
Sprachen; Konvergenzen und Divergenzen in Ausdrucksverfahren und Konditionierung. Kontrastive
germanistische Linguistik, vol. 2, 587-642. (Germanistische Linguistik 206-209). Hildesheim [etc.]:
Olms.
Sutter, Gert. 2009. Towards a multivariate model of grammar: The case of word order variation in Dutch clause
final verb clusters. In Andreas Dufter, Jrg Fleischer & Guido Seiler (eds.), Describing and Modeling
Variation in Grammar, vol. 204, 225-254. Berlin, New York: Mouton de Gruyter.

Old English Possessed Body Parts in Contact

Cynthia L. Allen
Australian National University
A striking syntactic change which took place in Middle English is the loss of the
Dative External Possessor (DEP) found in Old English examples like scearpa him a
scancan (Balds Leechbook) scarify his legs, literally, scarify him the legs. In Early
Middle English, the Internal Possessor (IP, used in the translation of this example)
increasing became dominant, and the External Possessor is limited to a few fixed
expressions such as look x in the eye and look x in the mouth in Modern English.
Ahlgren (1946) suggested that the loss of the accusative/dative case marking
distinction is the key to the loss of this construction. This explanation does not fare well
under close scrutiny and explanations in terms of contact with other languages have
been proposed as more promising in accounting for why although most Scandinavian
languages have had similar losses of case distinctions and have lost the DEP, they have
developed another External Possessor (EP) construction using a locative expression for
the possessor. McWhorter (2002) argued that contact with Scandinavian precipitated
this change, which must then be seen as an instance of simplification caused by
language contact rather than transfer of a construction from one language to another, but
others (e.g. Vennemann 2002, Filppula et al. 2008, Hickey 2012) point to the lack of
EPs in the Brythonic languages as a similarity with English which is documented as
unusual situation in Europe (Haspelmath 1999) and thus unlikely to be a coincidence.
By this view, the fact that the DEP only disappeared from texts in the Middle English
period is treated as stemming from a discrepancy between spoken language, where the
IP is assumed to have been preponderant because of Celtic influence, and the more
Germanic written language.
This paper furnishes new data which must be taken into account when evaluating
the role of contact in the loss of the DEP in English. A corpus study into the syntax of
body parts in Old English which shows variation between DEPs; both the IP and the
DEP are possible in some similar situations even in the same text. However, when we
make a distinction between DEPs involving direct argument (subject or object) body
parts as opposed to prepositional objects, we see that were already becoming more
restricted in their range at an early stage in OE direct argument body parts, being found
almost exclusively with adversely affected possessors. Both case marking loss and
contact with Scandinavian are therefore excluded as triggering the decline of the DEP,
whatever role they may have played in its eventual complete loss. Contact with Celtic
cannot be ruled out at least as a contributing factor, in this initial decline. However, if
we assume that the early favouring of the IP over the DEP in early OE writings is due to
Celtic contact, we must abandon the notion that Celiticised English was not
considered suitable English for writing in OE. I argue furthermore that in focusing on
the loss of the DEP in English, we should not lose sight of the fact that the DEP has
also changed in languages like German from the situation documented by Havers
(19011) for the early Indo-European languages generally. At an early stage, the IP and
DEP were in variation to describe essentially the same event but the DEP made the
speakers explicit focus on the fact that the whole possessor and not just the body part
was affected. In the European languages generally, the semantic distinction between the
IP and the DEP became sharpened to the point where optionality in the choice of
construction has disappeared in German, for example. In English, on the periphery of
the European Sprachbund and possibly influenced by contact with Celtic, the DEP
became an increasingly marked construction which was used to emphasise the effect on
the whole possessor, leading to its lowered frequency and eventual loss.

Slavic and the birth of Standard Average European

Henning Andersen, UVLA

The changes in Slavic verbal categories that occurred from deep prehistory until the
Early Middle Ages, encourage examination in a spatial perspective. The parallel
developments in contiguous language areas suggest widespread bilingualism and
multilingualism. This is particularly so with respect to the most recent developments,
which coincide with the demographic mobility in the period of the Great Migrations.
1. In the earliest phase, developments differ in Slavic, Baltic and Germanic,
but the outcome of these developments is the same in all three dialect groups: [i]
Inherited aspect distinctions (PIE Stative, Aorist) are lost, and what remains is a
Preterite/Present distinction. During this phase, [ii] analytic expressions of Telic
aspect (terminativity, Russ. predelnost, Streitbergs perfectiv) are innovated
(Maslov, J. S. 1959). As they are grammatized, the adverbial Telicity markers
develop from words to clitics to prefixes (Pinault 1995). Typologically the
developments in Baltic, Slavic, and Germanic are parallel.
2. In the second phase, [iii] Progressive (durative, iterative) verbs are formed
from simplex and prefixed Telic verbs to express Atelicity. Eventually [iv] Telic action
verbs and their Atelic counterparts are reanalysed as Perfective vs. Imperfective
verbs. Other Telic verbs become Perfective procedurals, and Atelic verbs,
Imperfective procedurals (Maslov 1959/2004). This is a Slavic development, but the
categories of proceduals in East Baltic, though exprressed with different morphology,
are typologically similar.
3. In the third phase, the other aspects known from Old Slavic texts are
grammatized. [v] The CS Imperfect (vs. Aorist) has a parallel in the Lithuanian
progressive Preterite. [vi] The CS auxiliated Retrospective (Pluperfect vs. Perfect)
has parallels in Balkan, Romance and Germanic languages (Drinka 2003, 2007). [vii]
The Slavic auxiliated Prospective has regional variants (Futures) whose auxiliaries
translate into the neighboring languages, Greek (will, have), Romance (have),
Hungarian (begin, take), and Germanic (shall, will; become) (Andersen 2006,
2009).
In this presentation I will discuss to what extent the development of the most
recent of these typological affinities [vvii] corresponds to known geographical
contiguities and the textual record of the languages in question.

Variationist Analysis of Phonetic Reduction in the Grammaticalized Motion Verb in Japanese

Fumihito Arai
Kobe Shoin Womens University

This study addresses the phonetic reduction in the Japanese -te-iku -Con.-go construction as in (1)
which has been grammaticalized from the full-verb iku go. Using the variationist framework, we
argue that this is a typical case of change from below, that is of language change in which the
innovative reduced form (-te-ku) is more likely to appear in informal, spontaneous speech and in male
utterances.
(1)

Ensoku-ni-wa,
takusan okasi-o
mot-te-[iku/ku] tumori desu.
excursion-to-Top. many snack-Acc. have-Con.-go
will
be
I am going to bring a lot of snacks on the school excursion.

While the semantics of -te-iku has been much studied so far (Imani 1991 and others), the phonetic
reduction concomitant with the grammaticalization of -te-iku has been largely ignored, except for
Matsumura (1998) and Shibatani (2007). Matsumura (1998) uses a bibliographic survey to show that
the form with the reduced vowel was already present in Edo Japanese, the older form of modern Tokyo
Japanese. Shibatani (2007) argues that the reduced form is more likely when the link between the -te
conjunctive form and its preceding verb is semantically less congruous.
However, since the two forms (unreduced/reduced) are still in competition in present-day Japanese,
this case should be treated as linguistic variation governed by multiple factors in accordance with
Weinreich, Labov & Herzogs (1968) principle of language change and not by just a single factor as
presumed by Shibatani (2007). Thus, the present study aims to elucidate how the phonetic reduction in
-te-iku can be explained from the variationist viewpoint.
We examined 4,029 tokens (2,972 of -te-iku and 1,057 of -te-ku) amassed from the Corpus of
Spontaneous Japanese (NIJL 2004), and set three linguistic factors (verb length, affirmative/negative
context, verb frequency) and eight social factors (birth year, gender, geography, education, speech
spontaneity, speech style, speech skillfulness, speech experience) as possible predictors of the vowel
reduction in question. The result of a multivariate analysis shows that the innovative -te-ku form
appears at a higher probability in informal and more spontaneous speech, and it is preferred by men.
This indicates that the -te-iku/-te-ku variation is a typical case of change from below.
Language-internally, -te-ku is preferred in affirmative contexts, which supports the hypothesis that
language change behaves conservatively in negative contexts (Givn 1979). The rate of the incoming
variant in relation to verb length apparently shows the effect of this linguistic factor, where the vowel
reduction is most compatible with a bimoraic verb. However, looking closely, it behaves as such
because the bimoraic verbs appear most frequently in the data. The pattern of relative frequency of
-te-ku in terms of verb length is consistent across speaker age groups. It follows that this vowel
reduction is in conformity with the principle of lexical diffusion which states that high-frequency words
are more susceptible to sound change (Hopper 1976).
References
Givn, Talmy. (1979). On understanding grammar. Orland: Academic Press.
Hopper, Joan. (1976). Word frequency in lexical diffusion and the source of morphophonological change. In Christie, W.
(ed.), Current Progress in Historical Linguistics, 96-105. Amsterdam: North Holland.
Imani, Ikumi. (1991). V-tekuru to V-teiku ni tuite. Nihongogaku 9, 5: 54-66.
Matsumura, Akira. (1998). Zho edogo tkygo no kenky. Tokyo: Tokyodo Shuppan.
NIJL. (2004). Corpus of Spontaneous Japanese. Tokyo: The National Institute for Japanese Language.
Shibatani, Masayoshi. (2007). Grammaticalization of converb constructions: The case of Japanese -te conjunctive
constructions. In J. Rehbein, C. Hohenstein & L. Pietsch (eds.), Connectivity in Grammar and Discourse, 21-49.
Amsterdam/Philadelphia: John Benjamins.
Weinreich, Uriel, Labov, William & Herzog, Marvin I. (1968). Empirical Foundations for a Theory of Language Change. In
W. Lehmann & Y. Malkiel (eds.), Directions for Historical Linguistics, 95-195. Austin: University of Texas Press.

Xavier Bach & Louise Esher

Research Centre for Romance Linguistics, University of Oxford

On the paradigmatic status of infinitives in French and Occitan

Infinitival forms are typically presented as forming part of the inflectional paradigm of the verb, yet
the reasons for this choice rarely receive explicit consideration, much less justification. Our study
examines the relationship between the infinitive and the inflectional paradigm in French and Occitan
(Gallo-Romance), showing that the infinitive is best considered peripheral to the paradigm.
In Romance, the infinitive is certainly felt to be representative of the verb paradigm, witness its use
as the citation form. One may further note the presence of infinitives in periphrases expressing tense,
voice and aspect (categories prototypically associated with the verb), whether interchangeably with
finite forms, as in the compound perfectives (jai fait; avoir fait), or alongside finite forms, as in the
periphrastic future (je vais chanter). However, as Maiden (2011:208-9) observes, the morphological
properties of the infinitive in Romance differ substantially from those of finite wordforms: the
infinitive bears chiefly lexical information, and does not itself carry marking for tense, person,
number or mood. Maiden suggests that speakers are hence unlikely to analyse the infinitive as
composed of a lexical stem and inflectional endings, but rather as a single, atomic unit. This item is,
unlike other forms of the verb, readily converted into a verbal noun (le savoir, le manger).
If the infinitive retains any significant link with another area of the paradigm, it should be with the
synthetic future (SF) and synthetic conditional (SC), as these forms derive historically from parallel
periphrastic constructions infinitive + HABEO, etc. and infinitive + HABEBAM, etc. respectively. Yet
examination of regional variation in Occitan and of native speakers spontaneous errors in French
indicates that there is no longer any link between the infinitive and the SF/SC.
In Occitan, the infinitive never serves as a template for analogical remodelling of other parts of the
paradigm, including the SF/SC; where the SF/SC stem appears remodelled on the stem found in the
infinitive, this stem is never unique to the infinitive but always found elsewhere in the paradigm;
indeed, there are no cases where the SF/SC stem is uniquely spread to the infinitive, or vice versa.
The infinitive typically resists changes general to the paradigm (levelling, shift of conjugational
class) and may undergo idiosyncratic changes distinct from those affecting the rest of the paradigm.
In French, investigation of erroneous SF/SC forms attested on the Internet reveals that errors
replacing the SF/SC stem by the infinitive (je savoirai) are confined to L2 learners1 and deliberate
playfulness.2 Native speakers spontaneous errors typically involve the stem of the 1pl/2pl.prs.ind
(je voulerai), confirming Freis (1929:168-9) finding of a general tendency to remodel verb forms on
the stem of the 1pl/2pl.prs.ind; by contrast, the infinitive serves as a template only for erroneous
infinitive forms (avoirer). Although Frei cites infinitives remodelled on the 1pl/2pl.prs.ind stem, we
have not found any such examples.
The French and Occitan data support a view of the infinitive as peripheral to the paradigm
lexically and syntactically linked to the paradigm, but morphologically distinct from it.
References
Frei, Henri. 1929. La grammaire des fautes. Genve: Slatkine.
Maiden, Martin. 2011. Morphological Persistence. In M. Maiden, J.C. Smith & A. Ledgeway
(eds.), The Cambridge History of the Romance Languages. Cambridge: CUP. 155-215.
1

Typically on language-learning websites and discussion forums.

Eg. the dialogue of the MMORPG persona Halumgalom the Gnome exclusively uses forms remodelled on the
infinitive: Ca pouvoira se vendre qui savoire. Ou sinon a faira un trophe ! (<http://www.pathfinderfr.org/Forum/yaf_postst2904p11_-Poubelle--La-corbeille-du-forum.aspx?=>, 20 January 2013).
2

The Story of Woe

Jhanna Bardal (University of Bergen), Valgerur Bjarnadttir (Stockholm
University), Eystein Dahl (University of Bergen), Tonya Kim Dewey (University of
Bergen), Thrhallur Eythrsson (University of Iceland), Chiara Fedriani
(University of Bergamo), Thomas Smitherman (University of Bergen)

In contrast to the consensus in the field that historical-comparative syntactic

reconstruction is a futile and unavailing enterprise, we aim to show how syntactic
reconstruction may be fruitfully carried out with the aid of Construction Grammar
where complex syntactic structures are regarded as formfunction pairings (cf.
Bardal & Eythrsson 2012), here specifically using the formalism of Sign-BasedConstruction Grammar (Sag 2012).
As an exercise in syntactic reconstruction, with the goal to illustrate in detail
how such reconstruction may be carried out, we compare the syntactic environments
found for an Indo-European interjection, traditionally reconstructed as PIE *wai
woe. We investigate the syntactic environments that woe instantiates in a subset of
the daughter languages, i.e. Early Germanic, Latin, Old Slavonic, Baltic, and Old
Indo-Iranian. In addition to its bare extra-sentential use, woe also occurs
systematically with a dative subject-like argument, a construction which we refer to
as the DAT-(is)-woe construction.
Our investigation into these five Indo-European language branches reveals
that the DAT-(is)-woe construction is found in all these early daughters. The
construction was generally used when something bad happened to people, expressing
the speakers dismay. In addition to the systematic occurrences of the DAT-(is)-woe
construction, we present sporadic examples of woe with an accusative in Latin and
a genitive in Old Lithuanian.
An issue in need of explanation is related to the word order variation generally
found for this construction. It is manifested as a systematic variation in the order of
the dative subject-like argument and woe, i.e. as DAT-(is)-woe and woe-(is)-DAT.
We show that when woe is in first position it represents new information, not
mentioned earlier in the context. As such, woe in first position functions as fronted
focused material, or predicate focus in the terms of Lambrecht (1994).
Our examination suggests that the DAT-(is)-woe construction is inherited,
hence we reconstruct three constructions for Proto-Indo-European to account for the
behavior of DAT-(is)-woe: a) an argument structure construction consisting of only
one argument in the dative case, and a specific pragmatic value associated with the
construction, namely speakers dismay, b) a subjectverb construction which
specifies that the default word order in Proto-Indo-European was with the subject in
initial position, while the verb came later in the sentence (cf. Delbrck 1900), and c) a
predicate-focus construction, which places the predicate in first position, preceding
the subject, if the predicate was in focus in the discourse. These three constructions
are sufficient to capture the behavior of the DAT-(is)-woe construction across the five
Indo-European daughter languages discussed here, and hence to reconstruct the
behavior of the construction in Proto-Indo-European.
We thus show how syntactic units may be reconstructed for an earlier protostage, in this case a concrete lexically-specific argument structure construction, a
more schematic neutral word order construction and a schematic discourse-motivated
predicate-focus construction, illustrating at the same time that non-canonical subject
marking is likely to have existed in Proto-Indo-European.
References:
Bardal, Jhanna & Thrhallur Eythrsson. 2012. Reconstructing Syntax: Construction Grammar and
the Comparative Method. In Sign-Based Construction Grammar, 257308. Eds. Hans C. Boas &
Ivan A. Sag. Stanford: CSLI Publications.
Delbrck, Berthold. 1900. Vergleichende Syntax der indogermanischen Sprachen 5(3). Strassburg:
Trbner.
Lambrecht, Knud. 1994. Information Structure and Sentence Form. Cambridge: Cambridge University
Press.
Sag, Ivan A. 2012. Sign-Based Construction Grammar: An Informal Synopsis. In Sign-Based
Construction Grammar, 69202. Eds. Hans C. Boas & Ivan A. Sag. Stanford: CSLI Publications.

Syntactic and phonological evidence in favour of the grammaticalization of Northern Catalan negative poc/poca.
Montserrat Batllori & Assumpci Rost
Universitat de Girona - Universitat de les Illes Balears
montserrat.batllori@udg.edu - assumpcio.rost@gmail.com

This paper focuses on the analysis of Catalan poc/poca 'no'. Syntactically, it provides evidence in favour of the current
process of grammaticalization undergone by poc/poca and argues for its development into a head (equivalent to the negative
marker no) on the lines of van Gelderen (2004, 2008a, 2008b, 2009 and 2011). Crucially, our hypothesis is supported by the
results of a phonetic test on intonation that points out to a clear intonational difference between other Catalan emphatic
polarity particles, such as pla 'NOT', and poc.
As shown in Batllori & Hernanz (2013), in the northern region of Catalonia, poc/poca (< PAUCU, Latin quantitative adverb)
is used by some speakers as a negative emphatic polarity particle, see (1a). It coexists with poc, which still displays a
quantitative value, see (1b), but they can be easily set apart by their syntactic behaviour.
(1) a. En Pere poc ho ha fet, destudiar per a lexamen. [NEGATIVE EMPHATIC POLARITY PARTICLE]
b. En Pere ho ha fet poc, destudiar per a lexamen. [QUANTITATIVE ADVERB]

However, there is microvariation with respect to the values expressed by non-quantitative poc. Some speakers use poc, in
(1a), as a counterpressupositional emphatic polarity particle (for example, in Pla de lEstany; see Rigau 2004). Others,
though, use it as a plain negative marker without pressupositional value, especially in Girona and Figueres, see (2):
(2) a. On s en Joan? Poca ho s [= No ho s]
b. Ja ha arribat en Pere? No, poca ha arribat [= No, no ha arribat]
c. Avui poca hi anir al teatre; estic molt cansada [= Avui no hi anir al teatre; estic molt cansada]

We attribute the fact that poc/poca is used as a negative head in Girona and Figueres to a specifier to head
grammaticalization, in terms of van Gelderen (2004, 2008a, 2008b, 2009 and 2011). It is, in fact, another instance of the
negative cycle explained and widely illustrated by this author.
Our hypothesis gains further support from the fact that the negative emphatic polarity particle, pla 'NOT', despite showing a
similar behaviour concerning the licensing of negative polarity particles, see (3), still conveys a pressupositional value, is
emphatic in nature and, accordingly, not equivalent to the negative marker no, see (4).
(3) a. La Maria poc ha dit mai aix.
b. La Maria pla ha dit mai aix.
(4) Tinc por que en Joan li ho digui tot.
En Joan pla dir res. En Joan no dir res

From a phonological standpoint, we also show that poc/poca behaves as a conventional negative marker, such as no. As
known, negation sequences are comparable to declarative intonation patterns. In Catalan, the structure of the typical
declarative melodic pattern displays a descending body and final inflexion (vid. Martnez Celdrn 1994, Prieto 1999, JuliMun 2005, Font Rotchs 2007). Accordingly, a non-emphatic negative statement would accommodate to (5):
(5)

body
final inflexion/toneme

If poc/poca were emphatic, (2c), for example, it would exhibit a different pattern from that in (5). As illustrated in (6), the
body and the final inflexion of the curve would be ascending, instead of being a descending, and the final toneme would
show an abrupt descending shape (cf. Font Rotchs 2007: 118).
(6)

body
final inflexion/toneme

To carry out the phonetic test on intonation we recorded 2 speakers who were asked to produce 5 utterances containing
negative sequences with poc/poca in non-emphatic contexts and 5 more with the negative particle no. Besides, they were
asked to produce utterances with the emphatic negative particle pla. Hence, we could contrast their intonational features with
those of the statements with poc/poca. The melodic curves obtained provide us with evidence to pose that poc/poca is a
polarity head.
Bibliographical references
Batllori, Montserrat and Maria-Llusa Hernanz (2013) Emphatic polarity particles in Spanish and Catalan, Lingua.128 (May 2013): 9-30.
Font Rotchs, Dolors (2007) Lentonaci del catal, Publicacions de lAbadia de Montserrat, Barcelona.
Rigau, Gemma (2004) El quantificador focal pla: un estudi de sintaxi dialectal. Caplletra, 36: 25-54.
Van Gelderen Elly (2009) Feature economy in the Linguistic Cycle.In Crisma, P. & G. Longobardi (Eds.) Historical Syntax and Linguistic Theory, Oxford, OUP: 93-109.
Van Gelderen, Elly (2004) Gramaticalization as Economy, Amsterdam/Philadelphia, John Benjamins.
Van Gelderen, Elly (2008a) Negative cycles, Linguistic Typology. 12:. 195-243.
Van Gelderen, Elly (2008b) Where did late merge go? Grammaticalization as feature economy, Studia Linguistica. 62: 287-300.
Van Gelderen, Elly (2011) The Linguistic Cycle. Language Change and the Language Faculty, Oxford/New York, OUP.

Brace Constructions in Late Latin and Old French

Brigitte L.M. Bauer
University of Texas at Austin

In the shift from Latin to French we observe a slow but steady change in word order, which is
often referred to as a shift from OV to VO structures. It is manifest, for example, in the spread
of Verb Object sequences in both main and subordinate clauses. In addition, there is another
related change that affects verb phrases, but that is less well known: the emergence of brace
constructions, in which the (direct) object is located between the auxiliary and the perfective
participle or infinitive, as in:
(a) [AUXILIARY + (DIRECT) OBJECT + PERFECTIVE PARTICIPLE]
habeo
epistulam
scriptam
have-1sg.
letter-Acc.
write-Pf.Part.-Acc.sg.-F. I have written a letter
(b) [AUXILIARY + (DIRECT) OBJECT + INFINITIVE]
coepit
epistulam
scribere
start-Pf-3sg. letter-Acc.
write-Inf. he started to write a letter
This talk will compare constructions including habeo and perfective participles (cf. [a]) and
constructions including auxiliaries other than habeo and infinitives (cf. [b]).
Analysis of [HABEO + PERFECTIVE PARTICIPLE] constructions shows that once habeo had
emerged as an auxiliary in the later stages of Latin, the construction underwent a consistent
change in word order in Old and Middle French that can be summarized as follows:
EPISTULAM SCRIPTAM HABEO > HABEO EPISTULAM SCRIPTAM > HABEO SCRIPTAM
EPISTULAM. The brace construction therefore is part of a gradual development.

In contrast to [HABEO + PERFECTIVE PARTICIPLE] structures, [AUXILIARY +

INFINITIVE] constructions underwent a much longer andin the absence of nominal forms of
the verbpotentially less complex evolution. Yet their development in all likelihood is
similar because brace constructions are indeed attested in Latin, as in: incipiunt hymnos
dicere (Peregr.) they start to recite hymns .
Since this development has been a neglected topic in Latin and Romance linguistics,
we will focus in this presentation on these constructions, examining whether their history as
well shows a shift from [OBJECT + INFINITIVE + AUXILIARY] to [AUXILIARY + OBJECT +
INFINITIVE] to [AUXILIARY + INFINITIVE + OBJECT]. My own preliminary research indicates
that several of these structures indeed were well attested in the shift from Latin to Romance.
On the basis of data from Vulgar and Late Latin as well as Old French texts I will analyse
whether these varieties indeed reflect different chronological layers, relating where possible
my findings to the patterns observed for perfective habeo constructions and the overall spread
of VO.

When uniform stress placement collapsed: the history of English word stress told by Evolutionary Game
Theory
Andreas Baumann, University of Vienna
In this paper we propose an evolutionary approach to stress assignment in languages such as English, in
which stress is lexically fixed, and use it to explain the diachronic development of word stress in the
history of English. In this approach lexical stress patterns are seen as adaptations to utterance rhythm.
Since rhythmic structures normally involve sequences of two or more words, our approach differs radically
from most extant theories (such as Chomsky and Halle 1968, Halle 1973, Hammond 1999, Hayes 1981, or
Schane 2006), which derive lexical stress by means of rules operating on isolated lexical items, and which
typically fail to handle variability (as in English [re.search]N vs. [re.search]N) satisfactorily.
From the perspective of cultural evolution, the memorized phonological structures of words including
their stress patterns represent replicating competence constituents whose evolutionary stability
depends on their faithful expression and transmission. Since rhythmically motivated repair processes
(such as stress shifting or vowel lengthening) would clearly impede their faithful expression and
transmission, it makes sense to assume that words should be structurally adapted to the rhythmic
structures in which they come to be uttered. At the same time, utterance rhythm does not only represent
the environment to which words must adapt but is itself constituted through the combination of words.
The verification of this approach is done in two steps: In the first step, we use the formal language of
Evolutionary Game Theory (cf. e.g. Hofbauer and Sigmund 1998, Maynard-Smith 1982) to model
interactions between phonological structures: in our game-theoretical model, which is an extension of Ritt
and Baumann (2013), polysyllabic words figure as players, which can choose among different stress
placement strategies (e.g. always stress the first syllable). They combine to form utterances together with
(potentially unstressed) monosyllabic words, and receive rewards (so-called payoff) reflecting the
rhythmic quality of the utterances they form. The key assumption of Evolutionary Game Theory is that the
payoff a word receives that way reflects its fitness: thus, choosing a stress placement strategy incurring a
higher payoff will promote the faithful transmission and utterance frequency of a word. As we shall
demonstrate, the model allows us to derive interesting and testable predictions about probable
distributions of stress patterns in the lexica of natural languages, and about the dynamics involved in their
potential change. In particular, the model shows that the distribution of stress placement patterns among
polysyllabic words should crucially depend on the frequency of monosyllabic words.
In the second step, in order to test the predictions, we feed relevant parameter estimates drawn from
a historical corpus ranging from Old English to Early Modern English into the model. The results are
remarkable: the model predicts that along the measured trajectory characterizing the frequency
development of monosyllabic words, the distribution of stress placement strategies among polysyllables
changes from solely initial stress (i.e. Germanic word stress in OE) to mixed stressing strategies (as they
arguably characterize EME). Hence, the diachronic development of English word stress can be explained as
an epiphenomenon of a word-length reducing diachronic process, viz. phonetic erosion.

References
Burzio, Luigi. 1994. Principles of English Stress. Cambridge: University Press.
Chomsky, Noam and Morris Halle. 1968. The sound pattern of English. Cambridge, MA: MIT Press.
Halle, Morris. 1973. Stress Rules in English: A New Version. Linguistic Inquiry 4: 51-64.
Hammond, Michael. 1999. The Phonology of English: A Prosodic Optimality-Theoretic Approach. Oxford: University
Press.
Hayes, Bruce. 1981. A Metrical Theory of Stress Rules. Bloomington: Indiana University Linguistic Club.
Hofbauer, Josef and Karl Sigmund. 1998. Evolutionary games and population dynamics. Cambridge: University Press.
Maynard Smith, John. 1982. Evolution and the theory of games. Cambridge: University Press.
Ritt, Nikolaus and Andreas Baumann. 2013. Transferring mathematics to English studies. In: Markus, Manfred and
Herbert Schendl (eds.). Transfer in English studies. Vienna: Braumller.
Schane, Sanford. 2006. Understanding English word accentuation. Language Sciences 29: 372-384.

Kristin Bech and Christine Meklenborg Salvesen

University of Oslo
SUPERFICIAL TYPOLOGY: THE SO-CALLED VERB-SECOND CONSTRAINT IN OLD
ENGLISH AND OLD FRENCH

Old English (OE) and Old French (OF) are both referred to as verb-second (V2) languages,
i.e. languages in which the verb typically occupies second position even after a non-subjectinitial element, as exemplified in (1) and (2).
(1)

Mid twam wurscipum ge-glngde se lmihtiga scyppend s mannes sawle

with two dignities adorned the almighty God the mans soul
With two dignities Almighty God adorned mans soul (lfrics Lives of Saints)

(2)

Et por ce voudroit ele mieuz qu'ele fest Tristan morir en aucune maniere
and for that would she better that'she did Tristan kill in some way
And because of that she would rather have Tristan killed in some way (Tristan)

Another similarity between the two languages is that they both feature numerous non-V2
sentences in which several elements, including the subject, precede the verb. However, the
syntactic and information-structural characteristics of such clauses are completely different in
the two languages (Bech & Salvesen, forthcoming). Furthermore, OE is a much more
heterogeneous language than OF with respect to word order in general; it has word ordering
possibilities that do not exist in OF.
In this paper, we focus on the postverbal field(s) of the two languages, and our paper is
divided into three parts: 1) a presentation of empirical data on the postverbal position of
elements in OE and OF, using a corpus of ca. 1,500 main clauses from OE and ca. 1,800 main
clauses from OF; 2) a discussion of some syntactic implications of the empirical data, using a
cartographic generative framework and 3) a discussion of possible information-structural
factors that may have a bearing on element position in the two languages. We will show that
categorizing the two languages simply as V2 languages is an oversimplification of a
complex matter.
Reference:
Bech, Kristin & Christine M. Salvesen. Forthcoming. Preverbal word order in Old English
and Old French. In Information Structure and Syntactic Change in Germanic and Romance
Languages, ed. by K. Bech & K.G. Eide. John Benjamins.

The syncretism of genitive and dative in Old Iranian

Maria Carmela Benvenuto & Flavia Pompeo
Sapienza University of Rome
mariacarmela.benvenuto@uniroma1.it
flavia.pompeo@uniroma1.it
As is well known, case syncretism is a phenomenon that has taken place in many IndoEuropean languages, and involves a reduction in the number of cases compared to the system
usually reconstructed for Proto-Indo-European (Luraghi 2000). This paper will focus on the
syncretism of the Indo-European dative and genitive in Old Iranian languages, devoting
particular attention to the changes which occurred from Old Avestan to Young Avestan.
Old Persian does not have separate forms for genitive and dative as these previously
distinct grammatical categories merged into the genitive (Meillet & Benveniste 1931). Indeed,
a detailed examination of Old Persian textual data reveals that the genitive expresses a variety
of syntactic functions and semantic roles, cross-linguistically typical not only of the genitive,
but also of the dative (Pompeo & Benvenuto 2011; Benvenuto & Pompeo 2012).
As far as Avestan is concerned, the dative dies out in Young Avestan while the
genitive spreads to an increasing number of fields beyond its original sphere of meaning,
replacing the dative in many contexts. Regarding this, Reichelt (1909), in the section entitled
Der dativische Genitiv, suggests that the high frequency of the adnominal genitive
determined the progressive extension of this case at the expense of the original functions of
the dative. In Young Avestan, this phenomenon evolved and eventually comprised the
syntactic function of Indirect Object and the role of Recipient.
The study of the Old Iranian material makes it possible to demonstrate that the decay
of the dative is not due to phonological changes or to the progressive silencing of post-tonal
syllables. On the contrary, it suggests that the main cause of this syncretic process is to be
found in both the semantic and syntactic overlapping of genitive and dative cases. In
particular, as far as the syntactic level is concerned, it is well-known that languages have a
tendency to eliminate synonymous grammatical forms over time (Bardal & Kulikov 2009).
In both Old Persian and Young Avestan, the result of the syncretism is a new genitive case,
structured as a polysemic radial category, with the expression of the Possessor as its core
function, while the other attested functions constitute metaphorical extensions of the
Possessor. The direction of the syncretism in question is confirmed by the general drift in the
evolution of both languages.

Bardal, J. & Kulikov, L.I. (2009). Case in decline. In: Malchukov, A. & Spencer, A. 2009,
eds.), The Oxford handbook of case. Oxford, pp. 470-478.
Benvenuto, M.C. & Pompeo F. (2012). Il sincretismo di genitivo e dativo in persiano
antico. In: Vicino Oriente 16, pp. 151-165.
Luraghi, S. (2000). Synkretismus. In: Booij, G., Lehmann, C. & Mugdan, J. (2000, eds.),
Morphologie.
Ein
internationales
Handbuch
zur
Flexion
und
Wortbildung./Morphology. An international handbook on inflection and wordformation, Vol. 1. Berlin - New York, pp. 638-47.
Meillet, A. & Benveniste, . (1931). Grammaire du vieux-perse. Paris.
Pompeo, F. & Benvenuto, M.C. (2011). Il genitivo in persiano antico. Un caso esemplare di
categoria polisemica. In: Studi e saggi linguistici 49, pp.75-123.
Reichelt, H. (1909). Avestisches Elementarbuch. Heidelberg.

Ivar Berg
Department of Language and Literature, NTNU (Trondheim)
ivar.berg@ntnu.no
Stages in deflexion: The (case of) Norwegian dative (case)
The usual story tells that the Old Norse four-case system disappeared in Norwegian during the
Late Middle Ages. However, when one examines the process more thoroughly it becomes
clear that this was by no means a straight-forward process. The present paper examines what
happened in more detail, aiming to sort out discernible stages in the process of deflexion.
Some Modern Norwegian (and Swedish) dialects still retain a dative case, which in itself
shows that case inflection did not simply disappear all over. By comparing Old Norse and a
Modern Norwegian dialect, Sandy (2012) demonstrates how the dative is actually more
widely used now, having replaced the genitive (more rarely the accusative). Both in form and
function there is no doubt the old dative, though, not a new nominativeoblique system. Two
examples of changes will be discussed, with examples from 15 th century written sources
(found in the Diplomatarium Norvegicum):
a) The former genitive-governing prepositions til to and millum between are
increasingly found with dative complements, as still is the case in dialects retaining
the dative. When genitive lost its function as a lexical case, dative was the only
marked option left.
b) In the inflection of the demonstrative essi this (the older form sj disappeared early)
a new stem enn- (based on acc.sg.masc. ann) replaced ess- in nom.sg.masc. and
nom./acc.sg.fem. This established a contrast between nominative enn- and dative
ess- and indicates that marking the case distinction was decisive for the change of
stem form.
It thus seems that Norwegian at one stage, presumably much more widely than in present
dialects, had a two-case system which was established and strong enough to reshape
grammatical structure. This point is easily missed if one just compares the two well-known
stages Old NorseModern Norwegian without properly considering the intermediate stages
the linguistic development went through. It also means that nominative and accusative merged
before dative merged with either, which seems contrary to the case hierarchies given e.g. in
Blake (2001).
References
Blake, Barry J. 2001. Case. 2nd ed. Cambridge: Cambridge University Press.
Diplomatarium Norvegicum IXXIII. 18472011. Christiania/Oslo. Vols. IXXI online:
http://www.dokpro.uio.no/dipl_norv/diplom_felt.html.
Sandy, Helge. 2012. Syntaktisk og semantisk styrt dativ. In: Unn Ryneland and Hans-Olav
Enger (eds.): Fra holtijaR til holting. Sprkhistoriske og sprksosiologiske artikler til Arne
Torp p 70-rsdagen. Oslo: Novus, s. 319338.

Are there diachronic universals of agreement systems?

Balthasar Bickel, Alena Witzlack-Makarevich, Taras Zakharko, and Giorgio Iemmolo
University of Zurich

Reconstruction often relies on assumptions about what constitutes natural, and

therefore universally preferred, pathways of change. One method for testing such assumptions is the Family Bias Method (Bickel 2011), which infers likely diachronic trends
from distributional biases within established language families and applies probabilistic
reasoning to extrapolate trends in small families and isolates.
In this presentation we apply this method to typological data from agreement system
in order to evaluate two hypotheses that are generally accepted even though the evidence
for them is restricted only to a handful of languages.
The rst one, which we call Watkins Law, based on Watkins (1962), Benveniste
(1946), and Jakobson (1932), maintains that zero forms in agreement paradigms are more
likely to develop and be preserved in the third rather than in the rst or second person,
as has been documented, e.g. for the development of agreement morphology from ProtoIranian to Persian (Watkins 1962: 83.). The second hypothesis to be tested, which we
call Silversteins Law (Silverstein 1976), predicts that ergative alignment is more likely
to develop and be maintained in the third person rather than in the rst and second
person.
Based on a detailed analysis of agreement paradigms in over 120 languages worldwide,
we estimate diachronic trends on zero forms and alignment types. For alignment types,
we nd no evidence of a universal trend in line with Silversteins Law, thus conrming
earlier ndings based on dierent methods. Likewise, for Watkins Law, we nd no
evidence if the principle is taken in its original sense, as bearing on paradigm shapes.
There is no universal trend in our data for languages to (re-)structure paradigms so as
to have more zeros in the third than in the rst and second person. That is, we nd that
worldwide, there are about as many families showing such a trend as there are families
with a trend towards an opposite or equi-proportional distribution of zeros.
However, the principle receives (weak) statistical support if individual markers are
considered regardless of the rest of the paradigm. There is a (borderline) signicant
trend for rst and second person categories to develop and maintain overt agreement
more often than third person. This trend is likely to reect grammaticalisation processes
of independent pronouns developing into agreement markers (Givn 1976, Ariel 2000):
since third person pronouns tend to be less frequent in discourse, they have an inherently
smaller probability of developing into overt third person agreement markers than rst
and second person pronouns.

Plural allomorphy in German dialects: the case of subtraction

Magnus Breder Birkenes (University of Marburg)
German exhibits a great deal of plural allomorphy, both diachronically and synchronically.
The paper is concerned with the apparently subtractive plural forms in many German dialects,
e.g., dg d day, days or knd kn child, children (cf. Wurzel 1984, Golston and Wiese
1996). Drawing on the fact that the occurrence of such forms in contemporary dialects is
unpredictable, the hypothesis is raised that we are dealing with historically far-reaching
phonological processes. This can be seen from the fact that similar subtractions appear in
analogous phonological environments (cf. Holsinger & Houseman 1999). Historically, we
find variation between the plural endings -er, and subtraction within one and the same
lexeme, a diversity that has existed for hundreds of years. Subtraction and -e, however, never
co-exist.
This leads one to the assumption that all subtractive forms in German can be traced
back to the historical presence of a schwa ending which was lost during apocope, a process
common to most Low and High German dialects. Crucially, before the loss of the schwa,
assimilations in consonant clusters and the general process of lenition in sonorous
environments led to the elision of a stem-final consonant. It is a truly natural phonological,
polygenetic development, in that it is found in diverse Low and High German dialects, and
other Germanic languages like North Frisian. The genesis of subtractive forms can thus be
thought of as a historical accident.
For their synchronic existence, however, another explanation is demanded. The
analysis to be presented here differs from older work on this topic (cf. Wurzel 1984 and
Golston and Wiese 1996) in which an attempt is made to derive the subtractive forms using
existing synchronic rules or principles of well-formedness. From a synchronic point of view,
Natural Morphology views subtractive forms as counter-iconic: a marked form (plural) is
symbolized with less phonological material than the unmarked form (singular). It therefore
predicts subtractive forms to be replaced with modulatory or additive forms. While there is
much truth in this, Natural Morphology cannot easily explain why these forms have existed
for over 500 years, as examples from language enclaves of German and a reconstruction of
the processes show.
In this paper, which is founded on empirical data from over 300 dialect grammars
from all over the German speaking area, I will argue that local markedness resulting from the
higher frequency of a marked grammatical category (cf. Tiersma 1982) favours such
subtractive forms in cases where the plurals are more frequent than the singulars, e.g.
children, teeth and leaves. These plurals may, due to their high frequency, have been
reanalysed as the base forms. An analysis where the singulars are derived from the plurals is
also supported by productive uses of the pattern in modern Hessian dialects in cases where the
singular can be said to be marked. In this way, I argue, subtraction could prevail against
competing markers in many lexemes. Such an analysis poses a problem to traditional models
of morphology because of the many affixes or rules needed, but is easily reconstructable in an
associative model of morphology (cf. Bybee 1985).
Bybee, Joan L. (1985): Morphology: A study of the relation between meaning and form. (Typological studies in
language 9), Amsterdam: Benjamins.
Golston, Chris / Wiese, Richard (1996): Zero morphology and constraint interaction: subtraction and epenthesis
in German dialects. In: Booij, Geert / van Marle, Jaap (eds.) Yearbook of Morphology 1995, Dodrecht:
Kluwer, 143159.
Holsinger, David / Houseman, Paul (1999): Lenition in Hessian: cluster reduction and subtractive plurals. In:
Booij, Geert / van Marle, Jaap (eds.) Yearbook of Morphology 1998, Dordrecht: Kluwer, 159174.
Tiersma, Peter M. (1982): Local and General Markedness. Language 58, 832849.
Wurzel, Wolfgang Ullrich: (1984): Flexionsmorphologie und Natrlichkeit: ein Beitrag zur morphologischen
Theoriebildung. (Studia Grammatica 21), Berlin: Akademie-Verlag.

Decomposing the Old Spanish Left Periphery Using a Parallel Corpus

Miriam Bouzouita
Ghent University / University of Cambridge
Miriam.Bouzouita@UGent.be
The current study focusses on various Old Spanish phenomena in which one or more
constituents appear fronted at the left-periphery, with or without a following resumptive element.
More concretely, it examines so-called focalisations and left dislocations, such as Clitic Left
Dislocations (CLLD) and Hanging Topic Left Dislocations (HTLD). The aim of this study
consists in unveiling the syntactic and pragmatic differences that exist between these Old Spanish
phenomena and their contemporary counterparts. The current research has been carried out on the
basis of the parallel Biblia Medieval corpus in order to control more steadily certain variables that
can potentially alter the outcomes of this study.
Escobars (1995) study of various left-periphery phenomena in contemporary Spanish,
while verifying the different properties of CLLD-HTLD identified by Cinque (1990) for modern
Italian, concludes that for Spanish a-case-marking of the left-dislocated constituent is a
characteristic of CLLDs that distinguishes it from HTLDs. Diachronically, this criterion proves to
be untenable as the preposition a has not always been available as a grammatical marker of
accusative case in the history of Spanish (e.g. Laca 2006). Even in the 20th c. the
grammaticalisation of this preposition as an accusative case marker has not yet been completed
since inanimates do not need to co-occur with it. Similarly, focalisations have been distinguished
from CLLDs and HTLDs on the basis of lack of resumptive pronouns. However, contemporary
Spanish strong pronouns can no longer occur without a co-referential clitic and will always appear
doubled even when topicalised. Yet their Old Spanish counterparts had no such restriction.
Similarly, although contemporary Spanish CLLDs are said to be sensitive to strong islands, this
criterion does not appear to apply in Old Spanish (Bouzouita in press [2013]). In sum, this study
will show that, given conventional criteria, there appears to be considerable overlap between the
supposedly distinct structures of focalisation, HTLD and CLLD. Moreover, it will be shown that
the aforementioned phenomena can be observed in the same syntactic and pragmatic contexts.

References
Bouzouita, M. (in press [2013]) Left Dislocation Phenomena in Old Spanish: An Examination of
their Structural Properties In: Left Sentence Peripheries in Spanish. Dufter, A. & A. O. de
Toledo y Huerta (eds). Philadelphia/Amsterdam: John Benjamins.
Cinque, G. (1990) Types of A-Dependencies. Cambridge, Mass.: MIT Press.
Escobar, Ma A. (1995) Lefthand Satellites in Spanish. Utrecht: OTS.
Laca, B. (2006) El objeto directo, En: Sintaxis histrica de la lengua espaola, Company C.
(ed.). vol. 1, Mexico: UNAM, p. 421-475.

Rates of Irregular Change: Phonological Reduction and Grammatical Convergence in Iranian

Chundra A. Cathcart
University of California, Berkeley
Non-Persian Modern Iranian languages share many formally identical or similar functional morphemes with Modern Persian (NP). This paper asks whether this similarity is due to patterns of drift shared by Iranian languages
(cf. Sapir 1921, chs. 7-8), or contact with NP, either in the form of grammatical borrowing or accommodation. I
compare the sound changes, regular and irregular (viz., phonological reduction), that derive fifteen NP functional
items against the historical phonology of their cognates in other Iranian languages. Some functional items show
an EQUAL number of irregular changes across Iranian; when coupled with language-specific regular changes, such
cases are clearly due to drift. But some non-Persian items show a GREATER number of irregular reductions than
their identical NP cognates. Often, the non-Persian language is otherwise more phonologically conservative than
NP. I argue that such asymmetrical cases point to sociolinguistic pressure from NP.
Items that show an EQUAL number of reductions between NP and other Iranian languages often also show
distinctive regular changes which make it clear that the similarity is an effect of drift, e.g., Old Iranian *haca
with Mid/NP az, Parthian a, Baloc ac, Kurdish from. Here, OIr *h- has been irregularly lost by all
languages, and the languages have undergone language-specific treatments of *-c-. But in other items, the number
of reductions in a non-Persian language is GREATER than that of an identical form in NP. For example, OIr *radsake (OP radiy) > MP ray (dat.-abl. marker) > NP ra (obj. marker) via regular change. Parth ra (dat.-abl.)
shows a regular outcome, but Bal ra (obj. marker) does notas a rule, Bal preserves OIr *-d-. If we take Bal ra
to be an inherited form, it shows a greater degree of reduction than its NP cognate.
This is contradictory. Impressionistically, Baloc is more conservative in its historical phonology than NP, and
not prone to lenitionits most noteworthy innovation is the fortition of OIr fricatives. This impression is confirmed by comparing relative Levenshtein distances (Nerbonne et al. 1996) between OIr and Baloc (mean = .44)
and OIr and NP (mean = .56) for 50 reliable cognate sets. A Welchs 2-sample t-test shows that the mean relative
distance between NP and OIr is greater than that of Baloc and OIr (t = 2.79, df = 96.8, p < .01), indicating that Baloc shows significantly less regular change from OIr than NP. In items where Baloc shows greater
irregular change than NP, I argue that it shows either (a) grammatical borrowing from NP or (b) grammatical
accommodation, a special type of convergence by which languages in contact adjust to make their...grammatical
elements more similar in form and function (Johanson and Robbeets 2012:11). In the absence of distinctive NP
sound changes, it is difficult to determine whichstill, either way, it points to sociolinguistic pressure from NP.
This study finds that across Iranian, grammatical items fall into three categories: (1) identical (cross-linguistically), with an equal number of phonological reductions; (2) non-identical, with an equal number of reductions; and
(3) identical, with fewer reductions in NP (I have not encountered identical items with a greater number of reductions in NP, or non-identical items showing different numbers of reductions). (1) is an uninformative category;
while many of these items have been borrowed (from NP into Indic or Turkic), the historical phonology does not
clearly indicate whether it is an inherited form or an NP borrowing. (2) securely points to drift or independent
innovation, as language-specific changes have operated (the possibility of early loans notwithstanding). (3) points
to sociolinguistic pressure from Persian, especially if a language which superficially shows more reduction, for
instance Baloc, is known to be more phonologically conservative than NP. Thus we see that parallelisms in the development of Iranian grammatical items come about due to both drift (ostensibly an effect of genetic relatedness)
and sociolinguistic pressure.
References
Johanson, L. and M. Robbeets (2012). Bound morphology in common: Copy or cognate? In L. Johanson and
M. Robbeets (Eds.), Copies Versus Cognates in Bound Morphology, pp. 322. Leiden: Brill.
Nerbonne, J., W. Heeringa, E. van den Hout, P. van der Kooi, S. Otten, and W. van de Vis (1996). Phonetic
distance between Dutch dialects. In G. Durieux, W. Daelemans, and S. Gillis (Eds.), CLIN VI, Papers from the
Sixth CLIN Meeting, Antwerp, pp. 185202. University of Antwerp, Center for Dutch Language and Speech.
Sapir, E. (1921). Language: an Introduction to the Study of Speech. New York: Harcourt, Brace and Company.

Anticausativization between Late Latin and early Italo-Romance:

the semantics of predicates and the syntax of voice
(Michela Cennamo - University of Naples Federico II)
This paper investigates the role played by the interplay of the aspectual template of verbs, the verbs inherent
meaning (the 'root'), and the nature of the P subject (e.g., animacy and control) in determining the
distribution of the different strategies available in the diachrony of Latin and in early Italo-Romance to mark
anticausativization (the (medio-passive) -r form, the reflexive pattern and the active intransitive in Latin, the
active intransitive and the reflexive in Old Italian).
The different forms are usually described in the literature as interchangeable (Feltenius 1997 for Latin,
Brambilla Ageno 1964 for Old Italian). We will argue, instead, that the structural and lexical aspects of the
verb meaning (following Levin & Rappaport Hovav 2005, among others) as well as the inherent and
relational properties of verbal arguments affect the use of the different patterns, interacting with the encoding
of voice, both synchronically and diachronically.
In particular, building on Cennamo, Eythrsson & Bardal 2011, we will show, through a corpus-based
investigation, that the selection of the reflexive strategy in Latin was initially confined to inherently telic
predicates (achievements and accomplishments) (e.g., frangere 'break', dum calor se frangat - till heat
RFL breaks - 'Till the heat goes down', fervefacere 'heat up', se patinae fervefaciunt RFL - pans heat-up 'The pans heat up', Plaut. Pseud. 831-33), whilst the active intransitive mainly occurred with verbs of
variable/reduced telicity (e.g., lenire 'soothe', irae leniunt - anger soothes - 'Anger soothes', Plaut. Mil. 583),
with activities (e.g., volutare 'roll', saxa volutant - stones roll - 'Stones roll') and, marginally, with
accomplishments lexicalizing a reversible state (e.g., aperire 'open', foris aperit - door opens - 'The door
opens', Plaut. Persa 300) (see also Cennamo 1998: 88). Gradually, in the course of time, the reflexive
spreads to non-inherently telic (e.g., minuere 'decrease', minuente se morbo - decreasing RFL illness - Plin.
Nat. 23, 50) and atelic predicates (e.g., servare 'keep'- mala se servant - apples RFL keep - 'Apples keep'),
and the active intransitive expands to inherently telic predicates (e.g., rumpere 'break'), until in late texts the
three anticausative forms become truly interchangeable (rumpunt dentes/rumpuntur dentes/dentes se
rumpunt - break teeth/break-R teeth/teeth RFL break - Its teeth break' (sc. equus 'horse') (Chiron 775)).
The equivalence between the anticausative strategies obtaining in Late Latin is also found in some early
Italian vernaculars, e.g., Old Florentine, where the active intransitive and the reflexive alternate with all verb
classes (e.g., degree achievements, aumentare 'increase', le pene ... s'aumentano - the punishments RFL
increase -'The punishments increase' (Boccaccio, Esposizioni, 47, 664.20) vs. le biade aumenteranno - the
corn will-increase - 'Corn will increase' (Boccaccio, Filocolo, 5, 54, 624.2), with hints of the gradual gaining
ground of aspectual notions such as telicity in determining the obligatory occurrence/preference of the
reflexive form to mark anticausatives, interacting with the reconstitution of the voice system in the transition
to Romance (Cennamo 2012)
The Latin and Old Italian data, therefore, appear to offer an interesting contribution to the current debate
on the role played by the verb's inherent meaning and its interaction and integration with the event structure
template of predicates in determining argument realization, showing the relevance of these notions for the
diachrony of anticausativization in Latin and the transition to Romance.

References
Brambilla Ageno, F. 1964. Il Verbo nell'Italiano Antico. Milan/Naples: Ricciardi.
Cennamo, M. 1998. The loss of The loss of the voice dimension between Late Latin and Early Romance. In
Historical Linguistics 1997, ed. by M. S. Schmidt, J.R. Austin & D. Stein, 77107. Amsterdam:
Benjamins.
Cennamo, M. 2012. Aspectual constraints on the (anti)causative alternation in Old Italian. Transactions of
the Philological Society 110.3: 394-421. Thematic issue on Argument Realization and Change.
Cennamo, Eythrsson & Bardal 2011. The rise and fall of anticausative forms and constructions in
Indo-European: the context of Latin and Germanic, under revision for Linguistics.
Feltenius, L. 1997. Intransitivizations in Latin. Upsala: Almquist & Wiksell.
Levin B. & Rappaport Hovav, M. 2005. Argument Realization. Cambridge: Cambridge University
Press.

Dating Proto-Indo-European: A revised computational phylogenetic

analysis supports the steppe hypothesis
William Chang, David Hall, Chundra Cathcart, and Andrew Garrett
University of California, Berkeley
Within Indo-European studies and historical linguistics more broadly, despite long study and
dispute, it remains controversial when and where Proto-Indo-European (PIE) was spoken. According
to the hypothesis of Renfrew (1987, 1999, 2000, 2001), PIE was spoken in Anatolia and was spread
broadly with the diffusion of agriculture. Since the chronology of farming dispersal is understood, on
this hypothesis PIE was spoken in the 7th millennium BCE. An alternative hypothesis (Mallory 1989,
Anthony 2007, Parpola 2008) is that PIE was spoken in the steppe north of the Black and Caspian
Seas between about 4500 and 3500 BCE. According to this steppe hypothesis, language spread in
Asia and Europe was associated with horse domestication, wheeled transport, and other cultural
innovations. The two analyses of PIE origins thus differ in chronology by some 3000 years. In the last
decade this debate has been put on a new footing with research by Gray & Atkinson (2003), Bouckaert
et al. (2012), and other associated work by the same authors and their colleagues.
In this line of research, computational methods from biological phylogenetics are applied to
linguistic data (e.g. basic vocabulary lists) in order to infer phylogeny and chronology. The latter in
particular does not rely on the glottochronological assumption of a constant rate of change, but
assumes only that rates of change are distributed in a statistically tractable way. Without exception,
and using a variety of tools, datasets, and statistical and model assumptions, work along these lines has
pointed to the Anatolian chronology for PIE. Most Indo-Europeanists do not accept this chronology,
so the results have been controversial and widely reported.
In this paper, we use basically the same methods and tools as a recent study by Bouckaert et
al. (2012), which argued forcefully for the Anatolian hypothesis, to demonstrate that the data actually
support the steppe hypothesis. When the data are coded correctly, the phylogenetic tools applied by
earlier researchers in fact produce a quite different result, one that is consistent with the steppe
hypothesis. Earlier work made several errors in data coding that produce a direct effect on the
estimated chronology. We mention two such errors here.
First, we encoded known descent relationships. For example, Irish and Scots Gaelic are
descended from Old Irish, and English is descended from Old English. Such relationships are crucial
because rates of change in known historical periods should inform inferences about unknown periods.
But Bouckaert et al. (2012) left these relationships as a matter of inference to the software rather than
stipulating them; they were then systematically inferred wrongly. For example, in their analysis, the
latest common ancestor of Old and Modern Irish and Scots Gaelic was far earlier than Old Irish itself;
in turn this led prehistoric ancestral dates to be estimated as too early. The error arises from patterns of
borrowing, drift, and accidental convergence.
Second, from the dossier of innovations we did not exclude lexical changes involving
borrowed words. Crucially, borrowing can be recognized in medieval and modern languages, where
we know the linguistic milieu, but not in ancient IE languages, where the borrowing may be from
unattested substrate languages. Loans from unknown sources are analyzed as endogenous innovation.
If modern borrowings are excluded, rates of change are estimated wrongly and the overall chronology
extended.
When these and other linguistically justified changes in data coding are made, and
phylogenetic software run, the impact on chronology is significant and yields date inferences
consistent with the steppe hypothesis. Our presentation will detail the linguistic rationale for all our
coding changes. In addition, describing multiple software runs, we will show that our results are robust
to varying assumptions about IE subgrouping and about the intervals at which linguistic splits happen.

Directional phrases in the history of Dutch

Robert Cloutier
Universiteit van Amsterdam
Modern Dutch prepositional phrases can be ambiguous between a locational and a directional
reading, as seen in (1).
1. Hij loopt in het gebouw.
a. He is walking in the building (i.e. he is inside the building walking around)
b. He is walking into the building (i.e. he is moving from a location outside of the
building to a location inside the building)
However, if the prepositional element occurs after the noun phrase, only the directional reading is
possible, as seen in (2).
2. Hij loopt het gebouw in.
a. *He is walking in the building
b. He is walking into the building
Numerous scholars have studied the construction exemplified in (2) (Los et al. 2012; Belin 2008;
Blom 2005; Helmantel 2002 among others), yet the question about the constituency of such
constructions remains unresolved: Are these best analyzed as postpositional phrases or as complex
predicates? The various constituency tests used to argue for or against a particular analysis have
proven inconclusive for a variety of reasons (Belin 2008). Examination of the historical
development of these constructions may shed some light on this question.
Many scholars have observed that Middle Dutch did not have these constructions
(Hogenhout-Mulder 1983; Van der Sijs 2005; Cloutier 2006) and that these were rather expressed
with prepositions in Middle Dutch (Gerritsen 1978; Hogenhout-Mulder 1983; Van der Wal and Van
Bree 1992; Cloutier 2006). Cloutier (2006), focussing on the adposition in, does not find examples of
this postpositional construction until the 19th century. However, two scholars discuss Middle Dutch
examples with positional phrases: Blom (2005) found instances with the adposition door through,
and Belin (2008) discusses instances of postpositions occurring in conjunction with the preposition
te at. These data suggest a timeline of the rise of postpositional directional phrases, though a
comprehensive study of the relevant adpositions over time is still lacking. I will examine the use over
time of a number of adpositions that express direction in Modern Dutch in a corpus of Middle and
Early Modern Dutch texts available on the CD-ROM Middelnederlands and the Digitale Bibliotheek
voor de Nederlandse Letteren.

Eleanor Coghill, University of Konstanz

The Development of Ergativity in Aramaic
In eastern Aramaic, ergative alignment in the perfect (later the past perfective)
developed as a result of the loss of the old Semitic verbal conjugations and their
replacement with participle forms. The latest stage of the language (North-eastern NeoAramaic) has seen a shift back to fully accusative alignment in most dialects. Thus an
entire cycle of alignment change can be seen in the documented history of Aramaic. In
the dialects where ergativity still survives it is restricted to verbal agreement (there is
no case marking). In these dialects ergativity is being further eroded, both by movement
towards Split-S alignment and by the increasing use of separate markers for O.
There has been much discussion of the pathways through which a shift to tenseconditioned ergative alignment may take place: for Iranian, Indo-Aryan and Aramaic,
various possibilities have been suggested, including the reanalysis of passives or of
possessive predicates (cf., e.g. Benveniste 1952, Cardona 1970, Kutscher 1969). More
recently, a connection between dative (expressing roles such as experiencer) and
ergative has been noted (e.g. Butt 2006, Haig 2008) and this appears to be supported by
the Aramaic case. The lexical distribution of the ergative construction in earlier Eastern
Aramaic (as established by a corpus search) suggests that the original construction may
have been limited mostly to psychological verbs, such as see, hear and know, which
could take a dative argument for the experiencer. These experiencers were then
reanalysed as agents. There is also evidence suggesting that beneficiaries too were
reanalysed as agents. Aramaic is ideal as a case-study for alignment change, due to the
wealth of texts, dating from 3000 years ago to the present day. The paper will present
evidence from both ancient texts and modern dialects, in order to understand the
changes in alignment and their motivations.
References
Benveniste, mile. 1952 La construction passive du parfait transitif, Bulletin de la Socit de
linguistique 48, 52-62.
Butt, Miriam. 2006. The Dative-Ergative Connection. In Patricia Cabredo-Hofherr (ed.),

Proceedings of the Colloque Syntax-Semantique Paris (CSSP) 2005.

Cardona, G. 1970. The Indo-Iranian construction man (mam) ktam. Language 46, 1-12.
Haig, Geoffrey L. J. 2008. Alignment Change in Iranian Languages: A Construction Grammar
Approach. Berlin; New York: Mouton de Gruyter.
Kutscher, Y.E., 1969, Two passive constructions in Aramaic in the light of Persian, Proceedings
of the international conference on Semitic studies, Jerusalem, 132-151.

Change or contact? English vs Afrikaans

Jac Conradie
University of Johannesburg

As the development of Afrikaans on the basis of 17th century Dutch is

often ascribed to language contact rather than diachronic change, a
series of similar developments in the history of English and that of
Afrikaans - in instances where the influence of English on Afrikaans is
excluded - offers an opportunity for identifying the role played by
language contact in effecting change. In the verbal system, the
following areas may be considered:

Loss of concord in the finite verb - more advanced in English than

in Dutch and complete in Afrikaans.
Virtual disappearance of the distinction between finite verb and
infinitive accompanied by a more rigid ordering of elements of the
verbal string in both English and Afrikaans.
Replacement of the BE perfect (Du. zijn, etc., Afr. is) by the HAVE
perfect (Du. hebben, etc., Afr. het) in so-called mutative verbs (e.g.
verbs expressing movement or change), a distinction still
maintained in Dutch.
The isolation of modal verbs as a separate verbal category in
English and Afrikaans to a much larger extent than in Dutch, e.g.
by their lacking present and past participles and by not allowing
ellipsis of the main verb.
The retention of relatively conservative forms of the past participle
in attributive function and/or with semantic specialisation or in a
figurative sense, e.g. Eng. drunken, beholden, ill-gotten, proven,
new-mown, molten, misshapen, shrunken, stricken, etc. and Afr.
gebonde (boek) vs gebind, gebroke (hart) vs gebreek, opgewonde
(kind) vs opgewen, etc.

In the nominal system, two salient developments are:

The loss of grammatical gender in both English and Afrikaans,

while Dutch retains a two-class distinction.
The frequent and very similar usage of the prenominal possessive,
as in Eng. the hunters shot, a years work, Afr. die jagter se skoot,
n jaar se werk.
-o0o-

Lexical expansion in the have and be perfect in Dutch

A diachronic collostructional analysis
Evie Couss
University of Gothenburg
The grammaticalization of the perfect is a well-studied topic within historical linguistics. It is widely
known that the perfect auxiliary have in for instance English or French has grammaticalized from a
possessive verb by losing its original lexical content and starting to function as an abstract grammatical
marker. Bybee & Dahl (1989) were one of the first to point out that the grammaticalization of the perfect
auxiliary is accompanied by changes in the surrounding grammatical context. This study will focus on
the properties of the main verbs that are combined with the perfect auxiliary in a perfect construction.
Bybee & Dahl (1989) and Bybee et al. (1994) hypothesize that the grammaticalization of perfect
auxiliaries is accompanied with a lexical expansion from telic change of state predicates to other types
of verbs.
This study investigates whether such a lexical expansion actually has taken place in the have and be
perfect in Dutch. Both perfect constructions combine the perfect auxiliary have or be with a past
participle. The be perfect is largely restricted to intransitive past participles that involve some change of
state (also known as unergatives or mutatives) whereas the have perfect combines with all other types of
past participles. This contribution presents a diachronic study of the past participles used in both perfect
constructions taken from a corpus of Dutch legal texts dating from 1250 till 1800. The sample under
investigation contains 1344 past participles featuring in the have perfect and 499 past participles in the
be perfect.
The collected past participles are submitted to an innovative statistical analysis, known as the diachronic
distinctive collexeme analysis. The analysis is developed by Hilpert (2006) as part of a larger family of
statistical measures called collostructional analysis. Collostructional analysis in general measures the
degree of attraction that lexical items exhibit to constructions. Diachronic distinctive collexeme analysis
in particular compares the attraction of lexical items to one construction over sequential periods of time.
The statistical method allows us to observe changes in the collocational preferences of a construction
such as the expected lexical expansion in the perfect construction.
The diachronic distinctive collexeme analysis of both perfect constructions indicates that they show
evidence of lexical expansion in the observed time period. The have perfect expands from verbs of
change of possession verbs of communication, possession and perception transitive activity verbs
intransitive telic verbs in irrealis contexts. The be perfect shows lexical expansion from change of
location verbs and change of state verbs verbs of occurence and verbs of continuation of pre-existing
condition atelic existence of state verb zijn be. These changes in collocational preferences will be
argued to reflect the ongoing grammaticalization of the have and be perfect in the observed time period
in Dutch.
References
Bybee, J. & . Dahl (1989) The creation of tense and aspect systems in the languages of the world. In: Studies in Language
13, 51-103.
Bybee, J., R. Perkins & W. Pagliuca (1994) The evolution of grammar. Tense, aspect, and modality in the languages of the
world. Chicago: University of Chicago Press.
Hilpert, M. (2006) Distinctive collexeme analysis and diachrony. In: Corpus Linguistics and Linguistic Theory 2, 243-257.

Phonetic Factors in Camuno *n-loss and *m-maintenance in word-final syllables

Michela Cresci
CUNY Graduate Center

Camuno, a variety of Eastern Lombard, is an endangered Romance language of northern Italy

spoken in Valcamonica. Though a great deal of work has been done on the history of coda-nasals in
Romance (e.g. Hajek 1997; Sampson 1999), the Camuno facts have yet to be integrated into this bigger
picture.
Fieldwork on Camuno has been on-going by this author for the past three years, and forms the basis
of this study. Internal reconstruction as well as comparative evidence demonstrate that historical loss of coda
nasals in Camuno has occurred, and involves four phonetic or phonological variables: stress; cluster vs.
singleton coda; voicing of the following obstruent in NC clusters; and place of articulation of the nasal. That
a singleton nasal is lost in a final stressed syllable, but not a final unstressed syllable is illustrated by pairs
like b good, m. (cf. < *bun, cf. bna good, f.) vs. zn donkey (< Lat. asin-). That nasals can be
maintained in word final codas of stressed syllables is illustrated by pairs like b good, m. (cf. < *bun, cf.
bna good, f.) vs. tnt round m. ( < *tnd, cf. tnda round f.). That obstruent voicing played a role in
coda nasal loss is clear from comparing forms like tnt round m. ( < *tnd, cf. tnda round f.), where a
nasal is maintained before a historically voiced obstruent, to those like tt a lot (cf. tant very much),
where the nasal is lost before a historically voiceless obstruent. This pattern is evident for both /n/ and /m/,
so we find /m/- loss in kp field (cf. kampt small field) before a historically voiceless obstruent, but /m/
maintained in gmp stem (cf. gambt small stem) where mp < *mb. While in clusters, the pattern of *n
and *m-loss are parallel, this is not the case for single-member codas. Word-finally, *n has been lost, as in
b good, m. (< *bun), but *m is maintained: fm smoke, d m dome, etc.
While some of these patterns are evident in other Romance languages, the absence of evidence of
vowel nasalization and the occurrence of historical word-final obstruent devoicing (Cresci 2012) make this
historical development of special interest. In this study I suggest phonetic factors that may explain: (1)
differential loss of *n vs. *m in word-final position of stressed syllables; (2) the role of obstruent voicing in
nasal maintenance; and (3) why, unlike French, Portuguese and other Romance languages, there is no
evidence of contrastive nasalized vowels in the history of Camuno (cf. Beddor 2009).

References
Beddor, P. S. (2009). A coarticulatory path to sound change. Language, 85.4, 785-821.
Cresci, M. (2012). Camuno Final Obstruent Devoicing. Poster presented at 2nd Workshop on Sound
Change, Institute of Phonetics and Speech Processing, Munich, Germany.
Hajek, J. (1997). Universals of sound change in nasalization. Oxford: Blackwell.
Sampson, R. (1999). Nasal vowel evolution in Romance. New York: OUP.

From Latin to Romance: discontinuities in passive BE-periphrases

Lieven Danckaert (GIST, UGent)
In this paper, I argue against the widespread view that the Romance analytic passive present tenses
with a BE-auxiliary, like French il est aim 'he is loved', have developed from the analytic Latin
passive perfect of the type amatus est 'he was loved' (see, among many others, Harris 1978: 188;
Winters 1984: 451; Hewson & Bubenk 1997: 315). The usual explanation offered by proponents
of this view is that Latin amatus est was ambiguous between a pure perfective past tense reading
('it was the case the was loved') and a resultative (adjectival), present tense reading ('it is now the
case that he is loved, as the result of a past action'). However, in addition to there not being any
independent evidence in favour of this putative meaning shift, the above explanation also fails to
take into account a set of word order facts that until now have not been noticed in the literature.
It is received wisdom that the evolution from Early to Late Latin witnessed a rise of headinitial syntax (see e.g. Bauer 1995). However, corpus data reveal that this is only partially correct.
As shown in Graph 1, there is a significant rise of head-initial TP's in clauses with a modal
auxiliary and an infinitival complement, but not in clauses with a BE-auxiliary and a (deponent)
past participle, which display the opposite tendency:
Average % of AuxV(P)
BE + past participle modal + infinitive
Classical Latin (Cicero, 85 BC - Gaius, 180 AD):
29,40%
37,50%
Late Latin (Egeria, 380 AD - Iordanes, 550 AD):
8,10%
78,02%
Table 1: Evolution of Aux-V(P) order in clauses with a modal and a BE- auxiliary

In other words, the word order development of Latin BE-periphrases is such that it moves away
from the 'Aux-V' order characteristic of (early) Romance languages (cf. amatus est vs. il est aim).
In addition, it has been pointed out that the replacement of (classical) Latin passive paradigms
happens in different stages: analytic passive perfects with a present tense BE-auxiliary (amatus est
'he was loved') are replaced by periphrases with a perfective auxiliary (fuit amatus) (cf. de Melo
2012) before the synthetic present passive (amatur 'he is (being) loved') is lost and (productively)
replaced by the Romance periphrases of the il est aim type (Muller 1924; Herman 2002).
In order to capture these two facts (viz. (i) the persistence of VAux word order and (ii) the
relative chronology of the changes), I will suggest that the Latin amatus est perfects disappeared
from the language, via a stage of verb incorporation (in the syntactic sense of Baker 1988). Only
later, when the synthetic present passives (amatur) are being lost, a new present tense analytic
paradigm (est amatus) replaces them. I will suggest this new paradigm is formed by analogy with
the existing fuit amatus periphrases, which themselves pattern with modal auxiliaries in appearing
predominantly in the order 'AuxVP' (as do the newly formed est amatus present tense passives).
Despite appearances, these innovative forms are unrelated to the Latin passive 'VAux' perfects.
References Baker, M. 1988. Incorporation. Chicago U.P. Bauer, B. 1995. The emergence
and development of SVO patterning in Latin and French. O.U.P. de Melo, W. 2012.
Kurylowicz's first 'law of analogy' and the development of passive periphrases in Latin. In: Laws
and rules in Indo-European, 83-101 Harris, M. 1978. The evolution of French syntax. London
Herman, J. 2002. La disparition du passif synthtique latin, ER 24, 31-46 Hewson, J. & V.
Bubenik. 1997. Tense and aspect in Indo-European languages. Benjamins Muller, H. 1924. The
passive voice in Vulgar Latin, The Romanic Review 15, 68-93 Winters, M. 1984. Steps toward
the Romance passive inferrable from the Itinerarium Egeriae, RomPh 37, 445-54

Serena Danesi University of Bergen

Between the historical languages and the reconstructed language:
The verbal adjective + dative construction in Latin, Ancient Greek, Sanskrit and Avestan

In existing historical linguistic research, syntactic reconstruction is generally regarded with

skepticism because, unlike with the lexicon, syntax does not provide well-defined entities to
compare. The Comparative Method presupposes a comparison only of items with similar form and
meaning. Since syntactic structures, allegedly, lack a meaning component, identifying cognate
entities appears to be an unattainable enterprise, and, consequently, syntactic structures have been
excluded from a historical-comparative framework (Jeffers 1976, Winter 1984 among others). In this
presentation, I aim to show, in the same vein as Bardal & Eythrsson 2012, that syntactic
reconstruction may be successfully accomplished; I demonstrate this with an analysis of verbal
adjectives across four different Indo-European languages, namely Sanskrit, Avestan, Ancient Greek
and Latin, as exemplified in (1 4).
(1) y

stotbhyo

hvyo

sti

yman

Sanskrit

REL.NOM singers:DAT to-be-invoked:NOM be:PRS3sg. sacrifice:LOC

He who is to be invoked by the singers at the sacrifice (RV. I, 33, 2d)

(2) tca.
voh
y. ..

Avestan

these:NEUT.PL goods:NEUT.PL which:NEUT.PL

friiatuua.

naire.

aaone.

to-be-worshipped:NEUT.PL man:DAT faithful:DAT

(We sacrifice to) all these good things, which the faithful man must worship (Yt. 13. 153)
(3) hmn pnta
poita
Ancient Greek
us:DAT

all:ADJ.NEUT.PL.NOM

to-be-done:NEUT.PL.

Everything must be done by us (Xen. An. 3, 1, 35)

(4) desperanda
tibi
concordia

Latin

to-be-despaired:NOM you:DAT harmony:NOM

You must not despair of harmony (Iuv. 6.231)

These adjectival forms (hvyo, friiatuua, pota, desperanda) qualify an entity, which should experience
the event expressed by the verbal root from which the adjectives derive. This entity, forming a
patient-relation with the verbal adjective, is in the nominative, while the entity carrying out the
event is expressed with the so-called dative of agent.
Developing an idea, originally suggested by Hettrich (1990: 64ff), I argue that this particular
combination mirrors a construction of Indo-European inheritance. The proposal will be advanced
with the aid of the theoretical framework of Construction Grammar in which, the basic unit of
language is the Construction, i.e. a formfunction correspondence, where no principled distinction
between lexical items and complex syntactic structures is assumed. I will provide evidence that the
structures investigated show similarities at a morpho-syntactic level (DAT VB.ADJ (be) NOM), at a
semantic level (modal meaning and low degree of transitivity), and also, to a certain extent, at an
etymological level. In sum, they constitute formmeaning pairings, available as units of
comparanda, as required by the Comparative Method, and can thus successfully be reconstructed
for a common proto-stage.
With regard to the semantics of the construction, I will show that the notion of participantexternal modality (van der Auwera & Plungian 1998) and the modal meaning entailed by the
verbal adjective are crucial for determining the argument structure of the construction. I will argue,
therefore, that an analysis involving a dative subject, a nominative object and a modal reading
instead of the agentive/passive reading, is better equipped to account for the verbal adjective +
dative construction than the standard analysis. Hence, this presentation offers a contribution to
current research on syntactic reconstruction of oblique subjects in Indo-European (Bardal &
Eythrsson 2009), as well as to recent typological studies on the close relationship between noncanonical case marking and modality (Narrog 2010).

How to change but stay the same

Clause coalescence in Papua New Guinea
Don Daniels
dondaniels@umail.ucsb.edu
Historical studies of complex constructions have focused mainly on subordination, and those
dealing with other complex structures have largely studied the origins of clause linkers. This study
investigates the diachronic behavior of complex non-subordinate structures by examining two
innovations that have happened to the clause chaining system of the Sogeram languages of Papua
New Guinea. In the first, sentence-final morphology, like the Manat suffix in 1SG.PST in example
(1), was reanalyzed as sentence-medial morphology with different-subject switch reference
meaning, as illustrated in (2).
(1)

(2)

Manat
Ini-n
p
mu kai aih-in=a
NEAR.DEM-ACC house SPEC LOC come-1SG.PST=INT
Ive come to another place here.

(authors field notes)

Manat
Kis=a,
vana
in-n
md-apar-in
mgu-g
therefore=INT speech NEAR.DEM-ACC plant-throw-1SG.DS
go.down-3SG.PST
Therefore, I brought up this matter again (lit. I erected it and it went down).
(authors field notes)

The second innovation involved a construction that employed two chained verbs to express
physical manipulation of an object, like the Apal example in (3)the first verb always referred to
the agent, while the second referred to the patient. In Nend, these common two-verb collocations
were reanalyzed as constituting a single compound verb stem with transitive meaning (4).
(3)

(4)

Apali
Ik-l
mg-ci
sukuala-c-in,
maci
cut-1SG.DS
move.down-3SG.DS
finish-FPST-1SG sago
I cut the sago down and finished it.

aga-d
DEF-OBJ
(Wade n.d.)

Nend
D-e-m
asamzgag
ha-dh
av-z-gw-em-or
walk-SS-CONT Asamzgag MID.DEM-setting throw-3SG.DS-go.inside-YPST-1PL
We walked and there at Asamzgag we threw it down.
(Harris n.d.)

The behavior of these clause chains affirms the cross-linguistic tendency towards increased
integration in complex structures, but also suggests that not all complex structures integrate in the
same way. In particular, the impulse towards subordination, seen as the endpoint on Hopper &
Traugotts cline of clause combining (2003:177), appears to be absent from the Sogeram data.
Indeed, the behavior of Nend suggests that the relationship between two clauses can remain
unchanged even as the construction in which they are found undergoes the change from a clause
chain to a single verb stem.
References
Harris, Kyle. n.d. Nend texts. Electronic files, Pioneer Bible Translators.
Hopper, Paul J. & Elizabeth Closs Traugott. 2003. Grammaticalization, 2nd ed. Cambridge: Cambridge
University Press.
Wade, Martha. n.d. Apali texts. Electronic files, Pioneer Bible Translators.

Nynke de Haas
CLS, Radboud University Nijmegen

The Northern Subject Rule in Northern Middle English: morphosyntactic conditions on verbal
inflection beyond subject type and adjacency
The Northern Subject Rule (NSR) is a pattern of variation in which verb endings are typically constrained by
two factors: subject type (personal pronoun versus NP) and adjacency (presence/absence of subject-verb
adjacency). The pattern applies variably in some contemporary dialects, but has traditionally been seen as
categorical in Northern Middle English (cf. Mustanoja 1960:481-482; LALME I:554). In modern dialects with
the NSR, personal pronoun subjects (I, we, you and they) trigger a zero ending on the verb (-) when it is
immediately adjacent to the subject, as in (1a). Elsewhere when the pronoun subject and the finite verb are
not adjacent, or when the subject is a noun phrase the verb ends in -s (1b,c) (cf. Pietsch 2005).
(1)

a. they sing
b. birds sings
c. they sing and dances

In Middle English, the NSR occurred with a number of different verb endings: -e and -en could occur instead
of - (cf. also Mustanoja 1960:481-482; LALME I:554,), and in the East Midlands, -th instead of -s (McIntosh
1983). I will present data from a detailed study of the NSR and related variation in Northern and
Northern/Eastern Midlands dialects of early Middle English (based mainly on data from the LAEME corpus;
cf. de Haas 2011), and relate it to a similar study of late Middle English data (from the MEG-C corpus and the
MEG Corpus Northern Documents Database).
I will show that in early Middle English, like in later dialects, many varieties with a subject effect did
not display an adjacency effect. Instead, even verbs with a nonadjacent pronoun subject often ended in -/e/en.
In addition, I will give a more fine-grained analysis of the late Middle English data, which has been analysed
for a variety of syntactic factors, to shed light on two questions. First, what syntactic structures led to
nonadjacency of subject and verb, and how did they influence verbal morphology? And second, to what extent
did earlier varieties show the same additional conditioning factors reported for present-day Northern and Scots
varieties (Buchstaller et al.2013)?
References
Buchstaller, Isabelle, Karen P. Corrigan, Anders Holmberg, Patrick Honeybone and Warren Maguire. (2013). T-to-R
and the Northern Subject Rule: questionnaire-based spatial, social and structural linguistics. English Language
and Linguistics 17:85-128.
de Haas, Nynke K. (2011). Morphosyntactic variation in Northern English: the Northern Subject Rule, its origins and
early history. Dissertation. Utrecht: LOT. Available online via <www.lotpublications.nl>.
LAEME: Laing, Margaret, & Roger Lass (2008-). A linguistic atlas of early Middle English 11501325. Version 1.1.
Online at http://www.lel.ed.ac.uk/ihd/laeme1/laeme1.html. Edinburgh: University of Edinburgh.
LALME: McIntosh, Angus, M.L. Samuels, Michael Benskin, with Margaret Laing & Keith Williamson (eds.). (1986). A
linguistic atlas of late mediaeval English. Vol. I. Aberdeen: Aberdeen University Press.
McIntosh, Angus (1983). Present indicative plural forms in the later Middle English of the North Midlands. In Douglas
Grey & E.G. Stanley (eds.), Middle English studies: presented to Norman Davis in honour of his seventieth
birthday. Oxford: Clarendon. 235244.
MEG-C = The Middle English Grammar Corpus, version 2011.1, compiled by Merja Stenroos, Martti Mkinen, Simon
Horobin,
&
Jeremy
Smith,
March
2011,
University
of
Stavanger.
<http://www.uis.no/research/culture/the_middle_ english_grammar_project/>. Accessed: 28 November 2012.
Middle English Grammar Corpus Northern Documents Database. (2009). Compiled by Merja Stenroos, Martti Mkinen,
Kjetil V. Thengs, & Nedelina V. Naydenova. Accessed 3 September 2012.
Mustanoja, Tauno F. (1960). A Middle English Syntax. Part I: Parts of Speech. Helsinki: Socit Nophilologique.
Pietsch, Lukas. (2005). Variable Grammars: Verbal Agreement in Northern Dialects of English. Tbingen: Niemeyer.

Language Dynamics in the Lower Congo region of Central-Africa:

A Phylogenetic Approach
Gilles-Maurice de Schryver*, Rebecca Grollemund & Koen Bostoen*
* KongoKing Research Group, Department of Languages and Cultures, Ghent University, Belgium
Evolutionary Biology Group, University of Reading, UK
The Lower Congo region occupies a pivotal position in wider Bantu history. It lies at the junction of two major Bantu
subgroups, i.e. North-West Bantu and West-Bantu, not only geographically, but also historically. Several studies have
located a secondary nucleus of early Bantu expansion in this zone (Heine et al. 1977). It is very poorly understood,
however, how and when Bantu speech communities peopled the Lower Congo area itself.
The languages that have been at the core of the Lower Congo region since at least the 15th century belong to the H10Kikongo group in Guthries referential classification of the Bantu languages (Guthrie 1971). What is commonly called
Kikongo can be considered as a large dialect continuum covering considerable parts of four Central-African countries,
i.e. Angola, Gabon and the two Congo. Its history is closely linked with the rise of the Kongo Kingdom and related
polities that emerged since the first centuries of the second millennium AD, long before the first Europeans arrived in the
area. The present-day linguistic landscape in the Kongo area suggests that both political centralization and economic
integration had a considerable impact on language evolution in the region. Only if we succeed in a better understanding
of these relatively recent convergence phenomena, can we factor them out to come to a better-founded insight in early
Bantu dispersals in this region and the Bantu expansion more generally.
In order to elucidate the external and internal classification of the Kikongo language group, we present in this paper the
first results of a new phylogenetic approach of the so-called West-Coastal Bantu languages (Vansina 1995), consisting
of the languages belonging to Guthries zone H and groups B40-80. It is based on the Tervuren list of 92 lexical items
belonging to basic vocabulary (Bastin et al. 1999), for which we identified cognate sets by means of the comparative
method. In order to build our trees, we applied to our data a Bayesian inference of phylogeny, using Markov chain Monte
Carlo (MCMC). Apart from lexicostatistical data already available from previous studies, such as Bastin et al. (1999) and
Grollemund (2012), two types of new datasets are incorporated in this study: (a) vocabulary from poorly documented
Kikongo varieties obtained through recent fieldwork in the Lower Congo, and (b) diachronic lexical data from historical
sources since the 17th century, such as the Vocabularium Latinum, Hispanicum, e Congense (Van Gheel 1652), the
oldest remaining Bantu dictionary. None of these data have ever been considered in a phylogenetic study.
The preliminary results regarding the external classification of Kikongo support the east-out-of-west model of Bantu
migration (Heine et al. 1977). The preliminary results regarding the internal classification of Kikongo indicate a language
frontier between the western Kikongo varieties, which are more closely related to West-Coastal Bantu languages further
north, and the other Kikongo varieties which manifest closer lexical affinities with West-Coastal Bantu languages further
East. This phylogenetic evidence from basic vocabulary will be checked against evidence from phonological and
morphological change obtained through the Comparative Method in order to discover correspondences and mismatches
indicative of different strata of language history in the Lower Congo.
References
Bastin, Y., A. Coupez and M. Mann. 1999. Continuity and Divergence in the Bantu Languages: Perspectives from a
Lexicostatistic Study. (Annales, Srie in-8, Sciences humaines). Tervuren: Royal Museum for Central Africa.
Grollemund, R. 2012. Nouvelles approches en classification: Application aux langues bantu du Nord-Ouest. Lyon:
Universit Lumire Lyon 2, PhD.
Guthrie, M. 1971. Comparative Bantu: an introduction to the comparative linguistics and prehistory of the Bantu
languages. Volume 2: Bantu prehistory, inventory and indexes. London: Gregg International.
Heine, B., H. Hoff and R. Vossen. 1977. Neuere Ergebnisse zur Territorialgeschichte der Bantu. In W. Mhlig, F.
Rottland & B. Heine (eds.), Zur Sprachgeschichte und Ethnohistorie in Afrika, 57-72. Berlin: Dietrich Reimer
Verlag.
Van Gheel, J. 1652. Vocabularium Latinum, Hispanicum, e Congense. Ad Usum Missionarior transmittendor ad Regni
Congo Missiones. Ms. Rome: National Central Library, Fundo Minori 1896, MS Varia 274.
Vansina, J. 1995. New Linguistic Evidence and the Bantu Expansion. Journal of African History 36: 173-195.

From 'habitually going' to 'maybe: grammaticalization and lexicalization of an epistemic sentence

adverb in Swahili.
Maud Devos (maud.devos@africamuseum.be)
Gilles-Maurice de Schryver(gillesmaurice.deschryver@Ugent.be)
In comparison to tense and aspect, modality has not been a popular research topic in Bantu
linguistics. The overall impression one gets from specialized as well as more general typological
studies is that the subjunctive plays a dominant role in the expression of Bantu modal categories.
Some recent dedicated publications (Devos 2008, Bostoen, Mberamihigo & de Schryver 2012)
emphasize the importance of modal auxiliaries in Bantu languages. This paper wants to add to this
emerging field in Bantu studies by looking at the origin and development of a modal adverb huenda
maybe expressing epistemic possibility in Swahili.
In the utterances in (1) and (2) huenda is used as a lexical verb expressing habitually going
and as a sentence adverb denoting epistemic possibility, respectively.
(1)
baada ya masomo Hasani huenda maktabani
after class Hasani usually goes to the library
(2)
huenda matatizo yao yatapungua akianza kupata mshahara
maybe their problems will grow less when he starts to earn wages
The usage patterns of huenda in the Helsinki Corpus of Swahili suggest the following semantic and
syntagmatic developments.
A. huenda as a lexical verb
habitually going
B. huenda + semantic main verb in consecutive epistemic possibility (might well)
C. huenda as a sentence adverb
epistemic possibility (maybe)
The first aim of the paper is to verify the hypothesized evolution with the help of a diachronic Swahili
corpus (currently under construction at Ghent University) as well as with comparative data from
Swahili dialects.
Next we take a closer look at the semantic changes involved. Although habitually going is not
a common source for epistemic possibility, a metonymically motivated change involving the
conventionalization of the implicature that someone who usually does something might well do it is
not unlikely. However, the complete bleaching of the lexical verb go is remarkable (see Hopper
1991 on persistence). We discuss whether the use of -enda go (to) should be attributed to the oft
assumed generality or basicness of verbs expressing go (Bybee et al. 1994) or whether the specific
semantics of -enda do play an important role in the change towards epistemic possibility. Important
in this respect is the occasional use of other verbs of motion in constructions of the type B.
Finally, we describe the evolution of huenda in Swahili as being on the interface of
grammaticalization and lexicalization (and pragmaticalization), with the change from A to B being
reminiscent of grammaticalization processes and the change from B to C being more akin to
lexicalization (or pragmaticalization) (Norde 2007, Diewald 2011).
References
Bostoen, K., F. Mberamihigo & G.-M. de Schryver. 2012. Grammaticalization and Subjectification in
the Semantic domain of Possibility in Kirundi (Bantu, JD62). Africana Linguistica 18: 5-40.
Bybee, J., R. Perkins & W. Pagliuca. The evolution of grammar. Chicago & London: The University of
Chicago Press.
Devos, M. 2008. The expression of modality in Shangaci. Africana Linguistica 14: 3-35.
Hopper, P. 1991. On some principles of grammaticalization. In E. Traugott & B. Heine (eds),
Approaches to grammaticalization, vol. 1. 17-35. Amsterdam: John Benjamins.
Diewald, G. 2011. Pragmaticalization (defined) as grammaticalization of discourse functions.
Linguistics 49, 2: 365-390.
Norde, M. 2007. Epistemic adverbs at the interface of lexicalization and grammaticalization. Paper
presented at TIN-dag, University of Utrecht, February 3, 2007.

Prehistoric Language Contact in Northeast Asia in a Single Lexical Item

Thomas Dougherty
University of Hawaii at Mnoa
A small number of loanwords shared between Japanese, Ainu, and Nivkh are attested (Vovin
1993, Tsumagari 2010, Vovin 2012). However, at least the borrowings from Nivkh into Japanese
via Ainu all appear to be recent, such as is the case for 'reindeer' (Japanese tonakai <Sakhalin
Ainu tunakay< Nivkh tlangi~c'olngi). I argue, based on morphological evidence, as well as
textual and distributional that there is at least one loan which is likely to be prehistoric: the Nivkh
verb hup-'to tie', which was borrowed into Old Japanese (OJ) via an Ainu intermediary.
Previous work on contact between Ainu and Nivkh has demonstrated a number of loansin both
directions, including morphological borrowings like the indefinite object marker i-,which is fully
productive in Ainu, but only applies to a subset of verbs in Nivkh (Vovin 1993:160, Shiraishi
2006:99-103). When theindefinite object marker i-prefixes the Nivkh verb hup-'to tie', it becomes
yup-,based on standard phonotactic process (Shiraishi 2006:102).Ainu has a yup-'to tie, to make
tight', essentially identical to the latter Nivkh form (Hattori 1964:293). That this is aborrowing
between Ainu and Nivkh was suggested by Vovin (1993), but no directionality was posited
(1993:160). As the Nivkh form is morphologically complex, and the Ainu form simplex, we can
be relatively certain that the directionality was from Nivkh into Ainu.
Recent workon the textual ancestors of Japanese, especiallybut not limited tothe group of
dialects spoken in the modern Kantregion, Eastern Old Japanese, also demonstrates plausible
borrowings from an ancestor of the Ainu languages and OJincluding at least one morphological
borrowing, as well as borrowings from OJ into Proto-Ainu (Vovin 2012:11-12, Vovin 2010:3435). In the Eastern and Western dialects of OJ, there is a doublet for the word 'to tie': musuNp-and
yup-. While cognates of both are attested throughout Japonic, the philological evidence is key, as
many modern varieties of Japonic use cognate forms of OJ yup-only in a much more restrictive
sense, meaning 'to tie up one's hair, to arrange one's hair' (Hirayama 1992:4978-4982, 52515253).
By adding another plausible borrowing between Nivkh and Ainu, and between Ainu and Japonic,
and demonstrating that the likely scenario is that this is a Nivkh loan into Ainu, and then into
Japanese, we hope to further illustrate three points: the existence of prehistoric language contact
in the Japanese archipelago, its nature as long distance contact, and the relatively high amount, as
evidenced by morphological borrowings.
Keywords: Ainu; Nivkh; Japonic; language contact; linguistic prehistory; borrowing.
References:
Hattori,Shir, ed. 1964. Ainugo Hgen Jiten[A Dialect Dictionary of Ainu].Tokyo:Iwanami
Shoten.
Hirayama, Teruo, ed. 1992. Gendai Nihongo Hgen Daijiten [A Large Dictionary of Modern
Japanese Dialects]. Tokyo: Meiji Shoin.
Shiraishi, Hidetoshi. 2006. Topics in Nivkh Phonology.PhD diss. University of Groningen.
Tsumagari, Toshir.2010. Long Journey of Walrus: A Linguistic Survey. In: Hopp Jinbun
Kenky[Northern Cultural Research]3, pp. 45-57.
Vovin, Alexander. 1993. A Reconstruction of Proto-Ainu.Leiden: E.J. Brill.
Vovin, Alexander. 2010. Koreo-Japonica.Folkestone: Global Oriental.
Vovin, Alexander. 2012. Man'ysh: Book 14.Folkestone: Global Oriental.

Preverbs in the history of Greek:

On predicting and changing idiomatic vs. compositional readings
Gaberell Drachman1, Nikolaos Lavidas2 & Angeliki Malikouti-Drachman1
1University of Salzburg, 2Aristotle University of Thessaloniki
gaberell.drachman@sbg.ac.at, nlavidas@auth.gr
The historical (like the synchronic) semantics of derivational morphology is usually assumed to be a
matter of postulating whether a given affix is understood idiomatically or compositionally, in a given
linguistic environment, at a given period.
In this paper we maintain that there is no need to postulate such meaning distinctions, at the
level considered. In fact, we attempt to distinguish between them on the basis of auxiliary hypothesis
fundamentally involving the place of morphology in the grammar.
Our example language is Greek, the data coming from the domain preverb-plus-verb for both
Classical and Koine Greek. We thus give emphasis to the historical dimension of the problem,
comparing diverse explanations for the various historical changes we observe: e.g. the rise and fall of
individual preverbal affixes, the (re)ordering among such affixes, and the historically attested semantic
shifts involved (see in 1a and b the contrast between the compositional and the idiomatic meaning of
sumperiphromai in Classical Greek; the compositional meaning is lost after the Hellenistic Greek
period).
(1) Classical Greek
a. sum-peri-phromai
together-round-carry.PASS.PRS.1SG to be carried round together
b. sumperiphromai
accomodate.PASS.PRS.1SG to accommodate or adapt oneself to circumstances (later: to
behave)
This variation was earlier accounted for under a syntax vs lexicon analysis (see, for example, Wasow
1977, among others). Alternatively, we reject the idea that there is separation in creativity (syntax vs.
lexicon), and we adopt a single source account of derived forms (Marantz 2001, Borer 2005). We
propose that change in case of preverbals is predominately a drift from transparency/compositionality
to opacity/idiosyncrasy. Classical Greek allows both types of affix-attachment by Merge (for the
prefixes sun- and peri-): (a) Affix + Root (idiomatic reading) and (b) Affix + Post-Cat(egory)
(compositional reading) (Drachman 2008). The loss of the compositional readings means a change in
the process of affix-attachment by Merge.
The proposed norm for change is, at a first approach, the loss of the compositional reading,
i.e., of the option of the Affix + Post-Cat Merge but not the reverse (there are no cases of loss of the
Affix + Root Merge to retain the Affix + Post-Cat Merge). We further argue that this contrast between
the types of affix-attachment should be evident in the differences between idiomatically-rich preverbs
and compositionally-rich suffixes. Hence, the changes in the verbal suffixes (for instance,
mediopassive suffixes in Greek) are not in the same direction and cannot be analyzed as a loss of the
Affix + Post-Cat process.

References
Borer, Hagit. 2005. Structuring Sense. Volume II: the Normal course of Events. Oxford: Oxford University
Press.
Drachman, Gaberell. 2008. Meaning variation and change in Morphology. In N. Lavidas, E. Nouchoutidou
& M. Sionti (eds.), New Perspectives in Greek Linguistics, 1-34. Newcastle: Cambridge Scholars
Publishing.
Marantz, Alec. 2001. Words (unpublished ms; WCCFL XX).
Wasow, T. 1977. Transformations and the Lexicon. In P. Culicover, T. Wasow & J. Bresnan (eds.), Formal
Syntax, 327-360. New York: Academic Press.

Contrastive Vowel Features in West Germanic and Old English

B. Elan Dresher, University of Toronto
The insight that phonological change may involve a reorganization of the contrasts of a
language goes back to Jakobson (1931), who argued for a structuralist phonemic approach
(see Salmons & Honeybone to appear). Hogg (1992) provides a number of interesting
illustrations of instances where his Neogrammarian predecessors have been unable to give a
satisfactory account of developments in early Old English because they lacked a phonemic
perspective. I show that some of these insights are not expressible in a theory that requires
full specification of underlying segments. These insights can be recaptured, however, if
underlying forms are specified only for contrastive features.
One example concerns the prehistory of early OE long . Since the corresponding vowel
in Proto-Germanic is assumed to have also been *, Wright & Wright (1925) propose that
persisted into the Old English period. Against this view is historical and comparative
evidence which appears to show that it was a back vowel, *a, in West Germanic. Most other
writers therefore posit that P-G * retracted to WGmc *a, then fronted again to OE *
when not before a nasal. Hogg (1992: 613) argues that the alleged shift of P-G * to
WGmc *a and then back to in Old English and Old Frisian emerges as an artefact of a
non-phonemic theory, once we consider the contrasts in play at each stage. He proposes that,
in the WGmc dialects from which Old English developed, *// is the only low long vowel
and there is no front/back contrast in operation. From the structural point of view, therefore,
the vowel as it develops in WGmc may be considered to be neutral in this last respect, that is,
*/a/, whatever its precise phonetic character. This suggests that */a/ (as well as short low
*/a/) should not be specified as being either [+back] or [back]; thus, its pronunciation could
have remained [] all along, while its contrastive feature specifications changed.
We can translate Hoggs insight into an explicit theory if we posit that contrastive specifications are assigned by ordering features into a hierarchy (Dresher 2009). On the assumption
that active features are contrastive (the Contrastivist Hypothesis, Hall 2007), phonological
activity can serve as a heuristic to ordering the features. One way of ordering them so that the
low vowels have no specification for [front/back] is shown in (1) (the length contrast is not
shown). This ordering, [low] > [back] > [high] > [long], requires [round] to be absent. Purnell
& Raimy (to appear) observe that this is supported by Lasss (1994) observation that rounding is non-distinctive in the West Germanic vowel system. Interestingly, there is evidence for
an active [round] feature in Old English, which had a different set of vowel contrasts; I will
argue that the OE order is [back] > [round] > [high] > [low] > [long]. Like the dog that didnt
bark, the absence of evidence for active WGmc [round] requires an explanation, which is
provided by the analysis in (1). It is significant that the evidence bearing on the activity and
inactivity of different WGmc vowels converges on the tree in (1).
(1) Contrastive hierarchy: WGmc vowels
vowels
wo
[+low]
[low]
/a()/
ei
[+back]
[back]
ty
ty
[+high] [high] [+high] [high]
/u()/
/o()/
/i()/
/e()/
Dresher, B. E. 2009. The contrastive hierarchy in phonology. Cambridge: CUP.
Hall, D. C. 2007. The role and representation of contrast in phonological theory.
Doctoral dissertation, U. of Toronto.

Hogg, R. M. 1992. A grammar of Old English. Vol. 1:

Phonology. Oxford: Blackwell.
Jakobson, R. 1931. Prinzipien der historischen Phonologie.
TCLP 4: 247267 (Copenhagen).
Lass, R. 1994. Old English: A historical linguistic companion.
Cambridge: CUP.
Purnell, T. & E. Raimy. to appear. Distinctive features, levels
of representation and historical phonology. In P. Honeybone & J. Salmons, eds., The handbook of historical
phonology. Oxford: OUP.
Salmons, J. & P. Honeybone. to appear. Structuralist historical
phonology: Systems in sound change. In Honeybone &
Salmons.
Wright, J. & E. M. Wright. 1925. Old English grammar (third
edition). London: OUP.

ROOF LANGUAGES AND THE EUROPEAN PERFECTS

BRIDGET DRINKA, UNIVERSITY OF TEXAS AT SAN ANTONIO

Johan van der Auwera (1998), in his important edited volume in the EUROTYP Series, makes
two especially noteworthy observations:

the area of western Europe where languages adhere to an array of similar grammatical
patterns, often referred to as Standard Average European (SAE), coincides fairly
closely with the area ruled by Charlemagne in the 9th century, hence, his name for it,
the Charlemagne Sprachbund
Standard Average European should more accurately be split into Standard Average
Eastern European and Standard Average Western European (Kortmann 1998b:
530-535).

In this paper, the validity of both of these claimsthe existence of the Charlemagne
Sprachbund and the division of SAE into East and Westis explored in detail, through a
close examination of the distribution of the periphrastic perfects across Europe. In western
Europe, for example, perfects are formed with HAVE and BE auxiliaries and past passive
participles, and show a layering of nuclear, core, and peripheral distribution like that noted by
Haspelmath (1998, 2001) for other structural features, and thus provide solid evidence for the
accuracy of the Charlemagne Sprachbund as a construct (Drinka, forthcoming). Within
eastern Europe, on the other hand, perfects are built especially with BE auxiliaries plus past
active participles, and show a different distribution, one reminiscent of that noted by
Kortmann (1998a, 1998b) for adverb clauses. While many factorsinternal as well as
externalcan be shown to have played a role in this development, what emerges as crucial in
the west / east split is the influence of the roof languages, Latin and Greek respectively, and
the early cultural divide connected with Roman Catholicism and Greek Orthodoxy.

Drinka, Bridget. (Forthcoming). Sources of Auxiliation in the Perfects of Europe. In: Freek Van de
Velde, Hendrik De Smet, and Lobke Ghesquire (eds.). Multiple source constructions in
language change. Studies in Language Series.
Haspelmath, Martin. 1998. How young is Standard Average European? Language Sciences 20: 27187.
Haspelmath, Martin, Ekkehard Knig, Wulf Oesterreicher, and Worfgang Raible. 2001. Language
typology and change. Sprachtypologie und sprachliche Universalien. La typologie des langues
et les universaux linguistiques. Vol. 2. Berlin / New York: Walter de Gruyter.
Kortmann, Bernd. 1998a. The evolution of adverbial subordination in Europe. In: Monika Schmid,
Jennifer Austin, and Dieter Stein (eds.). Historical Linguistics 1997. Amsterdam /
Philadelphia: Benjamins. 213- 28.
Kortmann, Bernd. 1998b. Adverbial subordinators in the languages of Europe. In: van der Auwera
(ed.). 457-561.
van der Auwera, Johan (ed.). 1998. Adverbial constructions in the languages of Europe. (Empirical
Approaches to Language Typology, EUROTYP 20-3) Berlin / New York: Mouton de Gruyter.

Being and becoming in Old Church Slavonic:

Grammatical and constructional profiles of byti
Hanne Eckhoff Laura Janda Tore Nesset
University of Troms
We examine the behavior of the Old Church Slavonic (OCS) verb byti be on the basis of data from
the Pragmatic Resources in Old Indo-European Languages corpus (PROIEL, foni.uio.no:3000/). Byti is
by many measures a unique verb. It is the most frequent verb in OCS (accounting for 13.8% of all
OCS verb attestations in PROIEL), it is used as an auxiliary and has entire subparadigms that other
verbs lack (future, subjunctive), it is historically derived from two different verbs (compare Latin esse,
fieri), and it appears in a large range of grammatical constructions. van Schooneveld (1951) claims
that at the time of OCS, byti should still be considered two verbs rather than one, an aspectual pair
rather than a single verb.
Taking inspiration from work on behavioral profiles (Gries and Divjak, 2009), we employ two
measures, each of which can be understood as a subset of behavioral profiles: grammatical profiles
and constructional profiles. A grammatical profile is the relative frequency distribution of the
grammatical forms of a word, and a constructional profile is the relative frequency distribution for the
grammatical constructions that a word appears in. We use these two measures to compare byti with
other OCS verbs and to investigate the constructions that characterize bytis use.
If one looks at the grammatical profiles of verbs attested >20 times in PROIEL, one finds a
statistical pattern that is remarkably similar (over 95% identical) to Dostl (1954)s classification of
OCS verbs according to aspect. Adding byti into the formula makes surprisingly little difference. In
terms of its grammatical profile, it looks much like an ordinary imperfective verb; its closest neighbors
are znati know and piti drink. When we split its forms and posit two verbs, however, they come out
as quite separate on the the axis that appears to separate verbs by aspectual properties in the same
analysis. This suggests that we are dealing with an aspectual pair, but semantically it is not clear that
this makes sense.
On syntactic grounds we identify six primary constructions of byti: copula, position, auxiliary
(with l-participle), auxiliaroid (with other participles), existential and happen (it happened that/when).
The constructions are quite distinct in terms of grammatical profile.
We conclude that it makes most sense to consider byti to be a single verb in OCS. The
grammatical constructions this verb appears in form a network of relatively distinct, yet related
functions in which the copula and existential constructions play a central role.

References
Antonin Dostl. Studie o vidovm systmu v starosloventine. Sttn pedagogick nakladatelstv,
Prague, 1954.
Stefan Th. Gries and Dagmar Divjak. Behavioral profiles: a corpus-based approach towards cognitive
semantic analysis. In Vyvyan Evans and Stephanie S. Pourcel, editors, Behavioral profiles: a
corpus-based approach towards cognitive semantic analysis. John Benjamins, Amsterdam,
2009.
C. H. van Schooneveld. The aspect system of the Old Church Slavonic and Old Russian verbum
finitum byti. Word, 7, 1951.

The diachrony of light verbs in English and Dutch

Marion Elenbaas
Leiden University
Light verbs are widely attested cross-linguistically (see e.g. Mohanan 2006). Often-cited
examples are take, give, make, have, and do, when found in combination with a predicative
complement such as a deverbal nominal, as in (1).
(1) a. He took a sip of her coffee.
b. She gave him a push.
The verbs in (1) are called light verbs because they lack the lexical semantic content of their
full verb counterpart (compare the light verb take in take a sip to the full verb take in take a
biscuit).
One of the central issues raised by light verbs concerns their categorial status.
Semantically, they are like auxiliaries in lacking full lexical meaning: just as auxiliaries have
to combine with a full (lexical) verb, light verbs cannot occur without a predicative
complement. Morphosyntactically, however, light verbs are like full verbs in that they are
form-identical to full verbs and show the same distribution. The fact that light verbs are
ambiguous between full verbs and auxiliaries has led to the idea that light verbs reflect a
transitional stage on a grammaticalisation cline from full verb to auxiliary. Hopper & Traugott
(1993), for example, argue that Hindi light verbs are an optional intermediate stage between
full verb and auxiliary on a grammaticalisation cline full verb > (vector verb) > auxiliary >
clitic > affix (from Hopper & Traugott 1993:108). In Hopper & Traugott (2003) this view is
reconsidered, and they state that it is not clear that auxiliaries developed from light verbs. This
view is closer to the one advocated by Butt & Lahiri (To appear), which holds that light verbs
in Indo-Aryan languages are not on the same grammaticalisation cline as auxiliaries and are
historically a dead end. In this view, light verbs are synchronic variants of full verbs.
In this talk I will investigate the diachrony of light verbs in two Germanic languages,
English and Dutch. I will do so by analysing corpus data, focusing on Middle and later
English and Dutch. Two examples, from Middle English and Middle Dutch respectively, are
given in (2).
(2) a. And there they hadde a grete batayle whythe the Quene,
and there they had a great battle with the Queen (cmgregor, 212.1922)
b. Doe gaf si enen groten krijt (493:37, Sente Lutgart, 1265-1270)
then gave she a big cry
Then she gave a big cry
I will show that the diachronic paths of light verbs and auxiliary verbs are distinct in both
English and Dutch: while auxiliaries developed from full verbs and may develop further into
clitics and affixes, light verbs are synchronic variants of full verbs, in line with Butt & Lahiri
(to appear). The diachronic study involves close examination of variation in the valency
patterns of light verbs. It is shown that light verbs consistently contribute agentivity and that
the choice of light verb is influenced by the degree of involvement of the Agent.
References
Butt, Miriam & Aditi Lahiri. To appear. Diachronic Pertinacity of Light Verbs. Lingua.
Hopper, Paul J. & Elizabeth Closs Traugott. 1993/2003. Grammaticalization. First and second edition.
Cambridge: Cambridge University Press.
Mohanan, Tara. 2006. Grammatical Verbs (with Special Reference to Light Verbs). The Blackwell Companion
to Syntax ed. by Martin Everaert & Henk van Riemsdijk, vol. II, 459-492. Malden, Mass.: Blackwell
Publishing.

New Perspectives on Language Change: L2 Transmission and the Cognitive Basis for ContactInduced
Differentiation of Lexical Forms
T. Mark Ellison (Psychology, UWA) and Luisa Miceli (Linguistics, UWA)
In this paper we focus on a phenomenon that has received little attention in historical linguistics:
contactinduced change that leads to decreased similarity between languages. We argue that while
contact usually results in convergent structure, the presence of a sizable group of L2 speakers in a
linguistic community can, at the same time, give rise to a rapid differentiation of lexical forms and
create highly distinctive vocabularies. We propose that this contact-induced differentiation has a
cognitive basis and make our case in three steps.
First, we offer a cognitive model of bilingualism which accords with findings of cross-linguistic
priming and interference (e.g. Grainger & Beauvillain 1988). In this model, lexical entries are stored
as members of a single lexicon, tagged for language. Language choice is mediated by these tags:
lexical items flagged with any language other than the target language are inhibited. Words with
similar forms and meanings, which we collectively refer to as doppels (close cognates, loan words,
chance resemblances), constitute mutually coactivating lexical entries marked for multiple languages.
The model predicts that doppels will consequently suffer interference in bilinguals.
Second, we discuss the results of an experiment to verify this prediction. Dutch native speakers using
English on a daily basis were asked to complete a task that involved reading a context paragraph in
Dutch, followed by an English sentence where they were required to fill in a gap. The Dutch/English
bilinguals were found to use doppels significantly less frequently than our control group of
monolingual English native speakers. This is consistent with data presented in Arnal (2011) on the
impact of native Spanish speakers on Catalan in recent decades (one of the outcomes described being
the loss of a number of close cognates in Catalan), and discussions in Franois (2011) and Harvey
(2011).
Third, we present the results of a simulation that allows the impact of these results to be extrapolated
over populations with a significant number of L2 speakers. These assumed a starting rate of 50%
lexical overlap, expressed as the percentage of doppels in the lexicon of each language: these
languages are either closely related or have borrowed heavily from a common source. Each generation
in the simulation builds a distribution over possible lexical forms by sampling of the distribution used
by the previous generation. We modified this sample with the bias against doppels found in the
Dutch/English experiment, for simulated bilinguals. With 5% bias against doppels among bilinguals,
and those forming half the speaker community, the simulation shows a drop in the proportion of
doppels found between these (simulated) closely related languages from 50% to 5% within 100
generations. With greater bias (e.g. when social factors favour differentiation), or more bilinguals, the
reduction is faster.
Finally, we present our plans for future investigations on the role of bilinguals in language change,
including a collaboration with a third colleague which focuses on speakers of Papua New Guinea and
Solomon Islands languages in those areas where high degrees of lexical differentiation are suspected.

References:
Arnal, Antoni 2011.Linguistic changes in the Catalan spoken in Catalonia under new contact
onditions, Journal of Language Contact 4: 5-25.
Franois, Alexandre 2011. Social ecology and language history in the northern Vanuatu linkage: a
tale of divergence and convergence, Journal of Historical Linguistics 1(2):175-246.
Harvey, Mark 2011. Lexical change in pre-colonial Australia', Diachronica 28:345-381
Grainger, Jonathan & Ccile Beauvillain 1988. Associative priming in bilinguals: Some limits of
interlingual facilitation effects. Canadian Journal of Psychology/Revue canadienne de
psychologie 42 (3): 261-273.

Vocabular Clarity and Faroese conjugation

Hans-Olav Enger, Univ. of Oslo
h.o.enger@iln.uio.no
A possible restriction on inflection, more specifically on the behaviour of affixes, is the No
Blur Principle (NBP) aka Vocabular Clarity (VC) (Carstairs-McCarthy 1994, 2010 and
elsewhere). The principle is based on a parallel between morphology and lexical semantics.
Simplified, the idea is that there are restrictions on the behaviour of affixes (morphological
elements) similar to restrictions on the behaviour of words (lexical elements): There can be
no more than one class-default affix, i.e., only one affix that signals a particular
morphosyntactic value (e.g. past tense) without also signalling inflection class. If there were
more than one default, theyd be analogous to exact synonymy, which is surely rare and
possibly non-existent in the lexicon. This approach has been useful, e.g. in the study of
Romance (e.g. Maiden 2009, da Tos 2012) and Scandinavian (Enger 2007), but it is
controversial (see e.g. Mller 2007, Carstairs-McCarthy 2010, Baerman 2012).
Dammel (2011: 289) suggests that Faroese verb conjugation is promising ground for
discussing the VC. Compared to Old Norse, Faroese shows spreading of the suffix -i in the 1.
sg. prs of verbs. This suffix is not characteristic of the most type-frequent verb class (kasta) in
Old Norse, from which Faroese and Norwegian both descend, but of the less type-frequent
duga class. On the assumption that type frequency has some influence on productivity, the
spread is somewhat unexpected.
However, also in some Norwegian dialects, the suffix from the Old Norse duga class
has spread. Faroese is particularly close to Western Norwegian, but the duga affix has spread
in both some Western and some Eastern dialects. If the two Old Norse patterns instantiated
by duga and erfa are considered two separate classes, then the suffix -ir is the class default in
Old Norse (Enger 2007). It is then less surprising that this suffix might spread.
Grammars of Old Norse differ as to whether duga and erfa should be considered one
or two classes. Noreen (1923) takes them as two separate ones, at least in principle; Haugen
(1993) takes them as subgroups of one and the same class. The only affixal difference
between them in Old Norse has to do with the imperative 2nd sg. For the purposes of VC, this
is clearly sufficient to call the two different, and thus to follow Noreen rather than Haugen.
Only then does -ir emerge as class-default, and then, its productivity is less unexpected.
My main claims, then, are as follows:

Faroese supports the previous assumption that -ir is a class default

The diachronic evidence from Faroese supports one particular understanding of

inflection class (at least for present purposes)

While type-frequency may influence productivity, it is not the only determinant (e.g.
Maiden 1996); another determinant may be class-default status.

The expression of possession in medieval Spanish or how a parallel corpus can bring new
insights into the study of complex variable phenomena
Andrs Enrique-Arias (andres.enrique@uib.es)
University of the Balearic Islands
This paper uses the expression of possession in medieval Spanish as a case study to demonstrate
how a parallel corpus of medieval translations of the Bible can bring new insights into the study
of complex sets of variable phenomena.
The expression of possession in medieval Spanish constitutes an intricate cluster of variable
phenomena. First, there is a wide number of possessive structures: a) possessive adjective with
or without article ((la) su casa); b) genitive phrase with a personal pronoun (la casa de l) c)
duplicative structures exhibiting both possessive and genitive phrase ((la) su casa de l), as well
as other expressions such as dative pronouns, or even zero marking when the relation of
possession can be inferred from the context. Furthermore, the distribution for each one of these
variants correlates with a complex set of structural and external factors. For instance, the
frequency of article + possessive (as opposed to possessive alone) is conditioned by features of
the possessor (person, number), the possessed entity (animacity) and the syntactic function of
the NP that contains the possessive structure (Wanner 2005: 39). At the same time the use of
article + possessor is related to contextual factors, such as expressivity, solemnity, reverence
and poeticality (Lapesa 2000: 422).
This study uses a parallel corpus of Spanish medieval translations of the Bible
(www.bibliamedieval.es) in order to analyze in a more controlled manner the different factors
that condition variation in the expression of possession in Old Spanish. By locating the
possessive structures in the Hebrew or Latin original and then looking at their equivalents in a
number of Spanish translations we can observe the variation exhibited by possessive structures -including zero marking-- that can occur in the same linguistic environment. Secondly, as the
Bible includes a variety of registers (narrative, lyrical poetry, wisdom literature, prophesies, and
legal codes) the corpus allows for a more controlled analysis of stylistic variation.
In sum, this study demonstrates how a parallel corpus of medieval biblical texts may enrich our
theoretical understanding of morphosyntactic variation and change in Spanish.
REFERENCES
Lapesa, Rafael (1971/2000): Sobre el artculo ante posesivo en castellano antiguo. In: Cano,
Rafael/Enchenique, Mara Teresa (eds.): Estudios de morfosintaxis histrica del espaol.
Madrid: Gredos, 413-435.
Wanner, Dieter (2005): The corpus as a key to diachronic explanation. In: Kabatek,
Johannes/Pusch, Claus D./Raible, Wolfgang (eds.): Romance Corpus Linguistics II: Corpora
and Diachronic Linguistics. Tubingen: Gunther Narr, 31-44.

The rise and fall of functional coherence in autonomous morphology

Louise Esher
Research Centre for Romance Linguistics, University of Oxford
Recent work in the field of autonomous morphology (eg. Maiden 2011, Esher 2012, ONeill 2013) points to a
continuum between extramorphological and the morphomic (Aronoff 1994; paradigmatic phenomena which
resist explanation purely in terms of extramorphological factors such as semantics and phonology). Smith
(2013) proposes that paradigmatic distributions on this continuum may be situated along a cline of increasing
or decreasing functional coherence. In this paper, I discuss how different types of functional coherence may
emerge and evolve, demonstrating that functional coherence is a complex notion, and exploring the
relationship between its constituent components.
A paradigmatic distribution may be considered more coherent if it presents any of the following properties:
i)
it has a partial or unique semantic correlate
eg. the constituent cells of the morphome Fuc comprising the Romance synthetic future and
conditional share certain semantic values (Esher 2012)
ii)
it has a partial or unique phonological correlate
eg. the 1sg/3pl.prs and 1sg/3sg/3pl.sbjv cells (U-pattern) in Romanian present a consistent
phonological context (Maiden 2011)
iii)
it has a partial or unique morphosyntactic correlate
eg. Fuc groups together entire screeves rather than odd cells as the U-pattern does; Fuc can be
defined in terms of paradigm categories associated with the verb, whereas the definition of the Upattern crucially refers to the agreement categories of person and number (Smith 2013)
I propose that these three elements of functional coherence constitute independent continuous variables, and
that overall functional coherence must be evaluated with respect to all three. I present diachronic data from
Romance showing that morphomic distributions may at once increase their phonological coherence while
decreasing their morphosyntactic coherence, or decrease their semantic coherence while their morphosyntactic
coherence remains unchanged. I also note suggestive correlations in the data between the development of
semantic coherence and high morphosyntactic coherence, and between the development of phonological
coherence and low morphosyntactic coherence, and discuss the possible reasons for this. My study
contributes to developing an overall model for morphomic space, which will provide insight into how
morphomic distributions change and develop over time.

References
Aronoff, Mark. 1994. Morphology by Itself. Cambridge, Mass.: MIT Press.
Esher, Louise. 2012. Future, Conditional and Autonomous Morphology in Occitan. Unpublished D.Phil
thesis, University of Oxford.
Maiden, Martin. 2011. Allomorphy, autonomous morphology and phonological conditioning in the history of
the Daco-Romance present and subjunctive. Transactions of the Philological Society 109. 59-91.
ONeill, Paul. 2013. Morphomes and morphosyntactic features. In Cruschina et al.(eds.), The Boundaries of
Pure Morphology. Oxford: OUP.
Smith, John Charles. 2013. The morphome as a gradient phenomenon: Evidence from Romance. In The
Boundaries of Pure Morphology. Oxford: OUP.

The development of grammatical asymmetries: unmarked 3PL objects in Oceanic

Bethwyn Evans
The Australian National University
Asymmetries of grammatical coding, including the lack of overt coding altogether, are
often considered to reflect change towards patterns of linguistic marking that are iconic
and/or economical (e.g. Bybee 1985, Koch 1995, Haspelmath 2008). In this paper, I explore
such processes of change with respect to the unusual lack of coding of 3rd person plural
within the object marking paradigms of a number of different Oceanic (Austronesian)
languages of Papua New Guinea and the Solomon Islands.
In Oceanic languages the object argument is often indexed within the verb complex, as
illustrated by (1) from Marovo (Solomon Islands), where the verb omi to see occurs with
the 3SG object suffix -a, which agrees with the object noun phrase tape stingray.
(1)

om i-a
see-3sgO

ia

tape

eni

mae.

3SG

stingray PRT

come

he saw a stingray coming.

(own fieldnotes)

Cognate forms and similar patterns of object marking are found in many Oceanic languages,
and a partial paradigm of object markers that includes 1st, 2nd and 3rd person singular forms
and a 3rd person plural form, can be reconstructed for Proto Oceanic (Evans 1995). However,
in a number of Oceanic languages, including Marovo, a 3rd plural object argument can be
indicated by the lack of an overt marker on the verb (Evans 2008), as in (2).
(2)

La

om i

ria

ihana

pu

soku

via.

ASP

see

ART:PL

fish

REL

many

INTENS

(We) saw lots of fish.

(own fieldnotes)

By briefly illustrating this pattern of asymmetrical object marking in Marovo and two
languages of Papua New Guinea, Manam and Torau, I will show that synchronically zeromarked 3rd plural objects display a range of conditioning factors both within and across
languages. And while economy will be shown to be relevant in the development of
unmarked 3rd plural objects, I will also show that these apparently similar asymmetries of
morphological marking have arisen in different ways that need to be understood within the
context of the formal and functional organisation of the object-marking paradigms and
their use in discourse over time.
References
Bybee, Joan L. 1985. Morphology. A study of the relation between meaning and form. Amsterdam/
Philadelphia: John Benjamins.
Evans, Bethwyn. 1995. Reconstructing object markers in Oceanic. BA Honours thesis. Canberra: The
Australian National University.
Evans, Bethwyn. 2008. Third person plural as a morphological zero: object marking in Marovo. In
Claire Bowern, Bethwyn Evans and Luisa Miceli (eds) Morphology and language history. In honour of
Harold Koch. Amsterdam/Philadelphia: John Benjamins, 281-298.
Haspelamth, Martin. 2008. Creating economical morphosyntactic patterns in language change. In Jeff
Good (ed.) Language universals and language change. Oxford: OUP, 185-214.
Koch, Harold. 1995. The creation of morphological zeroes. In Geert Booij and Jan van Marle (eds)
Yearbook of morphology 1994. Dordrecht: Kluwer Academic publishers, 31-71.

Analogy, Plain and Simple: The Development of Exceptions to Sievers' Law in Gothic
and
Umlaut Alternations in the Old Norse Short-Stem Class-1 Weak Verbs
David Fertig
University at Buffalo (SUNY)
This talk looks at two well-known puzzles in the older Germanic languages and proposes
solutions that incorporate important insights from recent work into an essentially
"proportional" approach to analogical change.
The first involves the emergence of forms traditionally characterized as "exceptions"
to Sievers' Law in Gothic, in particular gen. sg. forms of long ja-stem neuter nouns that end in
jis rather than expected -eis (=/i/; e.g. reikjis 'kingdom'). Kiparsky (2000) points the way to
a satisfactory account by recognizing that: 1) the original sound change that precipitated the
analogical innovation was the lengthening of the suffix vowel in the nom. sg. of the long jastem masculines (*hardis > hardeis 'herdsman'); 2) this change gave rise to a reanalysis
whereby learners took the (unchanged) accusative sg. to be the anomalous form in the
masculine paradigm, and this reanalysis set the stage for the overt change in the neuters. Once
we recognize these crucial points, I argue that the change is amenable to a much more
traditional analogical account than that proposed by Kiparsky.
The second puzzle involves patterns of umlaut alternation and non-alternation in Old
Norse, especially in the class-1 weak verbs, where long-stem verbs have umlauted vowels
throughout the paradigm (1st sg. pres. indic. heyre pret. heyra 'hear'), while short-stem
items have an alternation in the indicative between umlauted vowels in the present (tel '(I) tell,
count') and no umlaut in the preterite (tala). This is reminiscent of medieval German, except
that the correlation between stem weight and vowel alternation is reversed. The German
pattern can be explained straightforwardly in terms of sound change: Syncope removed
umlaut triggers in the preterite of long-stem verbs, but not yet in the short stems, before
umlaut had occurred.
If we assume the same chronology of syncope for Old Norse, the pattern there looks at
first glance utterly perverse. Iverson and Salmons (2012) are entirely on the right track in
proposing that analogy played a key role in the emergence of this pattern and in recognizing
the crucial importance of the fact that syncope in the long stems disrupted the predictability
(in one direction) of the umlaut alternation while that predictability remained fully intact in
the short stems. Their account can be greatly simplified, however, if we assume that apparent
"syncope", i.e. the -ia > -a change, in the preterite short-stem forms was an entirely
analogical matter. To the extent that the short stems did not yet show any hint of either
phonetic syncope or of the "impending phonetic demise" of umlaut (117), the direct
replacement of *telia by tala would have been a straightforward analogical change, and
hypothetical surface forms such as [tela] (116) would never have existed.
I examine a number of aspects of Old Norse verbal morphology that would have
favored such a change and show how this analysis can be extended to account for the
presence and absence of umlauted vowels in other word classes, including i-stem nouns (e.g.
light stem star; heavy stem gestr).
Iverson, Gregory K. and Joseph C. Salmons. 2012. Paradigm Resolution in the Life Cycle of Norse
Umlaut. Journal of Germanic Linguistics 24.101-131.
Kiparsky, Paul. 2000. Analogy as optimization: 'exceptions' to Sievers' Law in Gothic. In Analogy,
Levelling, Markedness, ed. by Aditi Lahiri, 15-46. Berlin: Mouton de Gruyter.

Jrg Fleischer (Marburg)

The fate of lexical hybrids and their agreement forms in continental West Germanic
Lexical hybrids (Corbett 2006: 213), whose (grammatical) gender is in inherent conflict
with (natural) sex, are attested in all old West Germanic languages. Their agreement forms, in
principle, can display formal or semantic agreement. This is illustrated by the following Old
High German example, illustrating the hybrid noun wb woman; wife (printed in bold): the
agreeing definite article and the first anaphoric personal pronoun are neuter (underlined),
whereas the second and the third anaphoric pronoun are feminine (doubly underlined):
Thaz uuib thannez birit gitruobnessi habet, uuanta quam ira zt: thanne siu gibirit then kneht
A woman when she is in travail hath sorrow, because her hour is come: but as soon as she is
delivered of the child (John 16,21)
In the history of German, semantic agreement reaches a peak in Middle High German, but
becomes less widespread in New High German (Fleischer 2012). Interestingly, however,
other West Germanic languages display quite different patterns. In some varieties, the
anaphoric agreement forms of hybrid nouns are nearly exclusively feminine, not neuter, while
the attributive forms are neuter:
Dt is dt Wiu, ju ik jrsene al blouked hbe (East Frisian (Saterlandic))
This is the woman whom I have seen already yesterday
In other varieties, lexical hybrids have disappeared insofar as the attributive agreement
relations, including, crucially, the definite article, display feminine forms. Since phonology
provides no cue to the gender of a noun in most modern West Germanic varieties, this means
that the former lexical hybrids have arguably become de-hybridized:
un az er hot di vayb zayne tsum ershtn mol aroysgefirt [] hot er geredt tsu ir tsu hilkhik un
tsu gikh (Yiddish)
and when he took his wife out for the first time [] he talked to her too high and too fast
Thirdly, in some varieties even neuter anaphoric forms prevail:
I bi bein Wei gwn und han ams gsoat und is hat gsoat (Bavarian)
I visited the woman and told her it and she said
In the paper I would like to describe and analyze the fate of hybrid nouns and their agreement
forms in continental West Germanic. Especially interesting are comparisons of closely related
varieties: For instance, North Frisian displays quite different patterns compared to
Saterlandic, and the same holds for different varieties of Yiddish. As it turns out, factors other
than morphology also play a role, such as the value of neuter gender and the Agreement
Hierarchy (Corbett 2006: 207). Also, in the case of German, the spread of formal agreement
seems to be especially typical of the standard language and thus might be attributed to
codification.
References
Corbett, Greville G. (2006): Agreement. Cambridge: Cambridge University Press.
Fleischer, Jrg (2012): Grammatische und semantische Kongruenz in der Geschichte des Deutschen: eine
diachrone Studie zu den Kongruenz von ahd. wb, nhd. Weib. Beitrge zur Geschichte der deutschen
Sprache und Literatur 134: 163203.

Diachrony and procedural meaning; the case of the Rioplatense preterit

Guro N. Flgstad & Ingrid L. Falkum
University of Oslo
A crucial assumption in grammaticalization theory is that not only the change in the direction from
lexical item to grammatical item, but also the further semantic development of a grammaticalizing
construction may be predicted. For example, a possessive construction is first assumed to become a
perfect (have gone), and later expand into becoming a perfective (Bybee et al. 1994; Traugott &
Dasher 2002). Such a development is assumed to occur due to domain-general processes such as
categorization, chunking, memory storage and analogy, and its validity, which allegedly has substantial
empirical basis, in e.g. Romance languages, is assumed to form the basis of diachronic universals
(Bybee 2010). In the explanation of the expansion of a perfect, it is observed that specific meaning
components are lost. This - a generalization of meaning observed in diachrony - is seen as typical to
grammaticalization.
In the present paper, we show, through a corpus of spontaneous speech data, a quite different
development, that the perfect has not expanded, but disappeared in the Romance variety Rioplatense,
spoken in Buenos Aires, Argentina (cf. Schwenter & Cacoullos 2010 for other Latin American
varieties). In this variety, the preterit (went) was previously restricted to past contexts with specific
time frame, such as in (1):
(1) Estuve en Nueva York en el 2000 I was in New York in 2000
Today, however, the preterit s scope has expanded, and in addition to being used in contexts such as
(1), it may also refer to action with current relevance, previously expressed with perfect, such as in
(2):
(2) Si, estuve en Nueva York Yes I have been to New York (lit. yes I was in New York)
We propose an analysis of the Rioplatense preterit that accounts for its expansion in meaning,
which takes as its starting point the relevance-theoretic distinction between conceptual and
procedural meaning (Blakemore 1987; Wilson 2011). The observation is that there are words and
morphemes whose function is to guide the inferential comprehension process by imposing constraints
on it; in the case of tense markers, this would be a constraint on the determination of temporal
reference (Nicolle 1998). Thus, the Rioplatense preterit can be seen as having gone from encoding a
single constraint, that the temporal reference should be located to the past, to encoding an additional,
partly overlapping constraint, that the temporal reference should be located to the past with current
relevance. We suggest that this change was instantaneous (cf. Givn 1991; Nicolle 1998) in the sense
that it occurred for the first time in a context in which a speaker used it with the intention to express
past with current relevance, and that this meaning was also recovered by the addressee. The initial
uses of the preterit expressing current relevance must have been quite costly in terms of the
processing demands they placed on the hearers pragmatic capacity, but as this use caught on in the
language community it became progressively more routinised, thus less effort demanding, until a
'pragmatic routine' or inferential shortcut developed. This could explain the naturalness with which it
is used among the youngest informants in our corpus.
At one stage, then, the expansion of the procedual encoding of the Rioplatense preterit
contributed to an additional underspecification of its semantic contribution. However, from a
synchronic point of view, the situation is likely to be one of polysemy, with the tense marker having
developed two distinct procedual encodings which are conventionalized for individual speakers of the
language.
References
Blakemore, D. 1987. Semantic Constraints on Relevance. Blackwell. Bybee, J. 2010. Language, Usage and Cognition.
CUP, Givn, T. 1991. Serial verbs and the mental reality of event: grammatical vs. cognitive packaging. In
Traugott, E. & Heine, B. (Eds.): Approaches to Grammaticalization, Vol. 1. John Benjamins, 81-127. Nicolle, S.
1998. A relevance-theory perspective on grammaticalization. Cognitive Linguistics, 9(1): 1-35. Schwenter, S. A.
& R.Torres Cacoullos. 2010. Grammaticalization paths as variable contexts in weak complementarity in Spanish.
In James A. Walker (ed.): Aspect in Grammatical Variation. John Benjamins. Traugott, E. & R. B. Dasher. 2002.
Regularity in Semantic Change. CUP. Wilson, D. 2011. The conceptual-procedural distinction: Past, present and
future. In Escandell-Vidal, V., Leonetti, M., Ahern, A. (Eds.). Procedural Meaning: Problems and Perspectives
(Current Research in the Semantics/Pragmatics Interface, Volume 25), Emerald Group Publishing Limited.

Looking Back to Move Forward: Adjectival Passives, Verbal Participles or just Deverbal Adjectives?
Josep M. Fontana. Universitat Pompeu Fabra

The standard view among linguists from many different traditions and theoretical persuasions is that
verbal participles and adjectival passives (however many types are posited) are homophonous expressions
that belong to two different syntactic categories and have markedly different semantic interpretations. This
categorical distinction between verbal and adjectival participles is often justified by the contention that only
verbal participles have a clear eventive interpretation whereas adjectival participles would have a noneventive or stative interpretation. Whether implicitly or explicitly, proponents of this dual analysis seem to
assume that participles are primarily and essentially verbal expressions whereas their use as adjectives would
be marked and more restricted than their use as verbs. According to this view, adjectival participles are then
derivative, synchronically and/or diachronically, from the primary or basic uses and meanings of verbal
participles.
While they dont necessarily question the dual analysis of participles, a growing number of authors
are providing evidence showing that the semantic interpretation of adjectival participles cannot be so sharply
distinguished from that of verbal participles (e.g. Gehrke 2012, Maienborn 2009, Marin 2009). These
findings have important theoretical consequences because if it turns out that the type of denotation adjectival
passives contribute cannot be clearly distinguished from that of verbal participles, their analysis as two
differentiated but homophonous lexemes becomes unjustified.
This paper presents a case study on how diachronic linguistics can contribute to a better
understanding of synchronic analytical problems and provide empirical evidence that is in fact essential in
the determination of the relative value of different existing proposals to analyze synchronic phenomena. I
will discuss a wide range of data extracted from five different diachronic corpora that suggest that most
current standard assumptions about the distinctions between adjectival passives and verbal passives are
fundamentally flawed and are in fact based on a misunderstanding of the nature of participles and of the
semantic interpretations and syntactic functions they had in languages like Latin or Classical Greek. The data
to be discussed comes from corpora of Catalan, Ladino (Judeo-Spanish), Spanish, Classical Latin, and
Medieval Latin. Both diachronic and synchronic data suggest that all perfect participles can be posited to be
members of the same single syntactic category (arguably they are deverbal adjectives) and that verbal
participles, to the extent that this category can be maintained at all even now, are synchronically derived
from adjectival participles. Some of these data show that the diagnostics used in many contemporary
theoretical frameworks to differentiate adjectival participles from verbal participles are inadequate and do
not justify positing two distinct syntactic categories for participial expressions, at least for the Romance and
Germanic languages as well as Greek and Latin. Since these are precisely the languages that have served as
the basis for most linguistic theorizing in the Western tradition, it is highly likely that the classifications of
lexical and syntactic categories that emerged from the analysis of those languages have also affected our
views on the nature of (past) participles in other language families that have a similar type of expression.
I will argue that this misunderstanding is in turn the product of another misunderstanding.
Contemporary linguistic theories inherited the notion of voice from traditional grammars without realizing
that traditional grammarians had misinterpreted the meaning of the terms passive and active used by Greek
and Roman grammarians. Essentially, medieval grammarians failed to capture the correct generalizations
with respect to the voice paradigms in the classical languages because they interpreted the notions of
passivity and activity used by the classics in too narrow a sense, one that crucially involved the exclusive use
of the notions of agent and patient to characterize the distinction between the two voices. Despite the
efforts of Humanist grammarians such as Petrus Ramus or Sanctius Brocensis to redress the situation, this
misunderstanding managed to survive and has had a significant effect on the way contemporary linguists
analyze how different languages encode events and states.
Bibliography
Embick, D. (2004). On the structure of resultative participles in English. Linguistic Inquiry, 35(3), 355-392.
Demonte Barreto, V., & Bosque, I. (1999). Gramtica descriptiva de la lengua espaola. Espasa Calpe.
Gehrke, Berit (2012). Passive states. In Telicity, Change, and State: A Cross-Categorial View of Event Structure, ed. Violeta Demonte & Louise
McNally, 185-211.
Oxford: Oxford University Press.
Klaiman, M. H. (1991). Grammatical voice (Vol. 59). Cambridge University Press.
Kemmer, S. (1993). The middle voice (Vol. 23). John Benjamins Publishing Company.
Kratzer, A. (2012, June). Building statives. In Proceedings of the Annual Meeting of the Berkeley Linguistics Society (Vol. 26, No. 1).
Levin, B., & Rappaport, M. (1986). The formation of adjectival passives. Linguistic inquiry, 623-661.
Maienborn, C. (2009). Building event-based ad hoc properties: On the interpretation of adjectival passives. Universittsbibliothek der Universitt
Stuttgart.
Marn, R. (2009). Del participio al adjetivo. In Fronteras de un diccionario: las palabras en movimiento (pp. 327-348). Cilengua. Centro
Internacional de Investigacin
de la Lengua Espaola.
Quirk, R., Greenbaum, S., Leech, G., Svartvik, J., & Crystal, D. (1985). A comprehensive grammar of the English language (Vol. 397). London:
Longman.
Snchez-Marco, Cristina. (2012). Tracing the development of Spanish participial constructions: An empirical study of semantic change. PhD. Thesis.
Universitat

Pompeu Fabra.
Sleeman, P. (2007). Prenominal and postnominal reduced relative clauses: arguments against unitary analyses. Bucharest Working Papers in
Linguistics, (1), 5-17.
Wasow, T. (1977). Transformations and the Lexicon. Formal syntax, 327-360.

Subgrouping without trees

A glottometric study of the northern Vanuatu linkage
Alexandre FRANOIS (CNRS LACITO; Australian National University)
Siva KALYAN (Northumbria University; Australian National University)
Since it was first developed by Schleicher in 1853, the family tree has become the most widely accepted
model for representing historical relations between languages. And yet, it has also been the object of
criticism, including among followers of the Comparative method (e.g. Ross 1988, Garrett 2006, Bossong
2009, Heggarty et al. 2010), for the problematic assumptions that underlie it: (1) that the history of
languages can be traced back by looking exclusively at divergence, to the exclusion of convergence and
diffusion; (2) that each modern language belongs to a single subgroup, which is itself nested in another
discrete subgroup, and so on and so forth.
The tree model may be appropriate when a speaker population undergoes successive splits, with
subsequent loss of contact among subgroups. For all other scenarios, it fails to provide an accurate
representation of language history. In particular, it is unable to deal with dialect continua, or with the
language families that develop out of them for which Ross (1988) proposed the term linkage. In such
cases, the scopes of innovations (their isoglosses) are not nested but persistently intersect, in ways which
cannot be accurately represented by any tree structure. Though Ross's initial observations about linkages
concerned the languages of Western Melanesia, it is clear that linkages are found in many other areas as
well such as Fiji (Geraghty 1983), Northern India (Toulmin 2009), etc.
In this presentation, we focus on the 17 languages of the Torres and Banks islands in Vanuatu, which
form a linkage (Franois 2011), and attempt to develop adequate representations for it. Our data, based on
a strict application of the Comparative method, consists of 474 linguistic innovations reflected in the area
whether phonological, morphological, lexical or otherwise. Based on these 17474 data points, we propose
to illustrate Historical Glottometry, a quantitative approach to language subgrouping in situations of
linkage. One tenet of glottometry is that innovation-defined subgroups may intersect, and define patterns
that can be quantified and measured. We calculate the cohesiveness of each subgroup (proportion of time
it is confirmed by the data), and define the relative strengths of all subgroups by calculating their
subgroupiness (number of exclusively shared innovations weighted by subgroup cohesiveness). The result is
a glottometric map of northern Vanuatu, in which the relative strength of subgroups can be visually
represented in ways more faithful to historical reality than what the tree model can do.
Overall, we hope to show that historical glottometry enables a fine-grained, reliable and testable
representation of language history in genealogical linkages, that combines the scientific power of the
Comparative Method with a diffusionist, non-cladistic model of language diversification.
Bossong, Georg. 2009. Divergence, convergence, contact. Challenges for the genealogical classication of languages.
In Convergence and divergence in language contact situations, ed. by K. Braunmller & J. House. Hamburg Studies
on Multilingualism, 8. Amsterdam-Philadelphia: John Benjamins. Pp.13-40.
Franois, Alexandre. 2011. Social ecology and language history in the northern Vanuatu linkage: A tale of divergence
and convergence. Journal of Historical Linguistics 1 (2):175246.
Garrett, Andrew. 2006. Convergence in the formation of Indo-European subgroups: Phylogeny and chronology. In
Phylogenetic methods and the prehistory of languages, ed. by P. Forster & C. Renfrew. Cambridge: McDonald
Institute for Archaeological Research. Pp.139-151.
Geraghty, Paul A. 1983. The Languages of Fiji. Oceanic Linguistics Special Publications, 19. Honolulu: Univ. of
Hawai'i Press.
Heggarty, Paul, Warren Maguire, & April McMahon. 2010. Splits or waves? Trees or webs? How divergence measures
and network analysis can unravel language histories. Philosophical Transactions of the Royal Society B: Biological
Sciences. 365:38293843.
Ross, Malcolm. 1988. Proto-Oceanic and the Austronesian languages of Western Melanesia. Pacific Linguistics.
Canberra: Australian National University.
Toulmin, Matthew. 2009. From linguistic to sociolinguistic reconstruction: the Kamta historical subgroup of IndoAryan. Studies in Language Change, 604. Canberra: Pacific Linguistic

Differential object marking in Old Japanese:

A corpus based study
Bjarke FRELLESVIG (University of Oxford & University of Oslo)
Stephen Wright HORN (University of Oxford)
Yuko YANAGIDA (University of Tsukuba)
Modern Japanese (NJ) has an accusative case particle, o, which is used like accusatives in
many other languages to mark direct objects, some locational and temporal adjuncts, and causees
in causative constructions. The NJ accusative particle o is the direct descendant of the Old
Japanese (OJ; 8th century AD) accusative case particle wo and superficially OJ wo and NJ o
overlap in usage to a large extent, but there are also some apparent differences. Most
conspicuously, compared with NJ, overt accusative marking of direct objects was less frequent
in OJ and many direct objects were not case marked in OJ. See (1)-(2) for contrasting examples
from the 8th century poetry anthology Manysh.
(1)

Akami-yama
kusane
kari-soke
Akami-mountain
grass
cut-remove
At Mount Akami I cut and removed grasses (Manysh 14.3479)

(2)

kwomatu ga
sita no
kaya wo
kara-sane
small.pine GEN
under GEN grass ACC
cut-RESP.OPT
Please cut the grass under the small pine (Manysh 1.11)

Even within the past few years, several quite different proposals have been made about
marking of objects in OJ (including Kuroda 2008, Yanagida and Whitman 2009, Wrona and
Frellesvig 2010, Kinsui 2011, Miyagawa 2012), but there is still no consensus about the exact
circumstances determining when direct objects are bare or accusative case marked in OJ.
For this paper we use the material in the Oxford Corpus of Old Japanese (OCOJ). The
OCOJ is an annotated searchable corpus of OJ texts, which is marked up for morphology and for
syntactic constituency. It contains all poetic texts from the OJ period, approximately 90,000
words, and the main OJ prose text, Imperial Edicts from the 8th century, approximately 10,000
words.
Using the comprehensive material in the OCOJ we examine in detail the distribution of
bare and accusative case marked objects in the OJ texts and show that OJ had differential object
marking (DOM) (Dalrymple and Nikolaeva 2011 for a recent overview), in the case of OJ
associated with the specific/non-specific distinction as suggested by Yanagida and Whitman
(2009). We argue that the OJ DOM pattern is broadly comparable to that observed in Turkish, as
described by En (1991). Thus, in OJ, accusative marked objects are specific, but bare objects
inside VP are non-specific and those that occur outside VP (i.e., by movement) are specific.
References
Dalrymple, Mary and Irina Nikolaeva. 2011. Objects and information structure. CUP.
En, Mrvet. 1991. The semantics of specificity. Linguistic Inquiry 22, 1-25.
Kinsui, Satoshi. 2011. Tgron [Syntax]. Bunpshi [Historical grammar], ed. by Kinsui,
Takayama, Kinuhata, and Okazaki, 77-166. Iwanami.
Kuroda, S.-Y. 2008. On the syntax of Old Japanese. Current issues in the history and
structure of Japanese, ed. by Frellesvig, Shibatani, and Smith, 263-317. Kurosio.
Miyagawa, Shigeharu. 2012. Case, argument structure and word order. Routledge.
Yanagida, Yuko and John Whitman. 2009. Word order and alignment in Old Japanese.
Journal of East Asian Linguistics 18(2), 101-144.
Wrona, Janick and Bjarke Frellesvig. 2010. The Old Japanese case system: The function of
wo. Japanese/Korean Linguistics 17, 565-580. CSLI Publications.

Lat. ecce / It. ecco: the long road of a primary cognitive scaffolding
Livio Gaeta, University of Turin
(livio.gaeta@unito.it)
In spite of its uncertain categorial status, the Latin particle ecce displays a long history which
significantly continues and expands in Italian. The particle is generally associated with a thetic
value because it normally introduces new referents whose existence has not yet been given for
presupposed in the text. Moreover, it is generally taken to be not fully integrated into syntax,
because it could only govern nouns (in earlier texts marked with the accusative, later with the
nominative, see (1a-b)) in the first position of main clauses (only later also in front of any focused
constituent, cf. Cuzzolin 1998):
(1) a. Plaut. Stich. 577 eccum tibi lupum in sermone heres the wolf in the tale
b. Cic. Att. 2, 15, 3 ecce tibi Sebosus!
heres to you Sebosus!
Note that an inflected form also occurred resulting from the fusion of the personal pronoun:
ecc(e)+(e)um (1a). In contrast with this, its direct descendant ecco shows a spectacular syntactic
expansion in early Italian texts as a thetic copula, because it also governs verbal infinitives (2a-c)
and past participles (2d-e):
(2) a. Ecco la potenzia dellamistade generare spregio di morte
Heres the power of the friendship generating contempt of death
b. ecco sonar un corno - e i can baiare,
Theres a horn sounding, and the dogs barking
c. Ed ecco verso noi venir per nave / un vecchio, bianco per antico pelo
And theres an old, white-haired man coming towards us by boat
d. Ecco compiute le quattro ragioni
There are the four reasons come true
e. Ecco giunta colei che ne pareggia
Theres come the one who makes us all equal
Notice that the construction with the past participle, besides displaying special aspectual properties,
still functions as a test for unaccusativity, because unergative intransitives are not possible (*ecco i
cani abbaiati there are the dogs barked). The paper will show that ecco as a predicate is largely
integrated into the constructional syntax of old and contemporary Italian: in fact it is not only highly
selective as for the governed verbs, but it can also be embedded into several subordinate clauses,
and in particular temporal, causal, consecutive, relative clauses, etc. Embedding is only possible
when the thetic value is preserved, i.e. when the subordinate clauses are asserted and accordingly
follow the main clause. At any rate, its predicative status is non-prototypical because ecco does not
display typical morphosyntactic properties of verbs like inflection or negation.
Finally, ecce and ecco are interesting from a more general perspective in the light of (i) their thetic
value which makes them powerful means of expression for basic communicative situations, and (ii)
their very long history. This suggests that they might be treated on par with those relics of earlier
stages of the language capacity remain[ed] as pockets within modern language, which according to
Jackendoff (2009: 113) testify to the role of a protolinguistic substrate lurking below modern
language. This primitive scaffolding still irradiates its cognitive strength in contemporary Italian.
References
Cuzzolin, Pierluigi (1998), Quelques remarques syntaxiques propos de ecce. In B. Garca
Hernndez (ed.), Estudios de lingstica latina, Madrid, 261-271.
Jackendoff, Ray (2009), Compounding in the parallel architecture and conceptual semantics. In R.
Lieber & P. tekauer (eds.), The Oxford handbook of compounding. Oxford, 105-128.

MIDIA: a balanced diachronic corpus of Italian

Livio Gaeta, University of Turin (livio.gaeta@unito.it)
Claudio Iacobini, University of Salerno (ciacobin@unisa.it)
Davide Ricca, University of Turin (davide.ricca@unito.it)
Marco Angster, University of Turin (marco.angster@unipv.it)
Aurelio De Rosa, University of Salerno (aurelioderosa@gmail.com)
Giovanna Schirato, University of Salerno (agschirato@libero.it)
Within a large research project funded by the Italian government and aimed at investigating Italian
word formation, a new research tool is being developed, namely MIDIA (Morfologia Italiana in
DIAcronia). This diachronic corpus of Italian fills an important lacuna: although a considerable
amount of texts are available electronically with special regard to the earlier stages of Italian, a
balanced diachronic corpus was until now a desideratum. MIDIA has been explicitly designed to
cover all periods of the linguistic history of Italian, divided in five homogeneous timespans and
collecting texts for seven different text types, going from prose and poems to chronicles, scientific
essays, etc. For each timespan and text genre 50 text chunks of about 8000 tokens have been taken
and acquired into a database. This amounts to a final corpus of about 14 million tokens, which is
hopefully large enough to perform a number of investigations on productivity of single word
formation patterns (cf. Gaeta & Ricca 2006, in press) as well as more in general on recurrent
constructional patterns including valence structure, idioms, etc. (Bardal 2008, Zeldes 2012).
Clearly, the database is independently searchable for any of the parameters indicated, namely
timespan and text genre.
The software utilized for building the database comes from the TreeTagger family as already
employed for the Repubblica corpus (cf. Baroni et al. 2004), enriched by more than 170,000
wordforms collecting all sort of allomorphs attested during the long history of Italian (about 26,000
nominal, 11,500 adjectival, and 98,000 verbal forms plus about 15,600 forms hosting clitics) and
including most grammatical words (about 1,800) and proper names (about 21,000). All of this
considerably enlarges the basic lexicon already present in the TreeTagger version of the Repubblica
corpus, and substantially improves its efficiency as a syntactic parser. Moreover, the users friendly
searching mask allows any possible research on single as well as on sequences of tokens or POStagged lemmas (or parts of them), employing all common logical symbols and extracting the data
into any sort of tables and graphs.
Besides presenting the new tool, the paper intends to offer a concrete exemplification of its
capacity, trying to empirically answer an interesting question: Given that the TreeTagger software
developed for the Repubblica corpus was originally designed to work on contemporary Italian, will
this modified version perform efficiently for old Italian as well? The answer is not trivial in the light
of the significant syntactic differences distinguishing old and contemporary Italian especially with
regard to word order (cf. Beninc & Poletto 2010 for an overview) which is a crucial cue for a
statistically driven parser like TreeTagger to work. The paper shows that in order to get a positive
answer the software needs to be trained specifically with text chunks taken from old Italian, which
is only possible if a balanced corpus like MIDIA is available. At any rate, enriching massively the
basic lexicon of TreeTagger significantly increases its efficiency, because especially proper names,
verbal forms hosting clitics and grammatical words behave as anchoring blocks facilitating the
recognition of syntactic structure.
Besides the obvious fact that adding wordforms to the basic lexicon improves the efficiency of the
software because more words are recognized, verbal forms hosting clitics were particularly difficult
to deal with by TreeTagger because the latter is unable to separate the clitic. Thus, from a verbal
form hosting a clitic like abbandonatala abandoned:her TreeTagger reconstructed a verb
*abbandonatalare. Besides the obvious lexical mistake, this also had immediate consequences for
the syntactic parser, because it took a past participle for an indicative, accordingly providing a

completely different syntactic environment. On the same track, proper names were difficult to deal
with because they normally occurred in a very different syntactic environment than common nouns
due to the lack of specifiers and modifiers typical of NPs, which rendered impossible their
recognition on the basis of the normal syntactic patterns statistically associated with a noun. Finally,
giving grammatical words as an input of the basic lexicon enormously facilitated the parsing
because of the unpredictability of their position and of the large graphic variation characterizing old
texts. The latter was a general problem particularly affecting the lemmatization of verbs in virtue of
their rich inventory of wordforms. This often created problems not so much for the parsing but for
the lemmatization, because for instance a form like abisognano need:3pl:prs:ind was correctly
identified as a verb but erroneously attributed to a new lemma abisognare and not to the correct
abbisognare to need, thus duplicating the entries. For this reason, we decided to massively add to
the basic lexicon a large number of verbal forms as they are recorded in philologically reliable
sources.

References
Bardal, Jhanna (2008), Productivity: Evidence from Case and Argument Structure in Icelandic.
Amsterdam/Philadelphia: Benjamins.
Baroni, Marco et al. (2004), Introducing the la Repubblica corpus: A large, annotated, TEI(XML)compliant corpus of newspaper Italian. Proceedings of LREC 2004, Lisbon: ELDA. 17711774.
Beninc Paola & Cecilia Poletto (2010), Lordine delle parole e la struttura della frase. In
Giampaolo Salvi & Lorenzo Renzi (eds.), Grammatica dellitaliano antico, Bologna: il
Mulino, vol. 1, 27-75.
Gaeta, Livio & Davide Ricca (2006), Productivity in Italian word-formation: A variable-corpus
approach. Linguistics 44.1: 57-89.
Gaeta, Livio & Davide Ricca (in press), Productivity. In Peter O. Mller, Ingeborg Ohnheiser,
Susan Olsen, Franz Rainer (eds.), Handbook of Word-Formation. An International Handbook
of the Languages of Europe. Berlin/New York: Mouton de Gruyter.
Zeldes, Amir (2012), Productivity in Argument Selection. From Morphology to Syntax. Berlin/New
York: Mouton de Gruyter.

Behind the scenes of inflectiona 2000-year timeline of evolution

Francesco Gardani
(Universitt Graz)
Due to its central role for morphological theory, productivity has been the object of a wealth of
discussions and analyses conducted from the most different theoretical and empirical angles (for a
survey, see Bauer 2005). Still, a fundamental dimension of productivity has not been adequately
enlightened in morphological research, viz. its role in the diachronic evolution of inflection. Based
on the investigation of the evolution of the noun inflectional classes of Latin and Old Italian
covering a time of approximately 2,000 years (Gardani 2013), this paper offers an analysis of longterm morphological change in terms of the role that productivity plays in the dynamics of
emergence, growth, decline, and loss of inflectional classes.
On a wider perspective, the paper follows the line of research on the evolution of morphology
recently propelled forwards by Carstairs-McCarthy (2010). More specifically, the investigation is
couched in the functionalist framework of Natural Morphology (Dressler 2003), from which also the
methodological approach is derived. Class productivity is measured, on historical synchronic cuts,
on the basis of the investigation of loanword integration, conversions, and class shift, with the data
on the integration of loanwords being drawn from the contact languages Ancient Greek, Germanic,
Arabic, Byzantine Greek, and Old French. The elaboration of the diachronic outline is then
encompassed by connecting the single synchronic cuts.
The paper shows that, in parallel with van Marles (1988: 148) results on derivation, also in
inflection there exists a link between changes in productivity and the naturalness parameters of
biuniqueness and morphosemantic transparency, relating, in particular, to the morphological
realization of the features of gender and number: for example, a higher level of biuniqueness in the
morphological realization of gender and number values determines an increase in productivity of
extant classes and also leads to morphogenesis.
References:
Bauer, Laurie. 2005. Productivity: theories. In Pavol tekauer & Rochelle Lieber (eds), Handbook of
Word-Formation, 315334. Dordrecht: Springer.
Carstairs-McCarthy, Andrew. 2010. The evolution of morphology. Oxford: Oxford University Press.
Dressler, Wolfgang U. 2003. Degrees of grammatical productivity in inflectional morphology. Italian
Journal of Linguistics 15(1). 3162.
Gardani, Francesco. 2013. Dynamics of morphological productivity. The evolution of noun classes from Latin
to Italian. Leiden/Boston: Brill.
Marle, Jaap van. 1988. On the role of semantics in productivity change. In Geert Booij & Jaap van
Marle (eds), Yearbook of Morphology 1988, 139154. Dordrecht: Foris.

Georgakopoulos Thanasis
Humboldt Universitt zu Berlin
The SOURCE-GOAL asymmetry in Ancient Greek motion events

This talk aims at examining the hypothesis of the prevalence of information relating to the
GOAL of motion over information relating to the SOURCE of motion in motion events of
Ancient Greek (henceforth AG). Although the phenomenon has been extensively
investigated, it is unclear whether SOURCE and GOAL expressions behave asymmetrically
(see, e.g., the articles in Kopecka & Narasimhan 2012). To clarify this issue, the following
two questions are addressed: 1) What are the factors that determine this asymmetrical
representation; and 2) What are the lexico-grammatical means by which AG encodes the
GOAL and the SOURCE of motion. These questions are assessed by the means of a tailor-made
special-purpose historical corpus constructed by myself. The corpus covers two different
stages of Greek, namely Homeric and Classical Greek, and comprises 32 works by 5
authors containing approximately 550,000 words in total. The results suggest that the
manifestation of the asymmetry is dependent on various intervening factors (in both
diachronic stages), e.g., the type of figures motion (self-motion vs. caused motion), the
lexical semantics of the motion verb (manner verbs [e.g., ban to walk, pl to navigate])
vs. verbs of inherent directionality [e.g. hk arrive, pheg to flee, take a flight, escape]),
etc. It is also shown that GOAL expressions exceed in number SOURCE expressions (e.g., 6
prepositions are SOURCE-denoting and 9 GOAL-denoting; if we take into account prepositioncase combinations, the inventory for the denotation of the GOAL becomes much richer). An
additional analysis is conducted, in which instances where a GOAL marker stands in the
place of a SOURCE marker are detected (cf. Ikegami 1987). The reasoning behind this
investigation is that in a language that GOAL and SOURCE are asymmetrically expressed, the
GOAL expressions will collocate with a greater number of verbs than the SOURCE
expressions. This is illustrated in the abstract caused-motion event below, whereby the
dative case -i.e., a marker that can be found in GOAL relations- is used in expressions of
receiving, although the verbs belonging in this semantic field require a complement with
an ablative value.
psou
prma
soi
t
khoirdia
Of.what.quantity:GEN.SG buy:1SG.SUBJ.M/P 2SG.DAT ART.ACC.PL
little.pigs:ACC.PL.N
For what sum will I buy from you the little pigs? (Aristophanes, Acharnians 812)

Based on these results, an account for the SOURCE-GOAL asymmetry will be proposed, and
implications of the role of other factors will be discussed.
References
Ikegami, Yoshihiko (1987). Source vs. goal: A case of linguistic dissymmetry. In: Dirven,
R. and G. Radden (eds.), Concepts of Case. Tbingen: Narr, 122-146.
Kopecka, Anetta & Bhuvana Narasimhan (eds.) (2012). Events of Putting and Taking. A
Crosslinguistic Perspective. Amsterdam/ Philadelphia: John Benjamins Publishing
Company.

The origins of clause-medial wh-relatives in Middle English.

Nikolas Gisborne, University of Edinburgh
Robert Truswell, University of Ottawa
This paper examines the diachrony of English relative clauses, focusing on the relationships
between lexical change and directional pathways of syntactic change. Headed relatives with
relative pronouns (as opposed to indeclinable relative particles like that) are common in IndoEuropean (IE), but almost nonexistent elsewhere: they are attested in 27 of 40 IE languages in
de Vries (2002), but only 7 of 132 other languages. However, such relatives were not plausibly a
feature of Proto-IE, as the oldest attested IE languages all use correlative relative clauses
(Haudry 1973, Kiparsky 1995, Bianchi 2000). A correlative is a biclausal paratactic structure
like (1), from Hindi. The second clause contains a demonstrative NP (vo in (1)), anaphorically
related to a relative element in the first clause (jo laRkii in (1)).
1) [jo
laRkii khaRii
hai] [vo lambii hai]]
Which girl
standing is
that tall
is
The girl who is standing is tall (Srivastav 1991: 639-40)
This entails that IE languages did not inherit headed relative pronouns from PIE. Rather, they
evolved from PIE correlatives independently several times, an example of a parataxis >
hypotaxis pathway.
The history of English shows this in microcosm. OE could form relative clauses with e (an
ancestor of that), with inflected demonstrative pronouns, as in (2), or with both.
2) Her Feng
to Dearne rice
Osric one
Paulinus r
gefullode
Here Succeeded to Deira
kingdom Osric that.ACC Paulinus earlier baptized
In this year Osric, whom Paulinus had earlier baptized, succeeded to the kingdom of Deira
(Peterborough Chronicle, 12th century, Allen 1977: 83)
In Middle English, inflected demonstratives disappeared, and a new series of headed wh-relative
pronouns gradually emerged in the 12th-16th centuries, (the only OE wh-relatives were
generalizing correlatives). Therefore, English has developed two independent sets of relative
pronouns, despite their typological rarity.
We show that this is due to lexical reanalysis, rather than constructional properties of
English relatives. First, the development of headed wh-relatives was not a case of lexical
replacement, because inflected demonstratives disappeared as the OE case system collapsed in
the 12th century, before headed wh-relatives emerged (Allen 1977). And secondly, headed whrelatives in MidE had a different distribution from that-relatives or demonstrative relatives:
they were initially restricted to sentence-final position, suggesting that they were paratactic
structures. Clause-final headed wh-relatives like (3a) emerge at a time when clause-medial nonwh relatives like (3b) were common, but it was fully 200 years before the emergence of
unambiguously hypotactic clause-medial wh-relatives like (3c).
3a) meister we wolden sen sum fortocne of e. Warbi we mihten cnowen gif it
master we would see some sign
of you whereby we might know if it
so were at u seist
true were that you say
Master, we would like to see some sign from you, whereby we might know if what you say
is true (Trinity homilies, c.1200)
3b) as & feola ore a
wron r kyninges eonestmen hit geotton ealle
Theseand many other dem.pl were there kings
attendants it confirmed all
These, and many others who were there of the kings attendants, all confirmed it.
(Peterborough Chronicle)
3c) And the Lord God bildide [the rib [which he hadde take __ fro Adam]] in to a woman
And the Lord God formed the rib which he had taken from Adam into a woman (Wycliffe,
Old Testament, late 14th century)
This demonstrates that wh-relatives in MidE did not assume the properties of other types of
headed relative: there is no constructional unity across types of MidE headed relative. Rather,
the emergence of headed wh-relatives was a product of gradual changes in the syntax of whpronouns (generalizing > definite, independent > subordinated).

Mnica Gonzlez Manzano

Universitat de Barcelona

Pragmaticalization and syntactic distribution. About the historical evolution of the Spanish
epistemic adverb de verdad

De verdad is an adverbial phrase which, apparently, shows an epistemic meaning very similar to other
adverbial elements in Spanish, as for example verdaderamente or en verdad. However, our goal in this talk is
to show that real synonymy does not exist in the language, and therefore explain the differences
concerning the syntactic distribution between de verdad and other similar epistemic adverbs. In order to
do so, we describe the historical evolution of de verdad, and we analyze the grammaticalization process that
this item experiences and which can be described as following this chain of grammaticalization:
prepositional phrase > verbal adverb > sentential adverb > discourse marker
Thus, firstly, on the one hand, de verdad can appear introducing different verbal complements, and on the
other hand, it can modify a noun or an adjective, in contexts like bueno de verdad. In these cases, an
inferential meaning arises, because de verdad, sometimes explicitly and sometimes implicitly, opposes a true,
reliable element, to another false, unreal one. This is a very relevant context, because data extracted from the
Corpus diacrnico del espaol (CORDE) shows that this is the main function of this phrase through history.
Secondly, de verdad becomes established as an adverbial phrase, and begins to function as an adverb (as a
verbal adverb, modifying the VP, as well as a sentential adverb, expressing epistemic modality). When de
verdad takes scope over the whole sentence, it also experiences a semantic bleaching1, that is to say, it
partially loses its modal meaning, equivalent to 'in a truly, authentic manner', and it develops a more
reinforcing semantic value.
Thirdly, de verdad grammaticalizes as a discourse marker with an emphatic, reinforcing meaning. It also
gains a more pragmatic value, as the discourse marker de verdad introduces the final, reliable argument, in
order to refute other possible false ones.
This final step was facilitated by the coappearance of de verdad with epistemic adverbs, which constitutes
the bridge context2 to develop this new discursive function. Furthermore, other features were involved in this
process, especially the semantic ambiguity that arises in some contexts where the reader cannot distinguish
whether de verdad takes scope over the sentence but is still in a propositional level, or it functions in an
extrapropositional level.
Finally, the data analyzed shows that de verdad, although developing a discursive meaning in the XVth
century, never consolidates as a discourse marker. On the contrary, de verdad functions in an intrasentential
level even during the XX-XXIth centuries, as the data extracted from the Corpus del espaol actual (CREA)
shows. This is why, although the adverbial form verdaderamente and de verdad exhibit a very similar
reinforcing meaning when functioning as discourse markers, we cannot consider them as synonyms, as de
verdad shows a very different syntactic distribution.
Corpus
REAL ACADEMIA ESPAOLA: Corpus diacrnico del espaol (CORDE), online version. <www.rae.es>
REAL ACADEMIA ESPAOLA: Corpus del espaol actual (CREA), online version. <www.rae.es>
References
SWEETSER, Eve (1988): Grammaticalization and Semantic Bleaching. Proceedings of the fourteenth annual meeting
of the Berkeley Linguistics Society, 389-405.
HEINE, Bernd (2002): On the role of context in grammaticalization. Ilse Wischer & Gabriele Diewald (eds.), New
Reflections on Grammaticalization. Amsterdam: John Benjamins, 83-97.

1
2

See Sweetser (1988) for a definition of semantic bleaching.

We have applied the term bridge context as Heine (2002) defines it.

The evolution of the pronoun le and the origins of lesmo in Spanish:

A diachronic study
Alejandra Gonzlez-Prez
University of Houston

In Spanish the use of the indirect personal pronoun le instead of the pronouns lo and la as
the direct verbal argument is known as lesmo. The present work addresses this
phenomenon of lesmo from the traditional perspective represented originally by Rufino Jos
Cuervo (1886), followed later by Salvador Fernndez Ramrez (1964), and most recently
defended by Rafael Lapesa (1986, 1993, 2000). For these authors, the loss of the Case
distinctions during the transition from Classical Latin to Vulgar Latin was due to the necessity
of having a pronominal reference that could distinguish both the gender and the [+/ animate]
features of the entity to which the pronoun was referring. This study analyses the parallel
developments of the personal pronoun le and the so-called personal a, and the influence that
this preposition could possibly have had in the change of the verbal argument requirements
from which lesmo was originated.
The main objective of this study is to explain how in Latin the use of the personal
preposition a with different types of verbs gave rise to both etymological and antietymological lesmo (Marcos Marn 1978). It also examines the influence that Latin verbs,
which required dative etymologically, had on those other verbs that required accusative case.
This phenomenon was a natural extension of the dative case given its higher frequency of
occurrence in the historical evolution of the Spanish language.
REFERENCES
Company Company, Concepcin. 1997. Cambios diacrnicos en el espaol. Mxico: Universidad
Nacional Autnoma de Mxico.
Davies,
Mark.
2001/2002.
Corpus
del
Espaol.
Illinois
State
University.
http://www.corpusdelespanol.org
Eberenz, Rolf. 2000. El espaol en el otoo de la Edad Media. Sobre el artculo y los pronombres.
Madrid: Gredos.
Fernndez Ramrez, Salvador. 1964. Un proceso lingstico en marcha. Presente y futuro de la lengua
espaola. Actas de la Asamblea de Filologa del I Congreso de Instituciones Hispnicas.
Madrid: Ediciones Cultura Hispnica, vol. II. 277-285.
Fernndez Soriano, Olga. 1993. Los pronombres tonos. Madrid: Taurus Universitaria.
Garca, Erica. 1995. La mejor palabra es la que no se habla. In Carmen Pensado, El complemento
preposicional. Madrid: Visor Libros.
Garca, Erica. 1995. Relevancia expresiva vs. Desambiguacin: el a personal. In Carmen Pensado, El
complemento preposicional. Madrid: Visor Libros.
Lapesa, Rafael. 2000. Estudios de morfosintaxis histrica del espaol. In Rafael Cano Aguilar &
Mara
Teresa Echenique Elizondo (eds). Madrid: Gredos.
Lapesa, Rafael. 1993. Sobre los orgenes y evolucin del lesmo, lasmo y losmo. In Olga
Fernndez Soriano, Los pronombres tonos. Madrid: Taurus Universitaria.
Lapesa, Rafael. 1986. Historia de la lengua espaola. Madrid: Gredos.
Lpez Bobo, Mara Jess. 1990. Sobre el lesmo en el Libro del Buen Amor. Verba 17. 343-361.
Marcos Marn, Francisco. 1978. Estudios sobre el pronombre. Madrid: Gredos.
Penny, Ralf. 1991. A History of the Spanish Language. Cambridge: University Press.
Pensado, Carmen. 1995. El complemento preposicional. Madrid: Visor Libros.

Pickbourn (1789) and the Aspectual Restriction on the Progressive

Mariko Higuchi Goto
Kyushu Institute of Technology, Iizuka

While the progressive construction is aspectually restricted as in *He is knowing, the reason
has been left unexplained. This paper proposes that it may be found in grammar writing from late 18th
century through the 19th century intertwined with the development of Standard English.
It seems to be Pickborn (1789) that touched upon the restriction for the first time. The
author states that as in (1).
(1) We do not say, I am loving, I am fearing, I am hating, I am approving, I am knowing; but we say, I
love, I fear, I hate, I approve, I know &c. (Pickbourn: 1789: 81-82).
Such an explicit prohibition does not seem to exist in earlier prominent grammar books. In fact, I am
loving is given as an example of the present imperfect in Lowth (1762) which is viewed as one of the
most authoritative grammar books in the 18th century. The prescriptive rule is presented as part of
Pickbourns description about distinctive usage of the simple present and the progressive as his own
observation. Considering that the two forms seem to be regarded as synonymous or free variants by
many grammarians in his time and even after the turn of the 19th century as Cobbett (1818) notes that
He works and He is working mean the same, the regulation on the progressive that we are familiar with
must have been still in an unbeaten track when Pickbourn started writing his treatise.
Later grammars, such as Brown (1823) and Bain (1863), seems to manifest Pickbourns
influence on them. Brown mentions that we say, I respect him; but not I am respecting him.
Similarly, Bain notes that an expression he is not intending violates the best English usage.
According to Grlach (1998), revised editions of Brown (1823) were published more than 20 times and
new editions of Bain (1863) 10 times, which may have lead to further proliferation of the rule. It might
also be worth paying attention to the word we employed in (1) as well as in Brown since the pronoun
was symbolic of users of correct English in the 18th and 19th centuries. The restriction in question
may gradually have infiltrated the grammar of the English language during the 19th century. The fact
that He is knowing is not acceptable in the present-day standard English may also indicate how
influential Pickbourns restriction has been.
Reference
Bain, Alexander (1863) An English Grammar, London: Longmans Green & Co.
Brown, Goold (1823) The Institutes of English Grammar, New York: the author; rpt. Delmar: Scholars
Facsimiles and Reprints (1982).
Cobbet, William (1818) A Grammar of English Language, New York: the author; rpt. Amsterdam: Rodopi
(1983).
Grlach, Manfred (1998) An Annotated Bibliography of Nineteenth-Century Grammars of English, Amsterdam/
Philadelphia: John Benjamins.
Lowth, Robert (1762) A Short Introduction to English Grammar, London: J. Hughs; rpt. Menston: Scolar Press
(1967).
Pickbourn, James (1789) A Dissertation on the English Verb, London: J. Davis; rpt. Menston: Scolar Press
(1968).

Null subjects and agreement in the history of Russian

Pavel Grashchenkov
Institue of Oriental Studies (Moscow) & Moscow State University
pavel.gra@gmail.com
There is a long-term story about relations between the verbal paradigm complexity and the
availability of null pronouns in a language. Languages that can not easily fit this hypothesis include
but are not limited to Chinese (Huang 1989), German (Mller 2005), French (Roberts 2007). Another
example of a problematic language, Russian, is being discussed in this paper.
Old Russian had null thematic referential subject pronouns that have been lost approximately in 16-17th cc., see (Borkovskiy 1949, Sukhoon 2003, Meyer 2011). The loss of null subjects was
preceded by the decline of the past tense system, see (Franks 1995, Mller 2006). The Old Russian
past tenses included aoirist, imperfect, perfect and pluperfect that had elaborated per-son-number
paradigm. Modern Russian lacks personal agreement for the past tense, but pre-served it for the
present. This calls in question Mllers (2006) account of a pro-drop loss as a result of
impoverishment and system-defining syncretism.
In spite of the loss of referential thematic null subjects, two prominent semantic types of
thematic pro are attested in Modern Russian, see (Franks 1995, Zimmerling 2009). These are a 3pl
non-referential subject, (1) and a 2sg generic subject, (2). It could be argued that the se-mantic
properties of these constructions are due to the interpretation of verbs only, but the standard tests on
reflexive binding, (3) and adjunct control, (4) show presence of null subjects.
The so-called semi-null subject phenomenon displayed by Modern Russian with 3pl and 2sg
null pronouns was reported in (Rizzi 1986), see (Biberauer 2008) for discussion. Semi-null subject
contexts impose semantic constraints on the distribution of pro elements. One of the most typical
cases of such constraint is non-referentiality.
The claim of the paper is that pro elements have not disappeared but have been reanalyzed on
the way from Old to Modern Russian. I will argue that null pronouns have undergone se-mantic
transformation after the past tenses decay.
The reanalysis proceeded in the following steps. At the first stage Old Russian had referential pro and personal agreement in the past (and other tenses):
(i) [pro{D} T{D} ]
At that stage null subjects bore D features licensed by T heads and thus introduced referential antecedents (5). As reported in (Borkovskij 1949), non-referential 3pl pro was already pre-sent in
Old Russian, see (6). At the same time, generic constructions with verbs in 2sg were not attested.
These facts can be reduced to the referential properties of pronouns: ability to denote non-referential
participants is a natural pragmatic extension of the 3rd person [-1,-2] pronoun, but not of the 2nd
person pronoun. After that, T lost D features in past tenses:
(ii) [pro{D} T{} ]
At this period (16th c.) Russian still had null subjects but in the past tense contexts they were
not properly licensed and interpreted, that resulted in D features degeneration from pro:
(iii) [pro{} T{} ]
This move led to another conflict state: the present tense could not discharge its D features:
(iv) [ pro{} T{D} ]
But the process of acquiring non-referentiality (loss of D features on T and pro) was supported in the present tense clauses by the fact that they regularly had habitual interpretation, (7).
Examples like (7) report facts that are actual for any participant from some class at any moment of
their existence. So, in some contexts verbs did no referential job (had no D features) even in Old
Russian, what simplified the transition from T{D} to T{}.
Thus we argued that development of non-referential 3pl and generic 2sg pro historically results from the two facts: a) the presence of referential null subjects in Old Russian and b) the loss of
person agreement in past tenses from Old to Modern Russian.

(1)

(2)

(3.a)

(3.b)

(4.a)

(4.b)

(5)

(6)

(7)

3Pl (pronon-ref)
Valenki nos-at zimoj.
felt.boots wear-3Pl in.winter
Everybody wears felt boots in winter.
2Sg (progen)
Posmotr-i na nego i grustno stanovitsa.
look-2Sg at him and sad become
Everybody, who looks at him, got sad.
3Pl (pronon-ref) Reflexive binding
Valenki nadevaj-ut dla samix seba a ne dla roditelej.
felt.boots put.on-3Pl for themselves but not for parents
People put on felt boots for themselves, not for their parents.
2Sg (progen) Reflexive binding
Poslua-e samogo seba i grustno stanovitsa.
Listen-2Sg himself and sad become
Somebody listens to him/herself and it is getting sad.
3Pl (pronon-ref) Adjunct control
Valenki zimoj nos-at toby ne zamerznut.
felt.boots in.winter wear-3Pl for not freeze
People wear felt boots in winter for not freezing.
2Sg (progen) Adjunct control
prid-e porane (toby) prigotovit uin, a sil ue net
come-2Sg earlier (for) cook dinner but forces already no
You come earlier to cook dinner, but you have no forces already.
Novgorod birch bark letters, 12th c., (Letopisi 1950)
A mn ne vda-st nito e.
And me not give-3Sg nothing EMPH
And he wont give me anything.
3Pl (pronon-ref) in Old Russian, 13th c., (Borkovskij 1949:106)
Kak to mesto zov-ut, gde stoim?
how this place call-3Pl where stay.1Pl
What is the name of the place where we are staying?
13th c., (Gorshkova & Haburgaev 1981:286)
Polane bo svoix otc obyai im-out krotok i tix.
Polans EMPH their fathers customs have-3Pl gentle and quiet
As though Polans have gentle and quiet customs of their ancestors.

References: Biberauer, T. 2008. Semi null-subject languages, expletives and expletive pro
reconsidered. || Borkovskij, V. I. 1949. Sintaksis drevnerusskih gramot. Prostoje predlozenie. ||
Franks, S. 1995. Parameters of Slavic Morphosyntax. || Gorshkova & Haburgaev 1981. Istoricheskaja grammatika russkogo jazyka. || Mller, G. 2006. Pro-Drop and Impoverishment. || Rizzi,
L. 1986. Null Objects in Italian and the theory of pro. || Roberts, I. 2007. Taraldsens Generalisation
and Language Change: Two Ways to Lose Null Subjects. || Zimmerling, A. 2009. Aggressive prodrop and the specificity of the 3rd person in Slavic languages.

New Bantu language tree derived from statistical model of phoneme evolution
Rebecca Grollemund, Simon Branford, Andrew Meade and Mark Pagel
University of Reading, Evolutionary Biology Group

Over the last fifty years a number of quantitative methods have been developed to compare and
classify the worlds languages. Recently, historical linguists have begun to explore the use of
sophisticated statistical methods borrowed from biology to produce phylogenetic trees of languages.
Several phylogenetic classifications have been established for Bantu languages (Holden, 2002,
Holden and Gray, 2006 or Rexov, Bastin and Frynta, 2006).
Here we present the first results for the Bantu languages from a new likelihood-based statistical
model that can detect and characterize both irregular and regular phonemic changes. Our model is
implemented in a Bayesian Markov chain Monte Carlo framework that allows us simultaneously to
estimate the phylogenetic tree along with the sound changes. Our results provide a new
classification of 250 Bantu languages covering the whole Bantu area (zones A, B, C, D, E, F, G, H, J, K,
L, M, N, P, R and S), based on phonemically transcribed data from a list of 100 basic vocabulary
words. In addition we derive estimates of the rates of change among phonemes and detect
evidence for regular sound changes that are consistent with linguistic expectations see a partial
example below.

Example of the classification of B50-60-70 languages, with regularizations found

We discuss the nature of the regular and irregular phoneme changes that characterize phoneme
evolution in this family. The analysis of these regular and irregular changes allows us to reach
general conclusions on the process of evolution of languages.
References
Holden, C. J. 2002. Bantu language trees reflect the spread of farming across sub-Saharan Africa: a
maximum-parsimony analysis. Proceedings of the Royal Society of London. Series B:
Biological Sciences, 269:1493, pp. 793-799.
Holden, C. J. & Gray, R. D. 2006. Rapid radiation, borrowing and dialect continua in the Bantu
languages. In: Forster, P. & Renfrew, C. (eds.) Phylogenetic Methods and the Prehistory of
Languages. Cambridge: MacDonald Institute for Archaeological Research, pp. 19-31.
Rexov, K., Frynta, D. & Zrzav, J. 2005. Cladistic analysis of languages: Indo-European classification
based on lexicostatistical data. Cladistics, 19:2, pp. 120-127.

Revisiting the Central Franconian Tonogenesis

Carlos Gussenhoven Radboud University Nijmgeen, The Netherlands
The scenario for the Central Franconian tonogenesis in Gussenhoven (2000) has been
supported by other research, but has also met with a number of objections. The 2000 account
is based the resolution of a conflict in 13th-century Cologne Franconian between the
analogical lengthening of singular noun forms whose plurals had undergone Open Syllable
Lengthening (as occurred in German, from dax - daa to daax - daa) and the avoidance
of homophony between singulars and plurals after a final schwa had been lost from the plurals
(from dax daax to [da:x] daax), where the phonetically lengthened vowel failed to merge
with long [aa] and was interpreted as tonally distinct). On the positive side, there is previously
inaccessible work by Garza (1992) in which forms that were inferred in 2000 are attested for
present-day Cologne. Also, recent work by Khnlein (2011) presents data for the dialect of
Arzbach (east of the Rhine) that are readily understood as a further step in the development
that was reconstructed for the dialects of Mayen and Roermond. In particular, one further step
in the phonetic truncation of the Accent 2 declarative contour that gave rise to the
Mayen/Roermond dialects will lead to a system in which the declarative forms for the two
word accents appear to have reversed, exactly as found in Arzbach. While this finding does
not directly support the tonogenesis, it is in agreement with the hypothesized developments
after the tone had arisen. This contribution will summarize the account and deal with a
number of objections that have been or could be raised against it:
1. The account does not explain that, in Limburgish dialects, open, long vowels that arose
through Open Syllable Lengthening are tonally distinct from originally long vowels, if tone
only arose after Open Syllable Lengthening had occurred (Boersma, voce; Roos 2009).
2. The prediction that singulars have Accent 2 and plurals have Accent 1 is disconfirmed for
vat vaten in a number of Limburgish dialects (Roos 2009).
3. On the assumption that Accent 2 is represented by a privative tone, singulars have a richer
representation than plurals, which goes against a Greenbergian universal (Roos 2009).
4. The scenario presupposes intensive dialect contact in 13th century Cologne, but none has
been advanced.
5. The reconstructed tonogenetic dialect closely resembles the present-day Cologne dialect,
while dialects that are further removed have undergone major phonological changes. This is
not in agreement with patterns of phonological change that are seen in centres of prestige and
more distant, less prestigeous areas.
Garza, P. (1992). Kontrastive Satzintonationsanalyse von Stadtklner Mundart und
Hochdeutsch. PhD dissertation. University of Cologne.
Gussenhoven, C. (2000a). On the origin and development of the Central Franconian tone contrast. In A. Lahiri
(Ed.), Analogy, levelling, markedness: Principles of change in phonology and morphology (pp. 215-260). Berlin:
Mouton Gruyter.
Gussenhoven, C. (to appear). An account of the Franconian tone reversal: From Cologne to
Arzbach. In Asu, E.L. & Lippus, P. (Eds.) Nordic Prosody. Proceedings of the XIth Conference, Tartu 2012.
Frankfurt am Main: Peter Lang.
Khnlein, B. (2011). Rule reversal revisited. Synchrony and diachrony of tone and prosodic structure in the
Franconian dialact of Arzbach. Utrecht: LOT.
Roos, Nieke (2009). The weak past tense in Dutch and Low German. PhD dissertation. Radboud University
Nijmegen.

Evaluating alleged typological changes which took place before the earliest
written documents
Jadranka Gvozdanovi, Heidelberg University

How can we evaluate hypotheses about language contacts which allegedly took place before
the earliest written documents? As a preliminary to answering this question, we must consider
the possibility that the alleged effects of contacts were in fact not due to contacts at all, but
rather changes which arose by spontaneous system-internal developments. This is the default
assumption if no change of type can be established. Clearly, we should watch out for the
pitfalls of circularity and compare contents, distribution and relative chronology of
developments in genetically related languages in order to evaluate any alleged modifications
of type. If then a hypothesis about effects of language contacts producing typological
modification(s) remains, this must jibe with a plausible source of contacts. Here, a minute
examination of the properties of the alleged source language(s) and the relative chronology of
their changes must be undertaken in comparison with the properties of the alleged target
language(s), as was done e.g. by Hewitt (2007) and Isaac (2007) who evaluated the HamitoSemitic hypothesis about modifications in Insular Celtic and pointed to essential differences
between the allegedly similar properties of Hamito-Semitic and Celtic. Moreover, likelihood
that a presumed source language may have been sufficiently dominant to trigger language
changes should be investigated language-internally on the basis of other language levels
potentially showing effects of the same contacts, and language-externally by examining
archaeological, historical and any other potential evidence of contacts (in line with
Gvozdanovi 2009).
The present paper follows the methodology proposed above by discussing critically the
proposed reconstructions of verb-phrase properties, including word order, which appear as
relatively specific in Indo-European. The result of the analysis shows that the distribution of
these phenomena is less restricted than supposed, and that they pertain to a conglomerate of
phenomena (partly connected with aspectual categories) whose origin can be ascribed to a
dialectal Indo-European development. Later developments exhibit geographic variation, but
neither their contents nor the external data provide sufficient evidence for deriving their origin
from contacts with non-Indo-European systems. The paper ends by discussing the alternative
solution of an early Indo-European dialectal differentiation and possibilities of its evaluation.
References
Gvozdanovi, Jadranka (2009) Celtic and Slavic and the Great Migrations. Heidelberg: Winter.
Hewitt, Steve (2007) Remarks on the Insular Celtic / Hamito-Semitic question, Karl, Raimund & David Stifter
Karl (eds.) The Celtic World: Critical concepts in historical studies,Volume IV: Celtic Linguistics, 230-268.
London & New York: Routledge.
Isaac, Graham R. (2007) Celtic and Afro-Asiatic, Tristram, Hildegard L.C. (ed.) The Celtic Languages in
Contact, 25-80. Potsdam: The Potsdam University Press.
Jadranka.Gvozdanovic@slav.uni-heidelberg.de

Eurolinguistics from a diachronic perspective

Sabine Husler, Halle/Saale
A new linguistic field has developed during the last 20 years - Eurolinguistics. For the
moment research concentrates mostly on synchronic linguistics, but as our former works in
adverbial subordination in Europe form a diachronic perspective has shown the position of the
centre vs. periphery of a linguistic Europe may differ during time (e.g. our data from the
middle ages compared to the results from Kortmanns mostly synchronic comparison of
European subordinators) and furthermore that not only converging but also diverting
structures can be found within the different branches (e.g. the Germanic languages).
The aim of the paper is twofold, 1st to check the established morpho-syntactic Euroversals
against diachronic findings and 2nd to compare them with the data from the Indo-European
languages outside of Europe to exclude a genetic origin.

Dcsy, Gyula (1973): Die linguistische Struktur Europas: Vergangenheit Gegenwart

Zukunft. Wiesbaden: Harrassowitz 1973.
Husler, Sabine, Polyfunktionalitt und Desambiguierung: Zum satzwertigen
Ausdruck von Adverbialrelationen an der Peripherie Europas : mit einem
vergleichenden Ausblick auf das Indoiranische, Hamburg, 2011.
Haspelmath, Martin. 2001. The European linguistic area: Standard Average European.
Handbuch der Sprach- und Kommunikationswissenschaft vol. 20.2, pp. 1492-1510.
Haspelmath, Martin et al. (eds., 2005): The World Atlas of Language Structures,
Oxford: Oxford University Press.
Heine, Bernd & Kuteva, Tania (2006): The Changing Languages of Europe, New
York/Oxford: Oxford University Press.
Hinrichs, Uwe (ed.), Handbuch der Eurolinguistik, Wiesbaden, 2010.
Kortmann, Bernd, Adverbial subordination : a typology and history of adverbial
subordinators based on European languages, Berlin, 1996.
Lewy, Ernst (1964): Der Bau der europischen Sprachen, Tbingen: Niemeyer.
Ureland, P. Sture (ed., 2003): Convergence and Divergence of European Languages
(Studies in Eurolinguistics 1). Berlin: Logos.

Keywords: Eurolinguistics, Diachronic linguistics, Language change, Comparative linguistics

Camiel Hamans

European Parliament/Anne Vondeling Stichting Brussels

& University of Amsterdam
From mono- to disyllabic clipping templates
A study of a recent language change

This paper aims at describing and explaining a recent change in preferences for certain minimal word templates.
The data which will be discussed are two types of clipping, the older one that results in monosyllabic clipped forms
and a more recent type that results in disyllabic clipped forms ending in a full vowel [o]. The data come from
English and Dutch. The theoretical approach is that of Prosodic Morphology, since traditional concatenative
morphology cannot account for clipping because the phenomenon does not rely on the chaining of morphemes
(Wiese 2001:131/2).
In English and Dutch one finds widely accepted truncated forms such as:
English
(1) temp
<

temperature

Dutch
(2)
Jap

<

Japanner

japanese

In addition to this type of examples one can find another disyllabic category such as:
(3) psycho

<

psychopath

(4)

aso

<

asociaal

antisocial

For Dutch examples as the example shown in (4) have been described by Van de Vijver (1997) and Hinskens
(2001). Their starting point is a disyllabic trochaic template, which is in accordance with the preferred metrical
pattern for minimal Dutch words (Kooij & Van Oostendorp 2003). However, this description fails to account for
the older group of Dutch clippings, as presented in (2).
Lappe (2007) is an extensive study of English truncated forms. The basic structure she starts with is monosyllabic.
Therefore she has no problem in explaining the examples in (1) or when applied to Dutch in (2). However, for the
examples in (3) and possibly in (4) she has to introduce a subsequent process of suffixation, which may work well
for examples as presented in (5) and (6) but leads to a strange reintroduction of a vowel [-o], which was already
present in the input form of (3) and (4).
(5) journo

<

journalist

(6)

alto

<

alternatief

alternative

Both descriptions, Van de Vijver and Hinskens on the one hand and Lappe on the other, do not take into account
the diachronic aspect and the historic origin of the o forms.
In this paper it will be shown that both languages show a development from an initial stage with a preference for
monosyllabic clippings to a later stage with an influx of o clippings, most likely under the influence of a HispanoAmerican English informal language (cf. Hamans 2012). This development, that also lead to an extension of this
kind of o suffixation to full words (as in sicko, weirdo and Dutch lullo dull person), can be expressed in terms
of a re-ranking of constraints, as in (7)
(7) Old system:
New system:

ANCHOR-LEFT >> TROCHEE

TROCHEE >> ANCHOR-LEFT.

REFERENCES
Hamans, Camiel (2012). From prof to provo: some observations on Dutch clippings. Bert Botma and Roland Noske (eds.).
Phonological Explorations, Empirical, Theoretical and Diachronic Studies. Berlin/Boston: Walter de Gruyter: 25-40.
Hinskens, Frans (2001). Hypocoristische vormen en reductievormen in het hedendaagse Nederlands. Neerlandica Extra Muros
39: 37-49.
Kooij, Jan and Marc van Oostendorp (2003). Fonologie. Uitnodiging tot de klankleer van het Nederlands. Amsterdam:
Amsterdam University Press.
Lappe, Sabine (2007). English Prosodic Morphology. Dordrecht: Springer.
Vijver, Ruben van de (1997). The duress of stress: On Dutch clippings. Jane Coerts and Helen de Hoop (eds.). Linguistics in
the Netherlands. Amsterdam/Philadelphia: John Benjamins: 219-230.
Wiese, Richard (2001). Regular morphology vs. prosodic morphology? The case of truncation in German. Journal of
Germanic Linguistics 13: 131-177.

Katja Hann, University of Cologne (Germany)

Traces of Pukina: Reconstructing Pukina on the basis of Kallawaya

The present paper will be concerned with the partial reconstruction of the extinct Pukina
language. Until latest the 19th century (see Adelaar and van de Kerke 2009: 125), Pukina was
spoken at the north-eastern shores of Lake Titicaca (Bolivia), but became extinct almost
undocumented. The only surviving material of Pukina is a collection of Christian texts and
prayers from the early 17th century (dOr 1607; see also Torero 1965), which contains about 260
words (see Cerrn-Palomino 2012: 273; for the most recent research on Pukina, see Adelaar and
van de Kerke 2009). However, it is assumed (see Stark 1972: 200, 206, 224) that considerable
parts of the Pukina lexicon are preserved in Kallawaya, which is a mixed and secret language of
the same region where Pukina was once spoken. The Kallawaya language is used only by
traditional herbalists and, as a mixed language, it takes its lexicon from different and genetically
unrelated languages: while the grammar of Kallawaya derives from Quechua, its lexicon comes
mainly from Pukina (see Stark 1972: 200, 224, 226). Quechua and Aymara contributed to the
Kallawaya lexicon to a minor degree (see Stark 1972: 206; for an overview of Kallawaya, see
Muysken 2009). Although Kallawaya is a secret language, it is reasonably well documented (see
e.g. Girault 1989).
In the present paper, it will be argued that parts of the Pukina lexicon can be reconstructed on the
basis of Kallawaya. For this, etymological and phonological features will be taken into
consideration.
First, it is assumed that those words of Kallawaya that have no recognisable etymology, i.e.
which are neither Quechua nor Aymara, are likely to be of Pukina origin.
Second, from the scarce material on Pukina it is evident that the language had a number of
phonological features not found in Quechua and Aymara. Among these features are:
o the phonemic value of /e/ and /o/;
o distinctive vowel length;
o word-initial consonant clusters.
The assumption is that Kallawaya words without an established etymological origin and which, in
addition, show at least one of the phonological features listed above, are actually Pukina.
In the presentation, the value of comparative etymology and phonology as a means for
reconstructing Pukina will be discussed as well as the degree to which Pukina can be
reconstructed on the basis of these methods.
References
Adelaar, Willem and Simon van de Kerke. 2009. Puquina. In: Mily Crevels and Pieter Muysken
(eds.), Lenguas de Bolivia. mbito andino. Vol. I. La Paz, Bolivia: Plurales editores. 125146.
Cerrn-Palomino, Rodolfo. 2012. Unravelling the Enigma of the Particular Language of the
Incas. In: Paul Heggarty and David Beresford-Jones (eds.), Archaeology and Language in the
Andes. Oxford/New York: Oxford University Press. 265-294.
Girault, Louis. 1989. Kallawaya: el idioma secreto de los Incas: Diccionario. La Paz, Bolivia:
Unicef, Panamerican Health Organisation (OPS) and World Health Organisation (OMS).
Muysken, Pieter. 2009. Kallawaya. In: Mily Crevels and Pieter Muysken (eds.), Lenguas de
Bolivia. mbito andino. Vol. I. La Paz, Bolivia: Plurales editores. 147-167.
Or, Luis Jeronimo de. 1607. Rituale seu Manuale Peruanum. Naples.
Stark, Louisa R. 1972. Machaj-Juyai: Secret language of the Callahuayas. In: Papers in Andean
Linguistics, Volume 2, No. 2. Madison, Wisconsin. 199-229.
Torero, Alfredo. 1965. Le puquina, la troisime langue gnrale du Prou. Doctoral dissertation.
Sorbonne, Paris.

The Peculiar Dative Marking of Giver with Verbs of Receiving in Japanese

Tomoko O. Hansen
University of Oslo
t.o.hansen@ikos.uio.no
This study aims at tracing the use of the peculiar dative marking of giver in Japanese by
using corpus of classical Japanese.
Dative particle ni in Japanese is used to denote a large number of different senses, but can
be attributed to two basic meanings, the goal meaning and the stative locative meaning. And
other senses can be accounted for as extensions of these two meanings (Hansen 2004). This
dative has also been used to denote recipient with verbs of giving, causee with causative
verbs, passive agent with passive verbs, and giver with verbs of receiving in Japanese.
Dative marks goal and recipient in many languages. Dative causee marking across
languages is also well attested in typological studies. Dative marking of passive agent is rare,
but is attested in a few languages. However dative marking of giver is extremely scarce. To
the best of my knowledge, only Korean (with a limited number of verbs) and Japanese mark
giver with dative. Many languages in the world use ablative such as from and of(f). This is
because objectively giver is the source of thing recipient receives, as in I received a book from
her (her = giver). Japanese uses also ablative to mark giver along with dative.
Dative marking of passive agent was already attested in the earliest literal period (i.e. Old
Japanese, 700-800 AD) (Kinsui 1997, Kawamura 2012), but dative marking of giver has
never been investigated. So where does this dative marking of giver come from? Is it a later
development? How long has it been used? In this presentation, the result of the corpus
investigation of major receive verbs from Late Old Japanese (800-1200 AD) will be presented
and the implication of the findings will be discussed. The corpus data for this study are taken
from the digital corpus of classical Japanese which is currently under development at National
Institute for Japanese Language and Linguistics (NINJAL).

References
Kawamura, Futoshi. 2012. Raru-kei Jutsugobunno Kenkyuu, Tokyo: Kuroshio Publishers
Kinsui, Satoshi. 1997. The influence of translation on the historical development of the
Japanese passive construction. Journal of Pragmatics 28: 759-779.
Hansen. Tomoko O. 2009. The Japanese Dative: A Cognitive Analysis. Saabrcken:VDM
Verlag.

Constructional Change at the Interface of Cognition, Culture, and Language Use:

A Diachronic Corpus Study of German Nominalization Patterns
Stefan Hartmann, University of Mainz
In the past few years, the diachronic development of German nominalization patterns has
been investigated from a variety of perspectives and in several frameworks (e.g. Barz 1998;
Demske 2000). In particular, the competing word-formation patterns of ung-nominalization
(e.g. Landung landing) and Infinitival Nominalization (e.g. (das) Tanzen dancing) have
been studied extensively with a focus on changes of word-formation constraints (e.g. Ehrich
& Rapp 2000; Werner 2010). However, some questions have not been answered satisfactorily
so far. For example, the constraints affecting ung-nominalization have not been described
exhaustively yet. Furthermore, there have only been tentative first steps towards an
explanatory account of the diachronic development of both word-formation patterns.
This paper tries to fill that gap by proposing a construal approach to word-formation change.
Drawing on insights from Cognitive Linguistics (e.g. Langacker 1987, 1991; Talmy 2000)
and usage-based approaches to language (e.g. Taylor 2012), a theoretical framework is
developed that allows for a coherent, cognitively plausible analysis of the phenomena in
question, which qualify as constructional change in Hilperts (2013) sense. On the basis of an
extensive corpus analysis of Middle High German (MHG, 1050-1350), Early New High
German (ENHG, 1350-1650), and New High German (NHG, 1650-today) texts it is shown
that the constraint changes affecting both ung-nominalization and Infinitival Nominalization
can be accounted for as changes in the availabilty of construal options: While the default
construal of MHG and ENHG ung-nominals is highly processual, their NHG equivalents
exhibit a bounded region construal (Langacker 1987), i.e. they are construed in a much more
reified fashion. Lexicalization of frequent word-formation products and subsequent reanalysis
give rise to new construal patterns, whereas others come out of use, rendering the respective
construal options unavailable. Infinitival Nominalization comes in as a replacement process
(Barz 1998) for ung-nominalization. These developments become obvious not only in the
analysis of the word-formation products, but also in a comparison of derivatives with their
respective base verbs in a force-dynamic framework (Talmy 2000; Croft 2012). Furthermore,
cultural influences governing the use of nominalization patterns are taken into account (cf.
e.g. Baayen 2009), thus considering the interrelation between cognition, culture and
[language] use (Bybee 2010).
References
Barz, I. (1998): Zur Lexikalisierungspotenz nominalisierter Infinitive. In: Barz, I.; hlschlger, G.
(eds.): Zwischen Grammatik und Lexikon. Tbingen: Niemeyer, 5768.
Baayen, R. H. (2009): Corpus Linguistics in Morphology. In: Ldeling, A.; Kyt, M. (eds.): Corpus
Linguistics. An International Handbook. Vol. 2. Berlin, New York: De Gruyter (HSK, 29.2), 899
919.
Bybee, J. L. (2010): Language, Usage and Cognition. Cambridge: Cambridge University Press.
Croft, W. (2012): Verbs. Aspect and Causal Structure. Oxford: Oxford University Press.
Demske, U. (2000): Zur Geschichte der ung-Nominalisierung im Deutschen. Ein Wandel
morphologischer Produktivitt. In: Beitrge zur Geschichte der deutschen Sprache und Literatur
122, 365411.
Ehrich, V.; Rapp, I. (2000): Sortale Bedeutung und Argumentstruktur. -ung-Nominalisierungen im
Deutschen. In: Zeitschrift fr Sprachwissenschaft 19, 245303.
Hilpert, M. (2013): Constructional Change in English. Cambridge: Cambridge University Press.
Langacker, R. W. (1987, 1991): Foundations of Cognitive Grammar. 2 Vol. Stanford: Stanford
University Press.
Talmy, L. (2000): Toward a Cognitive Semantics. 2 vol. Cambridge, Mass: MIT Press.
Taylor, J. R. (2012): The Mental Corpus. How Language is Represented in the Mind. Oxford: Oxford
University Press.
Werner, M. (2010): Substantivierter Infinitiv statt Derivation. In: Bittner, D.; Gaeta, L. (eds.):
Kodierungstechniken im Wandel. Berlin, New York: De Gruyter, 159178.

From Quantity to Height: Diachronic Change in the Preferences of Basic Scalar

Adjectives for Nouns Denoting a Gradable Property in Japanese
TADASU

HATTORI

Doshisha Womens College of Liberal Arts

In this study, using a corpus comprising the Minutes of the Japanese Diet from 1947 to
2006, I argue that there has been a change in the preferences of basic scalar adjectives
such as takai high, ookii large, tuyoi strong, koi dense, and ooi much used as the
predicate for nouns with an underlying gradable property, such as kanoosei probability,
possibility. For example, all of the following noun-adjective collocates are attested in
the corpus, and they are nearly synonymous in most of the cases.
(1) kanoosei-ga
probability-NOM

{takai/ookii/tuyoi/koi/ooi}
{high/large /strong/dense/much(many)}

Lit. (the) probability (is) {high/large /strong/dense /much(many)}

However, a diachronic analysis reveals that the relative frequency of takai high in the
above-mentioned collocation pattern has remarkably increased, while the relative
frequency of ooi much(many) has been in constant decline during the 19472006 period,
as I pointed out in Hattori(2011).
Examining the data more generally, we find that, out of the 168 nouns with a
minimum of 50 occurrences in the collocate patterns in question for each of three
20-year periods (19471966, 19671986, and 19872006), 63 nouns (e.g., kanoosei
probability, hituyoosei necessity, kikensei dangerousness, hiritu rate, nooryoku
ability) show a significantly increased frequency of co-occurrence with takai high,
while 84 nouns (e.g., kikensei dangerousness, yoti room, latitude, kanoosei
probability) show a significantly decreased frequency of co-occurrence with ooi much.
Therefore, we may speculate that the general tendency is as follows: the default mental
image utilized in expressing a larger value of some abstract scalar property has been
shifting from quantity to height or one-dimensional size. Partly, this change may be due
to the ambiguity of ooi (being) much resulting from the lack of mass-count distinction
in Japanese nouns.
Reference
Hattori, Tadasu (2011) Cooccurrence patterns between nouns denoting gradable
properties and scalar adjectivals: A diachronic study. (in Japanese with English
abstract). Gengo Kenkyu 140, 89116. The Linguistic Society of Japan.

Ulrike HEIDEMEIER & Mihaela-Mariana MORCOV

Laboratoire ATILF (CNRS & Universit de Lorraine, Nancy/France)

From Romance to Proto-Romance:

On reconstructing the conjugation of PRom.*/a-'e-re/ to have and */'ss-e-re/ to be
Since the founding of Romance philology by Friedrich Diez in the first half of the 19th
century, Romance etymology has been based on the written evidence provided by Classical
Latin. As a written language per definitionem cannot attest directly a spoken language, this
method goes against the principles of the reconstruction of proto-languages in historical
linguistics. This approach has not been accepted by the majority of Romance linguists of the
19th and 20th centuries, though: nearly any etymological discussion is focused on the
development of Romance languages in the direction from Latin to Romance. Only two
linguists recognised the need for comparative grammar in Romance etymology: Robert Hall
and Robert de Dardel. Since 2008, their work has been continued by an international group of
linguists, who are preparing the Dictionnaire tymologique Roman (DRom), under the
direction of va Buchi (Nancy/France) and Wolfgang Schweickard (Saarbrcken/Germany).
Their initial aim is to reconstruct the inherited lexicon of Proto-Romance, which survives in
all Romance languages (circa 500 words).
This talk seeks to go beyond the frontiers of lexical reconstruction and to analyse the
grammar of Proto-Romance. We will study two of the possibly most frequent words in all
languages, /to have/ and /to be/, in order to reconstruct their present tense conjugation. This
work is particularly interesting, as it since Hall (who presents some inexactness in his
analysis) has never been undertaken in Romance etymology: the discussion is still based on
the development of the written Latin forms habe, habs, habet and sum, es, est. We proceed
heuristically and try, in a first step, to identify the authentic inherited forms of the present
tense of PRom. */a-'e-re/ and */'ss-e-re/ in Romanian, Italian, Sardinian, French, Spanish,
Catalan, Portuguese etc. Our aim is to elaborate the definitive list of cognates by detecting and
excluding non-inherited analogical forms. On the basis of a micro-structural reconstruction of
each proto-form in each Romance language we can proceed to the macro-structural
reconstruction of the primitive inflected forms of PRom. */a-'e-re/ and */'ss-e-re/.
In this way, the talk aims to contribute to the global reconstruction of Proto-Romance
and to identify the rules of language change in Proto-Romance and Romance. It reveals the
advantages of comparative grammar, which has been in use in all language families since its
founding, and which is overdue to be applied and finally accepted in Romance etymology.
References:
Buchi, va (2010): Where Caesars Latin does not belong : a comparative grammar based approach to Romance
etymology. In: Brewer, Charlotte (dir.): Selected Proceedings of the Fifth International Conference on Historical
Lexicography and Lexicology held at St Annes College, Oxford, 16-18 June 2010, Oxford, Oxford University Research
Archive (http://ora.ox.ac.uk/objects/uuid%3A237856e6-a327-448b-898c-cb1860766e59).
Buchi, va (in press): Cent ans aprs Meyer-Lbke : le Dictionnaire tymologique Roman (DRom) en tant que tentative
darrimage de ltymologie romane la linguistique gnrale. In: Casanova, Emili & Calvo Rigual, Cesreo (dir.) : Actes
del 26 Congrs Internacional de Lingstica i Filologia Romniques (Valncia 2010), Berlin/New York, De Gruyter.
Campbell, Lyle (22004) [11998]: Historical Linguistics. An introduction, Edinburgh, University Press.
Dardel, Robert de (1996): la recherche du protoroman. Tbingen, Narr.
DRom = Buchi, va & Schweickard, Wolfgang (dir.) (2008): Dictionnaire tymologique Roman (DRom). Site Internet :
Nancy : ATILF (http://www.atilf.fr/DERom).
Fox, Anthony (1995): Linguistic reconstruction. An introduction to theory and method, Oxford, University Press.
Hall, Robert A. Jr. (1974): Comparative Romance Grammar: 1. External History of the Romance Languages, New York,
American Elsevier.
Hall, Robert A. Jr. (1976): Comparative Romance Grammar: 2. Proto-Romance Phonology, New York, American Elsever.
Hall, Robert A. Jr. (1983): Comparative Romance Grammar: 3. Proto-Romance Morphology, Amsterdam/Philadelphia, John
Benjamins Publishing Company.
Meillet 1984 = Meillet, Antoine (1984 [1925]): La Mthode comparative en linguistique historique, Paris/Genve,
Champion/Slatkine.
REW = Meyer-Lbke, Wilhelm (193019353 [191119201]): Romanisches Etymologisches Wrterbuch, Heidelberg, Winter.

Detecting diachronic signal obscured by synchronic noise

Rachel Hendery, The Australian National University
Palmerston Island is an isolated 54-person community in the Pacific. It was settled in the mid
19th century by a small group of settlers: one Englishman, a Portuguese or Portuguese creole
speaking man, and a small group of Cook Islanders (Hendery forthcoming). The current
residents trace their descent to the Englishman, Marsters, and his three wives. They are
monolingual in a dialect of English that shows some English/Polynesian mixing (see Ehrhart
1996, Hendery & Ehrhart 2013 for details).
In this paper I will use historical records from this island from 1888-1959, as well as
fieldwork data collected in 2009 and 2010 to look at the question of how we can distinguish
between language change and social or idiosyncratic variation (of the sort described in e.g.
Dorian 1994, 2010). This is a problem that all historical linguists face. Even when it is
possible to abstract across synchronic variation to describe a 'standard language' for a certain
point in time, it is dangerous to compare that with an earlier 'standard' which may not be a
direct antecedent, but might instead represent a different regional or social variety. Rather we
ideally compare, where available, a broad range of varieties at each level, concluding that
change has occurred only when the difference across time is greater than the variation within
a time period.
Palmerston Island magnifies this problem because of its small size and short history. The
nave approach would consider any differences between the earlier documents and the recent
recordings to be indicative of language change, while differences among speakers or writers
at each time period represent sociolinguistic or idiosyncratic variation. However, when I use
exploratory multivariate analyses (as used by e.g. Grieve et al 2011, Moisl et al 2006, or
Shackleton 2007) to detect natural 'groupings' among the speakers, I find that for at least some
features the earlier speakers are more similar to some of the current speakers than the current
ones are to each other. I discuss what we can infer from this about the historical development
of Palmerston Island English, how social, idiosyncratic and diachronic variation have
interacted with each other, and I conclude by talking about other statistical approaches which
can help us detect diachronic signal in synchronic noise.
References
Dorian, N. C. (1994). Varieties of variation in a very small place: Social homogeneity, prestige norms,
and linguistic variation. Language, 631-696.
Dorian, N. C. (2010). Investigating Variation: The Effects of Social Organization and Social Setting:
The Effects of Social Organization and Social Setting. Oxford: OUP.
Ehrhart-Kneher, Sabine. (1996). Palmerston English. In Stephen Wurm and Peter Mhlhusler, eds.
Atlas of Languages of Intercultural Communication in the Pacific, Asia, and the Americas:
Texts. Vol. 2. Mouton de Gruyter, 1996.
Grieve, J., Speelman, D., & Geeraerts, D. (2011). A statistical method for the identification and
aggregation of regional linguistic variation. Language Variation and Change, 23(02), 193-221.
Hendery, Rachel. (Forthcoming). Early documents from Palmerston Island and their implications for
the origins of Palmerston English. Journal of Pacific History.
Hendery, Rachel & Sabine Ehrhart. (2013). Palmerston Island English. In B. Kortmann and
Lunkenheimer, K. (Ed.) The Mouton World Atlas of Variation in English. Germany: De Gruyter
Mouton.
Moisl, H., Maguire, W., & Allen, W. (2006). Phonetic variation in Tyneside: exploratory multivariate
analysis of the Newcastle Electronic Corpus of Tyneside English Newcastle Electronic Corpus
of Tyneside English.
Shackleton Jr, R. G. (2007). Phonetic Variation in the Traditional English Dialects A Computational
Analysis. Journal of English Linguistics, 35(1), 30-102.

Stability of idiolects in unstable times: life stages and lifespan

changes of immigrants in the early modern Dutch urban context
Jennifer Hendriks
Australian National University
In this study I examine changes in the language of personal letters across the
lifespans of several individuals with the goal of shedding light on the contribution
these people may have made in the formation of Early Modern Dutch urban
vernaculars. The individuals selected for this study are of particular interest due to
their status as immigrants. Urban centers in Holland experienced a doubling and
tripling of populations over a short period of time (roughly 1580-1650). Given the
high urban mortality rates, such a population explosion could only have been
achieved by a massive influx of migrants. While the historical details of this period
have been extensively researched, the linguistic effects of massive migration on the
emerging urban Dutch vernaculars is virtually undocumented and remains poorly
understood. Using a single-genre corpus of several hundred autograph personal
letters written by men and women (family members) sharing similar demographic
characteristics, I explore changes across the lifespans of individuals who lived during
this period of extreme social, political, economic and religious instability. I discuss
three morphological changes (2nd person singular pronouns, diminutives and the 3rd
person reflexive pronouns), selected on the basis that if we assume there was no
possibility of idiolectal change later in life, these immigrants could not have
contributed to these changes. Preliminary analysis of the data, however, shows that
with respect to two of the changes (the diminutives and the reflexive pronouns), at
least some of the individuals could have participated in the changes taking place in
the Early Modern Dutch urban vernaculars.
It has long been assumed that idiolects become relatively fixed at a certain point in a
speakers lifeusually not later than post-adolescence. An increasing number of
contemporary as well as historical studies, however, offer evidence of idiolectal
change in later life stages (cf. Nahkola and Saanilahti 2004, Raumolin-Brunberg
2005, 2009, Sankoff and Blondeau 2007). Thus, the question has become not
whether idiolects can continue to change across a lifespan, but rather why and how
some individuals idiolects may change later in life and what this may mean in terms
of understanding changes observed for a given language period or in certain
linguistic communities. This study provides longitudinal data that helps to shed light
on the question of language change in individuals over time.

Nahkola, Kari, and Maria Saanilahti. 2004. Mapping changes in real time: A panel study on
Finnish. Language Variation and Change 16.75-92.
Raumolin-Brunberg, Helena. 2005. Language change in adulthood. Historical letters as
evidence. European Journal of English Studies 9(1).37-51.
Raumolin-Brunberg, Helena. 2009. Lifespan changes in the language of three early modern
gentlemen. The language of daily life in England (1400-1800), ed. by Arja Nurmi, Minna
Nevala, and Minna Palander-Collin, 165-196. Amsterdam/ Philadelphia: John Benjamins.
Sankoff, Gillian, and Hlne Blondeau. 2007. Language change across the lifespan: /r/ in
Montreal French. Language 83.560-588.
Unpublished archival sources:
Bibliotheca Thysiana, Universiteitsbibliotheek Leiden, The Netherlands
Collectie Daniel van der Meulen, Gemeentearchief Leiden, The Netherlands

An accessibility theory approach to distinguishing definite articles and third person

personal pronouns from demonstratives in Late Latin
Mari Johanne Hertzenberg, University of Oslo

Classical Latin had no definite article and no third person personal pronoun. The modern
Romance languages, on the other hand, have both. In all Romance languages, the definite article
and third person pronoun can be traced back either to the distal demonstrative ille or to the
intensifier ipse in Classical Latin (in Late Latin, ipse is also a demonstrative). At some point
between Classical Latin and modern Romance, then, ille and ipse were transformed into definite
articles and third person pronouns. It has long been a matter of debate when this change occurred.
Indeed, in the transitional stage between demonstrative and definite article / personal pronoun,
when there is no formal difference between the categories, it is not obvious when the linguistic
item in question is a demonstrative, and when it has become a definite article or third person
pronoun.

This paper is a case study of ille and ipse in the Late Latin text commonly known as the
Itinerarium Egeriae. I use data from the electronic PROIEL corpus1 to show that accessibility
theory and Grices maxim of quantity can determine the correct categorial status of ille and ipse in
their transitional stage between demonstrative and definite article / third person pronoun.

The development from demonstrative to definite article did not only occur in Latin/Romance. In
fact, demonstratives are a common source for definite articles cross-linguistically (see e.g.
Greenberg 1978, Diessel 1999). The method I propose for distinguishing demonstratives from
definite articles in Late Latin should be applicable to other languages as well.

References:
Diessel, Holger. (1999). Demonstratives: form, function, and grammaticalization. Amsterdam:
John Benjamins.
Greenberg, Joseph H. (1978). How does a language acquire gender markers? In J. H. Greenberg,
C. A. Ferguson & E. A. Moravcsik (Eds.), Universals of human language (Vol. 3, Word
structure, pp. 47-82). Stanford: Stanford University Press.

Available online at http://foni.uio.no:3000

Author
Title

Eugen Hill (Humboldt-Universitt zu Berlin, eugen.hill@hu-berlin.de)

The Prosodic Change Hypothesis from a Neogrammarian perspective

Abstract
Recently Richard Page formulated a hypothesis according to which sound change on
the level of segments may be irregular if it is caused by a change in prosody of a language. When
prosody changes, the distribution of prosodic material on the segmental level may change
accordingly. However, a prosodic requirement may often be met in more than one way. In other
words, prosody does not wholly determine the shape of the phonetic output. From this Page followed
that Unlike Neogrammarian sound change, prosodic change may be irregular in its implementation
on the segmental level. Prosodic requirements may often be satisfied in more than one way and
therefore do not wholly determine phonetic shape. According to Page, this hypothesis helps to
understand why some instances of sound change abound in what appears to be unexplainable counterevidence.
One instance of this kind is the well-known Open Syllable Lengthening in German, where we observe
lengthening of stressed vowels for instance in Middle High German vogel > Modern German Vogel
bird but not in MHG doner > MoG Donner thunder. In Pages view, the different outputs of MHG
vogel and doner in MoG are best explained by assuming that we are dealing with a change in the
prosodic structure of German which could be met equally well by lengthening the vowel such as in
vogel or lengthening the following consonant such as in doner. Similar explanations may be applied to
numerous other cases of apparantly irregular sound changes.
The goal of my paper is two-fold. First, I will show that Pages separation of prosodically motivated
segmental changes despite its obvious plausibility is not the only possible way of interpreting the
data. We know of sound changes which lack a prosodic background but neveretheless display
exceptions still awaiting explanation. We also know of changes in prosody which are perfectly regular
in their implementation on the segmental level. At the same time the most prominent instances of
prosodically caused irregularity seem explanable in a more traditional way. I will show that most
serious exceptions to the Open Syllable Lengthening in German can be accounted for by paying more
attention to correlations between the lack of lengthening and the consonantal environment of the
vowel (which was followed by a stop in MHG vogel but not in MHG doner).
Second, and more importantly, I will anlyse why sound changes with an obvious background in
prosody so often generate the impression of irregularity. I will demonstrate that this impression
usually emerges due to the following two factors:
(a) similar changes with different conditioning in closely related languages or dialects of one
language;
(b) lack of information on prosody which is typical for early stages of languages with an old written
record.
In pursuing this goal I will use particularly telling instances of prosodically motivated sound change
taken from languages of the Yenissean family (which are or were spoken in Southern Siberia), from
different branches of Slavonic and from dialects of Latvian.
References
Page, B. Richard. 1999. The Germanic Verschrfung and prosodic change. Diachronica 16: 297-334.
Page, B. Richard. 2007. On the irregularity of Open Syllable Lengthening in German. In Historical
Linguistics 2005. Selected Papers from the 17th International Conference on Historical Linguistics,
Madison, Wisconsin, 31 July 5 August 2005. Joseph Salmons Shannon Dubenion-Smith (eds.),
337-350. Amsterdam Philadelphia: John Benjamins.

Relative chronology of Burmese sound changes

Nathan W. Hill (SOAS, University of London)
Abstract: There is no tradition of named sound laws in the study of Sino-Tibetan languages. The
lack of this tradition impedes progress. Attempting to bring this august practice into the SinoTibetan family, Hill (2010) presented an inventory of Tibetan sound laws. In this paper I present a
similar list of sound changes for Burmese. However, whereas Hill's list was organized for the
convenient of presentation, I present the list of Burmese changes in chronological order, charting
the development of proto-Burmish to Old Burmese. The following changes, among others, will be
presented.
Initials
Schiefner's conjecture, *sC > *C
Burling's law, *C > Ch
Matisoff's law: *-, *s- > s- and *-, *ts- > tsRimes
Shafer's law, -*ik> -ac, *-i > -a
Maung Wun's law, *u>o before velars

RootClauseswithGerundVerbs
GabrielaAlboiuandVirginiaHill
Issue: Early Modern Romanian (EMR), as attested in the Moldavian chronicles (17th18thc.) displayed a
productive use of gerunds, in both matrix andadjunctclauses. Matrixgerundsmay occur irrespectiveof the
context, including as outoftheblue constructions, see (1a). Modern Romanian (MR) only preserved
gerunds as adjuncts. This paper aims to account for the mechanism that allows gerund verbs to generate
bothfinite/matrixandnonfinite/adjunctclausesinEMR.
Data: The EMR gerund has an invariable (ind(u)) form. It is purely verbal, being incompatible with
determiners (e.g., *mncndul the eating) (Caragiu 1957, Edelstein 1972), on par with its Latin gerund
ancestor (Miller 2000). Also, sentential complements are rare or missing, so gerunds are ruled out as
arguments and have an exclusivelyverbal function. EMR displays gerundsin:(i) rootclauses, in simple (1a)
andcomplexsentences(1b),wheregerundsarecoordinatedwithindicatives.
(1)a.Traianntiu,mpratul,supuindupredahii.
Trajanfirstemperor.theconqueringDOMDacians
First,Trajan,theEmperor,conqueredtheDacians.(Costin/Panaitescu1979:11)
b.IarvdznduccuprinduleiiTaraMoldovei,auornduitpreCaaziChereisoltan,hanul
andseeingthatinvadePolesMoldovahascooptedDOMCaaziChereiSultanhan
Crmului,itriminduii2.000deinicerictr70.000deoasteceaveattrasc.
Crimeeaandsendinghimand2.000ofsoldierstowards70.000ofarmythathadTartar
ButseeingthatthePolesareinvadingMoldavia,hecooptedtheSultanCaaziCherei,the
HanofCrimeea,andsenthim2.000infanterysoldiersthatwasaddedtothe70.000Tartar
armythathehad.(Costin/Panaitescu1979:1415)
Gerunds also occur in relative clauses and in a variety of adverbial clauses, either as absolute
constructions,withlexicalorprosubjects,whichareunexceptionalinRomance.Wefocuson(1).
Proposals: We argue for the following points: (1)Assertion in EMR root clauses can surface with either
gerund or indicative morphology (cf. Adger&Smith2005). Both are propositional, tensed domains that differ
only w.r.t. whether [uphi],an uninterpretablefeature with noeffecton meaning,is present (indicatives)or not
(gerunds). (2) The EMR gerundis underspecifiedfor TAM features hence,itsplurifunctionality yetprojects
to a full CP. (3) EMR gerunds could have an absolute or a relativetense, but the MR gerundlost its deictic
tense,whichexplainsitsabsenceinrootclauses.
Analysis. (i): EMR gerunds can occur with aspectual and temporal adverbs, andtheir aspectualand tense
interpretations are independent fromthose of matrixpredicates. (ii): EMR gerunds allowfor speakeroriented
adverbs (e.g. probably=epistemicmodality),andare precededby topicand focusconstituents. Hence,they
are in CP, in Fin (Rizzi 1997). (iii): The gerund precedes clitics.Since clitics attach to the highest Inflhead
in Romance (Kayne 1991, Uriagereka 1995, DobrovieSorin 1994 for MR), this confirms that the gerund V
moves to Finfor licensing purposes. (iv): EMR gerunds allow for Relative Operator (OP) therefore they
project up to ForceP (Rizzi 1997), although they stay inFinhence,they are CP/phasal domains.(v)While
finite verbs and infinitives in EMR take the free negative morpheme nu, EMR gerunds typically disallow nu
andinstead require the affixal ne. This is predictable from the Neg>T hierarchyin Romance(Zanuttini1997):
nu blocks Head movement above T (i.e. VtoFin, Isac&Jakab 2004 for imperatives), interfering with gerund
licensing in Fin. (vi) Semantically, the mood marker ind, merged in Fin, is underspecified for a particular
value, but its environment assigns it specific interpretation: assertive/realis or irrealis. Therefore, EMR
gerunds have a fully articulated clause structure equivalent to an indicative clause, with the only difference
that TAM features are intrinsic to the latter but not tothegerunds. Whilewith finite verbs T isvaluedvia the
inflectional endings on V (Pesetsky&Torrego 1994), these are absent with gerunds, so T(AM) has to be
recuperated contextually (syntactically or pragmatically). Guron&Hoekstra (1995) argue that Englishverbal
gerunds are headed bya Tense OPin Spec,CP resulting inaTchain for INFLnodes.We similarly adoptthe
requirement of an OP in Spec,ForceP (e.g. Rel OP) but argue instead that in rootclauses, whichhavetruth
values, this has to be an Illocutionary ForceOP, such as Meinungers (2004)Assert(ion) OP.The presence
of AssertOP binds all relevant TAM variables, thereby licensing feature values and deictic/absolute tense,
resulting in main clause status of EMR gerunds. Accordingly, root gerunds disallow questions, since
awhOP either interferes with the binding of TAM variables by the OP in Spec, ForceP, or semantically
contradictstheAssertOP,soitwillalwaysberuledout.

The spread of subjunctive clauses: a Romanian perspective

Virginia Hill
Subjunctive clauses emerged in the Balkan languages as a replacement to the nominalized
infinitive, so the value of T(ense) switched from [-] to [+] finite (Joseph 1983; Roussou 2009). In
Romanian (Rom), the nominalization applies to long infinitives. The emergence and spread of the
subjunctive clause occurred in Early Modern Romanian (EMR), and it is well attested in the 17th -18th
c. texts (Chivu 1997 a.o.). Crucially, the long infinitive in Rom had been nominalized and replaced by
short infinitives (1) or by de-indicatives (2), long before the subjunctive emerges (Frncu 2009).
Hence, the question is: why is it still the case that the subjunctive arises in EMR and replaces the
replacements of the nominalized infinitives (i.e., both (1)&(2)), as in (3)? How was that implemented?
(1)
duhul svnt puse [cercettori a
pasce beserica domnului]
spirit.the holly put shepherds to.INF guard church.the Lord.the.GEN
The holly spirit put shepherds to guard the Lords church.(Frncu 1969: 101/33)
(2)
pusr [de au
fcut peciate rii
Moldovei]
put.3pl
de have.IND.3PL made emblem country.the.GEN Moldova.the.GEN
they put (them) to make an emblem for Moldova(Ureche 1958: 72)
(3)
au pus craiul [pe oameni de ai lui s-i
porunceasc]
has put king.the DOM men of the his SUBJ-to.him order.SUBJ.3
the prince put some of his men to order him around (Costin 1979: 41)
On the basis of EMR texts, this paper argues (in the framework of generative grammar) that
the subjunctive clause emerged from a need to map the irrealis feature in an unambiguous way in
the sentential complements to verbs (main thesis). That is, the conditional complementizer s is reanalyzed as a subjunctive mood marker and replaces the ambiguous de, which served as a
complementizer in both realis and irrealis contexts, with finite or non-finite verbs. Both infinitives and
de-indicatives were replaced, so, in EMR, the triggers and targets for replacement are different than in
other Balkan languages; the replacement of the (short) infinitive is a side effect of a more general
change. Technically, the re-analysis of conditional s as a subjunctive mood marker arises from a
gradient distinction (Roberts 2010) between conditional [Operator] and [modal] features, which
triggers the re-classification of s as an inflectional versus clause typer; further on, there is a split
within [modal], between [irrealis] and [mode], the latter taking the value of beliefs/wishes versus
conditions/options (Saeed 2003), which were the previous values of s.
The analysis is based on observations gathered from an EMR corpus and from secondary
sources, as well as on tests w.r.t. word order, substitution and inflectional restrictions. The tests
identify changes in the left periphery of the s-clause within a cartographic framework (Rizzi 1997).
Briefly, the relevant observations are: (i) EMR has subjunctive verb forms inherited from
Latin (Fischer 1985), but they do not generate a subjunctive clause before the re-analysis of s as a
mood marker. (ii) The short infinitive (i.e., a to + infectum verb stem), which has replaced the long
infinitive, is productive in EMR (Sandfeld 1930), and so it stays in modern Rom, except in the
sentential complements to verbs. (iii) The complementizer de is rare in sentential complements to
verbs in modern Rom, although it was very productive in EMR. (iv) The statistics show that, by the
end of the 18th c., want verbs adopted the subjunctive complement (90%), whereas aspectuals were
very resistant to the switch (2.5%) (Frncu 1969).
The tests show that: (i) Topic and focus constituents follow conditional s but precede
subjunctive s, indicating hierarchical re-analysis at the left periphery. (ii) Restrictions on the mood
morphology of the verb applies to subjunctive s (only indicative and subjunctive) but not to
conditional s (almost any other mood form), indicating concordance between [modal] and
grammatical [mood] in the former but not in the latter. (iii) There is a short lived co-occurrence of de
and s (instead of complementary distribution in conditional clauses), indicating a finer-grained
analysis of [modal] (into [irrealis] de and [mode] s), after which de is dropped as redundant, and only
the marked mapping of [mode] is spelled out (as s).
In conclusion, this paper supports the accounts that argue against a language contact trigger
for the emergence of subjunctives in the Balkans (Demiraj 1970; Philippide 1927). It also supports
diachronic approaches based on gradient feature distinction (Roberts 2010); and contributes evidence
for a finer-grained representation of the syntactic projections that encode modality and mood (Rizzi
1997), and which may account for cross-linguistic variation in the structure of Balkan subjunctives.

The diachronic development of embedded yes/no in Dutch

Jack Hoeksema University of Groningen
Some predicates may embed yes/no answers, as in the following English and German examples:
(1) Q: Are you happy?
A: I believe so/not

Sind sie glcklich?

Ich glaube ja/nein.

Note that English uses polarity markers of indirect speech (so/not), whereas German uses direct
speech forms (ja/nein yes/no, doch actually, yes). In the period 1700-2000, Dutch underwent a
change from direct marking a la German to indirect marking a la English. This change is almost
completed now, except for two verbs (knikken/schudden nod, shake) which indicate yes/no answers
by head movement, and which retain the old marking by direct-speech forms. Old forms such as
early modern Dutch
(2) Hy seyde jae.
He said yes
He said yes

are gradually replaced by modern Dutch:

(3) Hij zei van wel.
He said of AFFIRM
He said yes

On the basis of a corpus of over 1200 examples from 1600 to the present, it is possible to establish
the following:
(a) The embedding predicates always embed propositions, not properties or questions (more
specifically, predicates that also embed finite dat-clauses); the head motion verbs knikken
and schudden form an exception to this generalization
(b) Factive predicates do not embed yes/no answers
(c) The quotative marker van slowly becomes obligatory
(d) Verbs of saying lag behind verbs of thinking in the change from direct to indirect marking
(e) The rate of change for neen/niet is slower than that for ja/wel, although both changes are
now virtually completed (contra Krochs 1989 constant rate hypothesis)
Quotative van can be found in other constructions as well (cf. Coppen & Foolen 2012), but is usually
not obligatory. The developments form an interesting case where Dutch is syntactically moving away
from German and English, by developing a special marker van. The change from direct to indirect
forms does not seem due to influence from English. English does not become a factor of importance
in the development of Dutch, as measured in loan words, loan translations and such, until the end of
the 19th century, and the change under consideration started in the 18th century, a little early for
English influence. French, which uses direct forms (je pense que non) is not a likely influence either.
References
Coppen, P.-A. & A. Foolen, 2012, Dutch quotative van: Past and Present. In I. Buchstaller & I. van Alphen, eds.,
Quotatives: Cross-linguistic and cross-disciplinary perspectives. Amsterdam/Philadelphia: John Benjamins, 259280.
Anthony S. Kroch (1989). Reflexes of grammar in patterns of language change. Language Variation and
Change, vol. 1,
199-244.

Horst Simon, Freie Universitt Berlin

How old is Standard Average European really?

The history of German as a test case

It seems to be clear from typological studies both qualitative and quantitative ones
that some North-Western European languages (German, Dutch, Frisian, French) stick
out as possessing a number of typologically unusual features (cf., e.g., Haspelmath
2001, Heine & Kuteva 2006, Cysouw 2011): the Standard Average European
sprachbund.
What is unclear, however, is the diachronic scenario that led to this convergence.
Candidates include: language contact in the early medieval Migration Period (van der
Auwera 1998, Haspelmath 1998), shared Latin influence in the Middle Ages and/or
Renaissance, Early Modern standardization processes (Kortmann 2009; assumption
behind Seiler 2011). None of these hypotheses has been supported by a significant
amount of empirical historical data.
In my paper, I am going to test the SAE features suggested in the literature against the
historical facts of German. In particular, I set up a SAE-typological profile of (Old
and) Middle High German. In order to check the standardization hypothesis, I will
also take into account a contemporary dialect, Bavarian, and a related West-Germanic
language with an entirely different standardization history, Afrikaans.
It will turn out that most SAE features are considerably older than standardization in
German.

van der Auwera, Johan. 1998. Conclusion. In: Johan van der Auwera, in collaboration with Dnall P.
. Baoill (eds.). Adverbial Constructions in the Languages of Europe. Berlin & New York: Mouton
de Gruyter. 813-836.
Cysouw, Michael. 2011. Quantitative explorations of the worldwide distribution of rare characteristics,
or: the exceptionality of northwestern European lnaguages. In: Horst J. Simon & Heike Wiese (eds.).
Expecting the Unexpected: Exceptions in Grammar. Berlin & New York: De Gruyter Mouton. 411432.
Haspelmath, Martin. 1998. How young is Standard Average European? Lnguage Sciences 20, 271-287.
Haspelmath, Martin. 2001. The European linguistic area: Standard Average European. In: Martin
Haspelmath & Ekkehard Knig & Wulf Oesterreicher & Wolfgang Raible (eds.). Language Typology
and Language Universals (HSK 20). Berlin & New York: de Gruyter. 1492-1510.
Heine, Bernd & Tania Kuteva. 2006. The Changing Languages of Europe. Oxford & New York:
Oxford University Press.
Kortmann, Bernd. 2009. Die Rolle von (Nicht-Standard-)Varietten in der europischen (Areal-)
Typologie. In: Uwe Hinrichs & Norbert Reiter & Siegfried Tornow (eds.). Eurolinguistik:
Entwicklung und Perspektiven. Wiesbaden: Harrassowitz. 165-187.
Seiler, Guido. 2011. Non-Standard Average European. Call for papers for a workshop held at FRIAS,
Freiburg, Feb. 2012.

The Diusion of Diatones and Frequency Eects

Ryuichi HOTTA (Chuo University)
chariderryu@gmail.com

This study aims to describe the historical growth of the diatonic stress pattern (e.g.
vs

recrd

rcord

[n.]

[v.]) from its rst attestations in the late sixteenth-century to the latest additions

in the earliest twenty-rst century. In describing the development, I pay particular attention
to two questions: how the pattern has diused over time and whether the diusion has been
aected by any frequency eects.
I address the rst question with reference to the theory of Lexical Diusion, proposed rst
by Wang, after Sherman's and Phillips's studies, but I attempt at a fresh evaluation of the
theory based on the evidence of the latest added diatones.

The second question concerns

a reconsideration of possible frequency eects that may have aected the order and schedule
of the diusion. Phillips considered word frequencies according to the prex they have and
concluded that less frequent noun-verb pairs turned diatonic earlier. Nevertheless, my enquiry
in consideration of word frequencies by themselves, without regard to their prex, reveals the
opposite: more frequent noun-verb pairs turned diatonic earlier.
In this study I wish to furnish a discussion about how the diusion of diatones can be
characterised in terms of order and schedule and what kinds of frequency eects have been
involved, if at all, in the language change.

References
Phillips, Betty S. (1984) Word Frequency and the Actuation of Sound Change, Language 60, 32042.
Sherman, D.

(1975)

Noun-Verb Stress Alternation: An Example of the Lexical Diusion of Sound

Change in English, Linguistics 159, 4371.

Wang, William S-Y.

(1977 [1969])

Competing Changes as a Cause of Residue, Language 45, 925.

(Rpt. in Readings in Historical Phonology: Chapters in the Theory of Sound Change, ed. Philip
Baldi and Ronald N. Werth, Pennsylvania State University Press, Pennsylvania, 23657).

Emphatic reduplication in Turkish: The multiple origins of linking consonant

Hyung-Soo Kim (Jeonju University, Korea)

In this paper I explore the origins of linking consonant in the emphatic adjectives of
Turkish, beginning with /p/, the de facto default consonant inserted. Vowel initial intensive
adjectives, e.g. eski old, epeski very old, reveal that the linking /p/ may have developed
from CV-reduplication (with subsequent reanalysis) of bases with Proto-Altaic initial *p,
which disappears in most Altaic languages (Ramstedt 1916; Aalto 1955): *pe-peski (CV
reduplication) > epeski (initial loss of *p) > e-p-eski (reanalysis). On the other hand, in the
so-called irregular emphatic reduplications (cf. Lewis 1967), e.g. sapasaglam~sapsaglam
robustly healthy, the same CV-reduplication occurs with an emphatic particle /pA/ whose
unstressed vowel optionally drops, leaving only /p/ to mark emphasis. Interestingly, this /pA/
alternates with /mA/ in languages such as Western Yugur (Roos 2000), suggesting that the
linking /m/ has developed in similar fashion. As for the linking /r/, the fact that its examples
are limited in number and of foreign origin suggests its exogenous beginning, while examples
such as srlsklam ~ srsklam sopping wet are compounds of two synonymic/alliterative
stems (cf. Erdal 1998), with the first stem reduced to CVC. Finally, the dissimilative origin of
linking /s/ is advanced on the fact that sibilant bases generally link with /p/ or /m/, while
examples with the linking /s/ mostly have a labial consonant in the base.
Overall, I propose the following as origins of emphatic adjectives: CV-reduplication
with subsequent reanalysis (as part of the grammaticalization routine: cf. Hopper & Traugott
1993); CV-reduplication with subsequent syncopation of the unstressed enclitic /pA/ and
/mA/ (cf. Menges 1968); reduction of reduplicative or synonymic/alliterative compounds and
dissimilation. Reduplication naturally adds emphasis to the base meaning, as does
compounding of synonymic bases. But its interaction with other morpho-phonological
processes may render such transparent synchronic encoding opaque, leaving in the process
many residues, as we see in the Turkish emphatic adjectives.

Economy in grammar: the emergence of inverse marking patterns in inflectional systems

Ivn Igartua
University of the Basque Country (UPV/EHU)
ivan.igartua@ehu.es
Economy manifests itself in several aspects of grammar or, to put it another way, grammatical
structures can be (partially) characterized by means of the principles of economy: grammar is not
(only) economical, but there is economy in grammar. Some of those principles have been primarily
applied to the synchronic description of languages (as the notion of paradigm economy developed by
A. Carstairs 1984, 1987 and others), but they can also be identified in the evolution of language
structures.
For instance, in inflectional systems economy of forms plays a diachronic role whenever a
syncretic fusional pattern arises and, as a consequence, a single inflected form fits more than one
morphosyntactic description. Another phenomenon apparently linked to morphological economy is
the emergence of inverse inflectional markers in number and gender structures. Number toggling or
inverse number is attested in the small Kiowa-Tanoan group of Amerindian languages. Other types
of morphological reversal characterize, among others, Estonian, the Labrador dialect of Eskimo, and
Luo, a Nilotic language (Smith 1979, Baerman 2007). A somewhat more restricted kind of marker
inversion (with semantic implications) affects number in the Sorbian languages (Slavonic), where
plural morphemes are used in paral reference, while the dual has its own suffixes (the morphological
distinction between paral and dual meanings can be illustrated with examples from Tocharian).
The traditional term genus alternans may also refer to a kind of morphosyntactic toggling. A
group of inflectional or agreement forms is shared by certain genders in the singular (masculine and
neuter vs. feminine), whereas in the plural the genders that share inflectional or agreement forms are
different from the former (feminine and neuter vs. masculine; Corbett 2009 talks about nonautonomous gender values). This is precisely what we find in such languages as Romanian,
Tocharian A and B, and Slovak (in the last case, from the sixteenth century onwards; cf. Igartua
2006). As a result of certain innovations, a third gender or agreement class arises which is essentially
a (rather economical) combination of the marker sets characterizing the other two genders or
agreement classes.
Number and gender toggling, alongside other morphological reversals that deviate from
canonical inflection (Corbett 2011), have deep implications for the principles on which inflectional
morphology is claimed to be based. Marker inversion can hardly be derived from such structural
properties as morphosemantic transparency, constructional iconicity, biuniqueness or uniform
encoding. In this sense, the principle of economy that underlies the emergence of inverse marking
patterns represents a sort of challenge to morphological theory.
References:
Baerman, Matthew, 2007, Morphological reversals, Journal of Linguistics 43:1, 33-61.
Carstairs, Andrew, 1984, Outlines of a constraint on syncretism, Folia Linguistica 18: 1-2, 73-85.
Carstairs, Andrew, 1987, Allomorphy in Inflexion, London: Croom Helm.
Corbett, Greville G., 2009, Universals and features, in Sergio Scalise, Elisabetta Magni & Antonietta
Bisetto (eds.), Universals of Language Today, Dordrecht: Springer, 129-143.
Corbett, Greville G., 2011, Higher order exceptionality in inflectional morphology, in Horst J. Simon &
Heike Weise (eds.), Expecting the Unexpected: Exceptions in Grammar, Berlin / New York: Walter
de Gruyter, 107-126.
Igartua, Ivn, 2006, Genus alternans in Indo-European, Indogermanische Forschungen 111, 56-70.
Smith, Lawrence R., 1979, Labrador Inuttut inverted number marking, exchange rules and
morphological markedness, Linguistics 17, 153-167.

Historicity of the Cumbrian Dialect of English: Dialectizing Process through Language Contact
Tsukusu Jinn It (Faculty of Arts, Shinshu University, Japan)
This paper argues the significance of the investigation of the linguistic substrata into the Place-Names and
the colloquial (documented) English dialect in Cumbria.

The Cumbrian dialect of English cannot simply

be regarded as a result of straightforward language contact but of the complex dialectizing process of the
languages in the area.

In such a process, there may have been a stage which must have provided the

language intermediately with the features of both English and medieval Norwegian.

The process can only

be specified by both a theory of dialects in contact (Trudgill 1986) and an interleaved Optimality Theory
(Bermdez-Otero and Hogg 2003).
Cumbria had experienced concentrated language contact among the British Cumbrian natives, the
Scottish Celts, the Northern dialect speakers of Old English and the settlers of Norwegian vikings with
sporadic Danish new-comers between early 10th and 11th centuries (Fellows-Jensen 1-5).

Studies of

language contact between Old English and Scandinavian such as Thomason and Kauffman (1988) or those
compiled in Ureland and Broderick (1991) rarely mention the situation around Cumbria, while the
situations in more Anglicized area of Danelaw have collectively been examined by, among others, Hines,
Townend, Pons-Sans, Bator.
Cumbria reveals the underlying process of dialectizing its language into English.

Some of the later

place-name records show an oscillating state of the sound: ModE Scalderscew was recorded, for instance,
as Scalderscogh in 1243, Skelderischoth in1287 and Skaldersko in 1303.

The wavering spellings of scogh

> schoth > sko may be reflecting the complex language contact in which phonologization endured a
trial-and-error process.

While this phonologization process can be explained by the interleaved Optimality

Theory, the phonetics of medieval Norwegian may well shed light on the process, despite the scanty data on
the aspect of grammaticalization of the Cumbrian dialect in this period.
References:
Bator, Magdalena 2010.

Obsolete Scandinavian Loanwords in English.

Bermdez-Otero, Ricardo and Hogg, Richard M. 2003

Frankfurt am Main: Peter Lang.

The Actuation Problem in Optimality Theory.

In Holt, D. Eric, ed. Optimality Theory and Language Change.

Dordrecht: Kluwer Academic P,

91-119.
Fellows-Jensen, Gillian 1985.

Scandinavian Settlement Names in the North-West Copenhagen: C. A.

Reitzels.
Hines, John 1991.
George, eds.

Scandinavian English: A Creole in Context.

Language Contact in the British Isles.

Pons Sanz, Sara Mara 2000.

Lindisfarne Gospels.

In Ureland, P. Sture and Broderick

Tbingen: Max Niemeyer, 1991, 403-27.

Analysis of the Scandinavian Loanwords in the Aldredian Glosses to the

Studies in English Language and Linguistics Monographs Vol. 9.

Valencia:

Universitat de Valncia.
Thomason, Sarah Grey and Kaufman, Terrence 1988.
Linguistics.

Language Contact, Creolization, and Genetic

Berkeley: U of California P.

Townend, Matthew

2002.

Language and History in Viking Age England: Linguistic Relations between

Speakers of Old Norse and Old English.

Trudgill, Peter 1986.

Dialects in Contact.

Turnhout: Brepols.
Oxford: Basil Blackwell.

Protolanguage or protolanguages?

Tore Janson, Stockholm University, Sweden

The word protolanguage has been used for two kinds of entities in the last decades.
There are really two different but homonymous terms, which may be denoted
protolanguage1 and protolanguage2. Within historical linguistics, a protolanguage1
traditionally refers to a reconstructed parent language such as Proto-Indo-European or
Proto-Bantu. In evolutionary linguistics, protolanguage2 has been used since the 1970s
to refer to a system of communication used during a stage in the evolution of man, more
advanced than the communication of other primates but less advanced than present-day
human linguistic communication. A reasonable question is whether the methods and
results of historical linguistics can contribute in any way to the study of protolanguage2.
Most historical linguists would think not. It is a well-known empirical fact that our
work cannot take us further back in time than about 10,000 years at the very most.
Attempts for reconstructions of properties of any protolanguage1 further back (such as
the notorious Proto-World) are based on wishful thinking rather than on reliable
methods. And protolanguage2 must be at a distance of at least 50,000 years or so,
perhaps very much more. Still, there may be something to say.
Protolanguage2 designates a system of language used during a certain period. There
are various views on what protolanguage2 may have been like; for overviews, see Fitch
(2010), McMahon and McMahon (2013). However, all scholars seem to agree that the
phonological system was not unlike present-day languages.
Now, the insights of historical linguistics are not limited to expertise about specific
changes and protolanguages1. Within the field, including the sister disciplines
dialectology and sociolinguistics, there is a large body of knowledge about how and why
languages vary, change, split, and influence each other, especially when it comes to the
sound systems. This knowledge may be applicable to protolanguage2.
For one may ask whether forms of protolanguage2 spoken in different places and at
different times varied and changed in the same way as languages do now, or varied less
or not at all. That is, was protolanguage2 the first and original protolanguage1?
It is not possible to know that for certain, of course; almost nothing is, when it comes
to protolanguage2. But if it did not vary and change as languages do now, that must be
because some fundamental property of sound systems changed between protolanguage2
and the very earliest protolanguages1. It is not easy to see what property that could have
been.
If change prevailed throughout the period of protolanguage2, there never was a first,
original language; the period of protolanguage2 then encompassed any number of (not
very advanced) languages and protolanguages1.
The possibility, or necessity, of assuming variation and change has not been discussed
much by those interested in protolanguage2. However, the issue should be put on the
agenda, in my opinion, and historical linguists could make a significant contribution.


References:
Fitch, W. Tecumseh. (2010). The Evolution of Language, Cambridge: Cambridge UP.
McMahon, April M. S. and Robert McMahon. (2013). Evolutionary Linguistics,
Cambridge: Cambridge UP.

Subjectivity in the Dutch mental state predicates.

Karolien Janssens and Jan Nuyts
University of Antwerp
Topic: This paper deals with the question how the subjective meaning of the mental state predicates (MSP) has
emerged historically, and what this tells us about the status of this semantic notion. We define subjectivity as in Nuyts
(2012): an evaluation is subjective if it is presented as being the assessors sole responsibility (as opposed to an
intersubjective evaluation, which is presented as being shared between the assessor and a wider group of people).
That MSPs such as think or believe can express subjectivity is generally accepted in the literature (e.g. Benveniste 1966,
Persson 1993, Aijmer 1997, Simon-Vandenbergen 1998, Van Bogaert 2006, 2009). An example from Persson (1993: 8) is
(1).
(1) Look at that girl. I think she is pretty.
Most authors relate subjectivity (defined in different ways though) to notions such as epistemic and deontic modality
(e.g. Lyons 1977, Coates 1983, Traugott 1989, Traugott and Dasher 2002, Palmer 2009). Also in the literature on the
MSPs, the subjective use is typically correlated with the epistemic use cf. the fact that think, e.g., is also a frequent
marker of epistemic modality, as in (2).
(2) A: Where is John? B: I think hes in the library, but you need to check.
Yet, unlike in (2), think in (1) does not appear to have an epistemic meaning: the speaker is not estimating the chances
that she is pretty. It is rather a pure marker of subjectivity here (in the above definition): it indicates that the aesthetic
evaluation (pretty) is the speakers personal opinion. This raises the question whether there is indeed a relationship
between the pure subjectivity use of the MSPs of the kind in (1) and their epistemic use of the kind in (2). We
investigate this issue by means of a diachronic corpus study of four MSPs in Dutch which can express subjectivity:
denken think, dunken think (impersonal), geloven believe and vinden find.
Method: We have analyzed the meanings of these predicates, and the historical evolution in them, in corpora from four
stages of the language: Old Dutch, Middle Dutch, Early New Dutch and Present Day Dutch. We have collected 200
instances for each period, for each verb (according to criteria such as representativity, e.g. in terms of text genres, and
comparability across the periods). For Present Day Dutch we have selected two different samples, one written and one
spoken.
Results: The data show that subjectivity is probably the youngest meaning category in all verbs, although it is not always
a recent meaning (dunken is already used as a subjectivity marker in Early Middle Dutch), and although it is not equally
strongly present in all these MSPs. But more importantly, the data show that the use as a subjectivity marker may but
need not emerge out of an epistemic use of these verbs: in some of the verbs it does, but in others it does not. Our
investigation thus offers support for Nuyts (2012) assumption that subjectivity constitutes a meaning category on its
own, separate from epistemic modality.
Aijmer, Karin. 1997 I think an English modal particle. In Modality in Germanic Languages, T. Swan and O. Westvik
(eds), 1-48. Berlin: De Gruyter.
Benveniste, Emile. 1971. Problems in General Linguistics. Miami: University of Miami Press.
Coates Jennifer. 1983. The semantics of modal auxiliaries. London & Canberra: Croom Helm.
Lyons, John. 1977. Semantics. Cambridge: Cambridge University Press.
Nuyts, Jan. 2012. Notions of (inter)subjectivity. English Text Construction 5: 53-76.
Palmer, Frank R. 2009. Mood and modality. Cambridge: Cambridge University Press.
Persson, Gunnar. 1993 Think in a panchronic perspective. Studia Neophilologica 65: 2-18.
Simon-Vandenbergen, Anne-Marie. 1998 I think and its Dutch equivalents in parliamentary debates. In Corpora and
Crosslinguistic Research, S. Johansson and S. Oksefjell (eds), 297-317. Amsterdam: Rodopi.
Traugott, Elizabeth C. 1989. On the rise of epistemic meanings in English. Language 65: 3155.
Traugott, Elizabeth C. and Richard B. Dasher. 2002. Regularity in semantic change. Cambridge: Cambridge University
Press.
Van Bogeart, Julie. 2006. I guess, I suppose and I believe as pragmatic markers: grammaticalization and functions. BELL
New Series 4: 129-149.
Van Bogaert, Julie. 2009. The Grammar of Complement-Taking Mental Predicate Constructions in Present-Day Spoken
British English. PhD thesis, University of Gent.

Regrammation and paradigmatization

Reanalyses of the deictic relative construction with progressive function in French

Kirsten Jeppesen Kragh & Lene Schsler, University of Copenhagen

During the history of French, progressivity has been expressed in different ways, both by means of
simple tenses (atelic tenses like the present and the imperfect tenses) and by means of analytic
constructions. The first known progressive construction is a combination of an auxiliary (a form of
to be or of a verb of movement) and a present participle: Pierre est / va / s'en va / vient / s'en vient
chantant, meaning Peter is singing, in use from the early texts until the 18th century. Later, from the
16th century, other constructions with progressive function arise: first a form of to be, a preposition,
and the infinitive: Pierre est / aprs chanter, still later, from the 18th century, a more complex
construction: Pierre est en train de chanter, see Schsler (2006). Recently in Kragh & Schsler (to
appear), we have proposed an analysis of the deictic relative construction as yet another way of
expressing progressivity: (Je vois) Pierre qui chante.
The purpose of the present paper is to characterize each of these constructions as members of a
progressive paradigm. Our use of the term paradigm conforms to that found in (Nrgrd-Srensen
et al. 2011), according to which paradigmatic structure is common to morphology, topology (word
order) and constructional syntax and all grammatical changes involve paradigmatic restructuring.
We aim at identifying differences and similarities of these progressive constructions, their
distribution over time, place and text types.
The method of investigation is corpus based, by means of data mainly found in Frantext, the largest
collection of texts from the earliest period till today with a large variety of text types
(http://www.frantext.fr).
References:
Kragh, Kirsten Jeppesen & Schsler, Lene. to appear. Reanalysis and grammaticalization of
constructions.
Nrgrd-Srensen, Jens, Heltoft, Lars & Schsler, Lene 2011. Connecting grammaticalization. The
role of paradigmatic structure. Amsterdam/Philadelphia: John Benjamins Publishing
Company.
Schsler, Lene. 2006. Grammaticalisation et dgrammaticalisation. Etude des constructions
progressives en franais du type Pierre va / vient / est chantant, in Labeau, Emmanuelle,
Carl Vetters & Patrick Caudal (eds), Smantique et diachronie du systme verbal franais
Vol. 16. Amsterdam/New York: Rodopi, 91-119.

Eroding Ergativity: Revealing Structure?

Alana Johns, University of Toronto
Inuktitut is known for being an ergative language, as shown in (1), where the agent of the
transitive construction in (1a) has a different case (ergative) from that of the theme of the
intransitive construction (absolutive) in (1b), which not coincidentally is the same case as that of
the patient in the transitive construction (1a).
(1) a. anguti-up nanuq kapi-jaa
[Transitive construction]
man-erg. polar.bear(abs) stab-tr.part.3s/3s
The man stabbed the polar bear.
b. angut pisuk-tuq
[Intransitive construction]
man(abs) walk-intr.part.3s
The man is walking.
What is less known is that Eastern Canadian dialects have become more nominative-accusative, in
that a transitive clause typically utilizes the antipassive construction, (Johns 2001), as in (2) which
functions as an alternative transitive construction and has less agreement.
(2)

angut nanur-mik kapi-si-juq

[Antipassive construction]
man(abs) polar.bear-MIK stab-AP-.intr.part.3s
The man is stabbing a/the polar bear.

It is therefore interesting that with this erosion of canonical ergativity in Eastern Inuktitut, we
observe a patterning that resembles one found in Unangax or Aleut (Bergsland 1997), which has
been described as anaphoric (Fortescue 1995). The revealed patterning entails that the ergative
construction in (1a) is more likely to be found in second mention, referring back to a previously
mentioned nominal. That two varieties of the same language family (Eskimo-Aleut) should
demonstrate a similar pattern, even though they are geographically very distant, and are only
distantly related suggests that this pattern is a deep property of the language family. If so, why
then do intervening languages and dialects not show the same property? I will argue that in fact
they do, if we consider the canonical ergative construction as involving clitic doubling in the
verbal inflection (which is linked to two arguments). Unangax and Eastern Inuktitut have come to
disallow clitic doubling, thus necessitating the non-use of the ergative construction where a
nominal is introduced (non-clitic doubling). This provides a simple explanation for the related fact
that the ergative construction functions as anaphoric agreement, or a clitic (Merchant 2011) when
an argument has already been introduced.

Bergsland, K. 1997. Aleut Grammar: Unangam Tunuganaan Achixaasix. Fairbanks: Alaska Native
Language Center.
Fortescue, Michael. 1985. Anaphoric Agreeement in Aleut. In Predicates and Terms, eds. A. M.
Bolkestein, C. de Groot and J. L. Mackenzie, 105-126. Functional Grammar Series 2.
Dordrecht: Foris.
Johns, A. 2001. Ergative to accusative: comparing evidence from Inuktitut. In J. T. Faarlund (ed),
Grammatical Relations in Change, 205-221. Amsterdam: John Benjamins.
Merchant, Jason. 2011. Aleut case matters. Pragmatics and Autolexical Grammar: In honor of Jerry
Sadock, E. Yuasa, T. Bagchi and K. Beals (eds), 193-210. Amsterdam: Benjamins.

A methodology for quantifying intuitionistic claims about morphological change

Cynthia A Johnson, Rory Turnbull, Rachel Steindel Burdin, & David M Howcroft
Morphological changes have been described as processes that promote clarity and/or
redundancy (e.g. Hock 1991:446, Joseph 1998:365). These notions are invoked on an intuitionistic
basis and are not defined or quantified. We argue that information theory can be used to quantify and
assess these claims. Information theory is a mathematical framework for quantifying the amount of
information conveyed by a message given its context (Shannon 1948). Metrics from information theory,
such as entropy, provide a basis for quantifying and evaluating such claims about morphological change,
enabling us to make comparisons across paradigmatic changes that proceed in opposite directions (e.g.
grammaticalization vs. lexicalization).
We examine the development of the adjectival system from the bipartite strongweak system of
Middle High German (MHG) to the tripartite strongmixedweak system of New High German (NHG). The
choice of a strong or a weak adjective ending depends on the form of the preceding article: in MHG, the
weak declension is used following definite articles while the strong is used elsewhere. In NHG, the weak
is still used with definite articles, but the new mixed class is used following indefinites, and the strong
continues to be used elsewhere.
The information theoretic notion of entropy is a measure of the overall unpredictability in a given
communicative system. We calculate the expected relative entropy of the adjective form given the form
of the preceding word in order to compare the predictability of the adjective form in general in MHG to
its predictability in NHG. We find that the shift to a tripartite system has resulted in an increase in this
predictability.
Because morphological change can proceed in different directions (e.g. the strongweak
distinction has been lost in Afrikaans), we contrast the actual state of affairs with other plausible
developments from MHG to NHG. We call two of these alternatives NHGw and NHGs. In NHGw, the
bipartite system of MHG collapses to a single system based on the weak declension, while in NHGs, the
MHG system collapses to a single system based on the strong declension. We use expected relative
entropy to compare NHGw and NHGs with MHG, demonstrating that entropy can also increase locally as
a result of other plausible paradigmatic changes.
We present a methodology for applying information theoretic notions to morphological change,
utilizing a framework which has already been applied with some success to the description of synchronic
variation (Jaeger 2010), as well as diachronic variation in phonology (Tsui 2012). The successful
application of the methodology here not only provides a common underlying motivation behind different
types of morphological change, but also provides a potential unifying factor to change at all levels of the
grammar, as the same notions can be used to describe phonological, syntactic, and morphological
variation and change.

References
Hock, H. H. 1991. Principles of Historical Linguistics, Berlin: Mouton de Gruyter.
Jaeger, T. F. 2010. Redundancy and reduction: Speakers manage syntactic information density. Cognitive
Psychology, 61:2362.
Joseph, B. 1998. Historical Morphology. In A. Zwicky & A. Spencer (eds.), The Handbook of
Morphology, Blackwell Publishers.
Shannon, C. E. 1948. A mathematical theory of communication. Bell System Technical Journal,
27:379423, 623656.
Tsui, TH. 2012. Tonal variations in Hong Kong Cantonese: Interactions of acoustic distance and
functional load. Presented at
CLS48, Chicago IL.

LocalanalogyasasourceofsuppletioninCatalan
MatthewL.Juge,TexasStateUniversitySanMarcos
Suppletionresearchhasnotpaidadequateattentiontoanalogyasasourceofsuppletion(Juge1999,
forthcoming).Previousresearchhasalsofailedtorecognizetheroleoflocalchange(asdescribed
byJoseph1992).InthispaperIexamineinteractionsamongsoundchange,analogy,andthe
developmentofsuppletioninCatalanverbsinsupportoftheclaimthatanalogicalchangesofthistype
arenotteleologicalbutrathertheundirectedresultofpatternsoflexicalstorageandsemanticand
formalassociations.AseriesofregularsoundchangesinthehistoryofCatalancreatedconditionsthat
ledtosuppletivepatternsnotfoundintheotherRomancelanguagesoftheIberianPeninsula.The
mostimportantofthesechangesinvolvethelossofnonlowfinalvowels(Latindc>CatalandicI
say)andeventualvocalizationofthevelarbeforefrontvowels(Latindcis>Catalandiusyousay).
Suchchangesestablishedthefinalvelar/k/asamarkerofthefirstpersonsingularinthepresent
indicativethathasspreadtootherformssuchasscIam.Becausesomeperfectformsalso
developedfinalvelars(e.g.,Latinhabuit>OldCatalanhac,3spreteritofhavertohave),the
statusofthesevelarsisnotentirelyclear:theycanbeconsideredinflectionalendingsorstemfinal
consonants(inalternationwithvoicedsegmentsreminiscentofpatternsfoundelsewhereinthe
language).Thestatusofthefinalvelaralsorelatestothedialectalvariationinthefirstpersonpresent
indicativemarker,sincesomedialectshavenosuffix(e.g.,AlgueresecantIsing),whileothershave
vocalicsuffixes(e.g.,oinStandardCatalan,iinNorthernCatalan).
Whilethespreadofchasincreasedthenumberofverbswiththiselement,ithasnot
regularizedtheverbsinquestionandthusrunscountertothewidelyheldviewthatanalogycreates
regularity.Mostofthesecasesarebetteranalyzedascontaminationthanfourpartanalogy,bothof
whichIclassunderanalogymorebroadly(followingHock2005)sincetheybothcreateirregularity
(cf.dialectalGalicianiaIwent(fourpartanalogy)andfeaImake(contamination)(Juge1999,
forthcoming)).Aski(1995)hasproposedatemplatebasedapproachtosuppletiveverbsin
Romance,butheranalysisdoesnotfitthedata.IarguethattheCatalandatasupporttheviewthat
suchchangesdependonsubconsciousassociationsofformalandsemanticpatternsratherthan
systemicconcernsperse.Thesedevelopmentsillustratethepotentialformultiplefactorstoinfluence
morphologicalchangeandchallenge
traditionalviewsofanalogicalchange.Theyalsofavorthehypothesisthatanalogicalchangedepends
onpatternssometimeslocallydefinedoflexicalstorageandisnotteleological.
References
Aski,JaniceM.1995.Verbalsuppletion:AnanalysisofItalian,FrenchandSpanishtogo.Linguistics
33.403432.
Hock,HansHeinrich.2005.Analogicalchange.InRichardJanda&BrianD.Joseph(eds.),Handbook
ofHistoricalLinguistics,441460.Malden,MA:Blackwell.
Joseph,BrianD.1992.Diachronicexplanation:Puttingspeakersbackintothepicture.InGarryW.
Davis&GregoryK.Iverson(eds.),Explanationinhistoricallinguistics,123144.Amsterdam:John
Benjamins.
Juge,MatthewL.1999.Ontheriseofsuppletioninverbalparadigms.InChang,SteveS.,LilyLiaw&
JosefRuppenhofer,(eds.),ProceedingsoftheTwentyFifthAnnualMeetingoftheBerkeley
LinguisticsSociety,183194.Berkeley:BerkeleyLinguisticsSociety.
Juge,MatthewL.forthcoming.Analogyasasourceofsuppletion.InLaurenceReid&Ritsuko
Kikusawa(eds.),HistoricalLinguistics2011:SelectedPapersfromthe20thInternational
ConferenceonHistoricalLinguistics,Osaka,August2011.Amsterdam:JohnBenjamins.

OntheRiseofNonCMintinHungarian:ACorpusBasedDiachronicApproach.
GergelyKntor
ResearchInstituteforLinguistics,HungarianAcademyofSciences
1. Introduction. Hungarian has a construction not wellattested in formal linguistics. After describing the
phenomenon, I turn to data found in corpora from various stages of Hungarian, which will answer how, when and
whythepresentsituationevolved.
2. The phenomenon. Mint1 (than) in presentday Hungarian (PDH) is a C0 that introduces finite, comparative
subclauses(Kenesei1992).Still,mintisnonclausalinanotherconstruction:
(1) a.t mint ldozatot hallgattkki.b.
Jnossal
mint gyanstottal beszltek.
him as
victimACC they.listenedVMJohnCOM
as
accusedCOM they.talked
Hewasinterrogatedasthevictim. TheytalkedtoJohnastheaccused.
In line with Bnrti (2007), mint2 in (1) is a binary conjunctiongeneratingordinarybalancedcoordination(parataxis),
in which the same features are realised on both conjuncts. It is not a preposition, as Hungarian has none. It isnota
C0, because mint ldozatot receives main stress in (1a), followedbythereverseorderof theverb andverbmodifier
(VM), indicating that the mintphrase is focussed (Hungarian CPs cannot be focussed .Kiss 2002). Mint2 in (1)
must be followed by one DP only, and the same morphological case must be realised on the DPs preceding and
following mint, as determined by the predicate. Also, binary conjunctions encode conventionalimplicatures(e.g., but
in she is a top movie star, but she is humble clash between being a top movie star and being humble cf. Grice
1975).Paratacticmint2encodestwoconventionalimplicatures:
(A) the first conjunctDP canbe predicatedover by thesecond one(e.g., in(1a),myfatherisavictim,regardlessof
beinginterrogatedornot),and
(B) the secondconjunctrefersto astate/characteristicofthe individual inthe first conjunct,whichisa cause/basisof
the event defined by the proposition (they interrogated my father because he was the victim in (1a), they
talkedtoJohnbecausehewassuspectedofbeingguiltyin(1b)).
3. The corpora. The corpora used are mainly Bible translations from various stages (Old Hungarian (OH
8961526), Early Middle Hungarian (EMH 15261590) and Late Middle Hungarian (LMH 15901772)). Biblical
texts include many similes and metaphors metaphors may represent both conventional implicatures typical of
paratactic mint (A+B), while similes only (B). The research investigates the historical synonyms of mint, either
paratacticorhypotactic,whoseratiosandchangescanhelptounderstandhowparatacticmint2evolved.
4.Languagechange.OldHungarian(OH)hadfourelementsthatwereabletointroducesimiles:
(2) a. sem feneseitnk, mint nap(ViennaCodex,Baruch6:66)
mint
not they.light
as
Sun
theycannotshineliketheSun
b. a vizek megerosodnenc
monnal kofal(ViennaC.,Jud5:12) monnal
the waters they.would.harden as
stone.wall
thewaterswouldstandfirmasawalloneitherside
c. befedec
foldnc orcaiat mikent saskac(ViennaC.,Jud2:11)
miknt
they.would.cover Earths face as
locusts
theycoveredthefaceoftheEarthlikelocusts
d. eltauoztattnyaz penzt
mykeppen ewrdewgewt(JkaiC.79:24) mikppen
VMsendthe moneyACC
as
devilACC
togetridofthemoneyliketheDevil(or:asifitweretheDevil)
As whoperators, they appeared in full, overt subclauses too (see (3a)) the underlying adverbial nature of these
subclauses is supported by the optional appearance of the resumptive demonstrative pronoun of manner adverbials
(PDHgy,OHvg(an))inthemainclause(see(3b)):
(3) a. bankodnacorajtamintzoctacbankodnielozolutnechalalan(ViennaC.Zach.12:10)
they.mournhimSUPasthey.domournfirstbornsdeathSUP
theyshallmournforhim,asonemournsforanonlychild
b. medeneketvgantezenvalamintazvdobenzorgalmazikuala(ViennaC.Est.2:20)
allACCsoDEMhe.didV.AUXasthetimeINEhe.urgedV.AUX

shedideverythingthewayheurgedhertodo
Corpus data reveal a tendency: if there is no resumptive pronoun in the main clause, the subclause is elliptical with
only one nominal expression overt (true for 76% of the cases), as in (2). Still, OH already had some elliptical
examples in which the bare nominal after theoperatorcould be takenas predicative,predicating overaconstituent in
themainclause(conventionalimplicature(A)above):
(4) micoralkototuolnamnalostortazkotelecbol(MunichCodex86rb,John2:15)
whenhe.madeaswhipACCthecordsELA
whenhemadesomethinglikeawhipofcords
By EMH, (i) mint had undergone a relative cycle (van Gelderen 2009) and had become a C0 head
(BcskaiAtkri 2011) (ii) monnal had become extinct (iii) miknt and mikppen had started to withdraw from
elliptical clauses: in the corpus used, less than 7% of miknt and mikppen constructions involve only a nominal
expression all other examples are clausal. However, as for mint, the tendency that the lack of the main clause
resumptive came with a bare nominal after mint was stronger than in OH (true for over 90%). Also, the subset of
elliptical mintconstructions representing conventional implicature (A) had become statistically muchmoresignificant
thaninOH(over10%ofallellipticalmintconstructions).
By LMH, all but one characteristics were given for paratactic mint2to break away from its C0 parent: (i) mint
was already a head (ii) in a subset of mint constructions,itcould encodeboth implicatures(A+B),notonly(B)asa
C0 (iii) this subset is alwayselliptical withonlya nominalexpressionafter mint(iv)this nominalexpressionisalways
predicative. The last step on mint2s evolution into a binary conjunction was category change: C0 Conj0. (This
happened only to mint2: monnal became extinct miknt and mikppenremainedoperatorsappearingin subordinate
clauses.) I claim that this change had occurred by LMH: paratactic mint2 is even more statistically significant now
(e.g., over 15% of all elliptical mint constructions in the Kroli Bible), and this is the first time a mintconstruction
appearedinafocussedposition:
(5) tet
mint rckvalt fogadnad
be(KroliBible,1590,Philemon1:15)
himACC as
eternalACC
you.may.accept VM
youmayhavehimassomeoneeternal(youmayhavehimforever)
In (5), rckvalt is followed by a reverse VVM order, as in (1a). The reason for this reanalysis is Lightfoots
(1979) Transparency Principle: in a special subset, a C0 followed by obligatorily elliptical subclauses, encoding an
extraconventionalimplicaturewasmoreopaquethananewcategory.
Selectedreferences:
BcskaiAtkri, J. (2011) A komparatv opertor esete a mondatbevezetvel. [The Comparative Operators Case
with the Complementiser.] In: K. . Kissand A.Hegeds eds.NyelvelmletsDiakrnia.BudapestPiliscsaba:
PPKE.pp.103119.
Bnrti,Z.(2007)Amellrendelss azellipszisnyelvtana amagyarban [TheGrammar ofConjunctionandEllipsisin
Hungarian.]Budapest:Tinta.
.Kiss,K.(2002)TheSyntaxofHungarian.Oxford:OxfordUniversityPress.
Grice, P.H. (1975) Logic and Conversation. In: P. Cole and J.L. Morganeds.SyntaxandSemantics3.New York:
AcademicPress.pp.4158.
Kenesei,I.(1992)OnHungarianComplementizers.ApproachestoHungarian4:3750.
Lightfoot,D.W.(1979)PrinciplesofDiachronicSyntax.Cambridge,CUP.
van Gelderen, Elly (2009) Renewal in theLeftPeriphery: Economyand theComplementiserLayer.Transactionsof
thePhilologicalSociety107.2:131195.
ThisresearchisfundedbyOTKA78074.

The grammaticalization of the English progressive: cross-linguistic perspectives

Killie Killie
University of Agder, Norway
kristin.killie@uia.no

In Killie (forthc.) I suggest that the English be + -ing periphrasis has undergone the following
development:1
(1)

(A) EModE
(B) EModE19th century
(C) 19th century

FOCUS MARKER, used with any predicate regardless of aktionsart

PROGRESSIVE MARKER, emphatic and mostly optional
PROGRESSIVE MARKER, increasingly obligatory with concomitant
loss of emphasis in its obligatory uses

Killie (forthc.) is based on the historical part of the Helsinki Corpus. In this corpus it is
difficult to find proof of a stage B. However, in my paper I will argue that cross-linguistic
evidence may shed light on the Early and (early) Late Modern progressive.
Williams (2002) draws attention to an interesting point concerning the use of the
progressive in Italian. He notes that its use is not fully optional; rather in Italian it is not
possible to use the non-progressive form in certain cases when we specifically wish to stress
the ongoing nature of the situation (2002: 38). Thus, the second sentence in the exchange in
(1) below, which requires a progressive form in English, also requires a progressive in Italian.
1) English: Make me tea, will you? Actually, Im already making you one.
Italian: Mi fai un t, per favore? A dire la verit, te lo sto gi facendo.
In my paper I argue that the Norwegian progressive works in a similar way as the Italian
one. I also provide evidence from Early Modern English, from the Old Bailey Corpus, which
suggest that the Early Modern English progressive may have had a similar function. The
absence of evidence of such a stage in the Helsinki Corpus may thus be explained by a
paucity of situations in that corpus in which immediacy is sufficiently stressed, as in (1)
above.
References
Killie, Kristin. forthc. The development of the English BE + V-ende/V-ing periphrasis: from
focus construction to progressive marker. To appear in English Language and Linguistics.
Williams, Christopher (2002) Non-progressive and progressive aspect in English. Biblioteca
della ricerca, Linguistica 12. Fasano: Schena Editore.

This scenario disregards the development of the periphrasis as a subjective marker, which will not be treated
here.

Null arguments in Old Norwegian in an early Germanic perspective

Kari Kinn, University of Oslo
Although there is a vast literature on null arguments within the framework of generative grammar, the null
argument properties of early Germanic languages have received relatively little attention. A recent exception
that approaches early Germanic null arguments comparatively is Walkden (2012). He takes Old Icelandic data
into account, while its close and contemporary relative Old Norwegian remains unexplored. The aim of this
paper is to supplement and revise our picture of early Germanic null arguments, making use of evidence from
the Old Norwegian part of the Menotec/ISWOC corpus.
Walkden (2012, 216222) develops an analysis in which Old Icelandic, Old English, Old High German and
Old Saxon are considered partial null argument languages. I will argue that this analysis, albeit capturing
important distributional facts, faces empirical problems when confronted with Old Norwegian.
According to Walkdens analysis, the licensing of null arguments (in the early Germanic languages mentioned
above) depends on an Aboutness topic operator in SpecShiftP (Walkden, 2012, 219). This implies that there
should be no more than one null argument in each clause, as there is only one ShiftP (Walkden, 2012, 221).
However, this prediction is not borne out in Old Norwegian, where examples containing null objects in addition
to null subjects are found. Cf. (1) below, where the null subject refers to a group of Wends, while the null
object refers to the tongue of a prisoner:
(1)

pro skaro pro ar af

[they] cut [it] there off
There, they cut it off. (The legendary saga of St. Olaf, 70941)

The view that null arguments are not conditioned (exclusively) by an Aboutness topic operator in SpecShiftP
is supported by the fact that Old Norwegian exhibits sentences with a single null argument where it is highly
debatable whether this argument is an aboutness topic. In (2) below the null object refers to sott disease, but
the clause containing the null object seems to have he purpose of encreasing our knowledge about Egill rather
than the disease (i.e. Egill is the aboutness topic):
(2)

Sva er sact at gil tkr nu sotti

sva hara at igi er inn mir farenn.
En sva bar
so is said that Egill takes now diseasei so hard that not is one more ill-treated but so bore
hann pro i prudlega at ngi mar hyri hann ymia.
he [it]i stately that no man heard him cry
It is said that Egill now became so ill that no man was struck harder than him. But he bore it with
such dignity that no one heard him cry. (The legendary saga of St. Olaf, 22292)

I will discuss a revision of Walkdens analysis that can account for the Old Norwegian facts, but still preserve
the advantages of the partial null argument approach. This will involve extending the range of operators in the
C-domain able to license null arguments. The view that null arguments in Old Norwegian are not necessarily
aboutness topics presses the question of how their antecedents are successfully identified. As a possible approach
to this problem, I will discuss a mechanism of context scanning under control or by extrasyntactic means, in
the spirit of Sigursson (2011, 283).

References
Sigursson, H. (2011). Conditions on argument drop. Linguistic Inquiry, 42(2):267304.
Walkden, G. (2012). Syntactic Reconstruction and Proto-Germanic. PhD thesis, University of Cambridge.

The un-grammaticalization of rather: evidence against grammaticalization

Rachel Klippenstein, The Ohio State University
klippenstein.1@osu.edu
Grammaticalization is often conceptualized as a theoretically significant category of linguistic
changes. The category is typically defined in terms of morphemes coming to express more
grammatical contenteither through lexical morphemes coming to express grammatical
functions, or through already-grammatical morphemes coming to express new grammatical
functions. Arguments for grammaticalization as a theoretically significant category depend on
claims that grammaticalization has distinctive characteristics beyond the definitional property
of generating new grammatical material.
In some varieties of English, a verb meaning roughly prefer has developed out of the
adverb rather, as seen in sentences like When my brother finished high school he wanted to
travel but my parents rathered he would stay at home. (Strong 1926). This development took
place through many incremental smaller changes of reanalysis and analogical extension (Juge
2002, Klippenstein 2012).
I argue that this development is distinguishable from cases of grammaticalization only
by the definitional property: it does not create new grammatical material, but new lexical
material. A main verb meaning roughly prefer is no more grammatical in content than an
adverb (but also no less grammatical; this is not degrammaticalization). In other respects, this
development is exactly like grammaticalization. It takes place through the same mechanisms
of reanalysis and analogy that are widely recognized as those by which grammaticalization
takes place (Traugott 2011, and see Norde 2009:1819 for more references), including
semantic reanalysis via pragmatic inferencing; repetition, which Traugott 2011 also presents
as a core mechanism of grammaticalization, is not at work in all cases of grammaticalization,
so its absence from the case of rather does not distinguish this case from grammaticalization.
In addition, the incremental steps by which it takes place are exactly the kind of incremental
changes (sometimes called gradual) that is often taken to be typical of grammaticalization
and to distinguish grammaticalization from other kinds of change.
This case thus provides evidence that grammaticalization should not be conceptualized
as a theoretical category that is fundamentally different from other kinds of linguistic change,
though it may remain a useful term for describing some linguistic phenomena.

Juge, Matthew. 2002. Unidirectionality in Grammaticalization and Lexical Shift: The Case of
English Rather. Proceedings of the Annual Meeting of the Berkeley Linguistics
Society. 28:147154
Klippenstein, Rachel. 2012 The behavior-before-coding principle in morphosyntactic
change: evidence from verbal rather. Paper presented at the Linguistic Society of
America 86th Annual Meeting, Jan. 5, 2012, Portland, Oregon.
Norde, Muriel. 2009. Degrammaticalization. Oxford: Oxford University Press.
Strong, Mabel. 1926. New Verbs. American Speech 1:292.
Traugott, Elizabeth Closs. 2011. Grammaticalization and Mechanisms of Change. The
Oxford Handbook of Grammaticalization, ed. by Heiko Narrog and Bernd Heine.
1930. Oxford: Oxford University Press.

Contact,shiftandvowellengthinShetlandNorn
REMCOKNOOIHUIZEN
RijksuniversiteitGroningen
JakobJakobsensEtymologicalDictionaryoftheNornLanguageinShetland(190821)isaveritable
treasuretroveforhistoricallinguistics.Basedonfieldworkcarriedoutinthe1890s,thedictionarycontainsc.10,000
lemmaswithdetailedlocalisedphonetictranscriptionsandetymologicalinformationonwordsofScandinavianoriginin
Shetland.AsthedatawascollectedoveracenturyafterthedeathofNorn(Barnes,1998),itisusefulforthestudyof
languagecontactandshiftinducedattritioninahistoricalcontext(e.g.,Campbell&Muntzel,1989Sasse,1992).
Jakobsensphonetictranscriptionspaylittleattentiontoanysystemapparentinthelanguage,butitisinthesystemsthat
wecanfindveryrelevantinformation,especiallywhenthelanguagesincontacthadvery
differentwaysofgoverningaparticularphoneticfeature.Thefeatureinvestigatedinthispaperisvowel
length,wherewemaydistinguishthreerelevantsystems:
Free(unconstrained)vowelandconsonantlength,asinOldNorse.
Scandinaviancombinatoryvowellength,whereeitherthevowelorthefollowingconsonant(s)islong,
butneverbothorneither(rnason,1980).Itisgenerally(implicitly)assumedthatthiswasthesystem
inprelanguageshiftNorn.
TheScottishvowellengthrule,whereasubsetofvowelsislengthenedwhenfollowedbyavoiced
fricative,/r/,oramorphemeboundary(Aitken,1981).Withminormodifications,thisisthesystemin
presentdayShetlandScots(Mather&Speitel,1986vanLeyden,2002).
ThepicturepaintedbytheNorninJakobsensDictionaryisnotasclearasanyoftheseidealisedsystems.
Infact,tracesofallsystemscanbefound,aswellaslocalinnovations:
LongvowelsinNornoftengobacktoOldNorselongvowelsordiphthongs.
Vowelsfollowedbytwoormoreconsonantsaremostoftenshort,alsowhentheywerelonginOld
Norse.
OldNorseshortvowelsmaybelengthenedinNornwhenfollowedby/r/.
Therearesignsofcompensatoryvowellengtheningwhenthefollowingconsonant,often//,has
disappeared.
Inthispaper,IpresentadetailedaccountoftheconstraintsonvowellengthinNorn,basedontheG,H,IandJsections
ofJakobsensDictionary(c.2500tokens).Theconstraintsareanalysedinacontextofcontactandattritioninduced
languagechange.IthendiscussthestatusofthelanguageinJakobsensDictionaryisitNornorScots?and
assesshowthisdatahelpsconnectearlier(prelanguageshiftNorn)andlater(ShetlandScots)stagesofthelanguage
situationinShetland.
REFERENCES
Aitken,A.J.(1981).TheScottishvowellengthrule.InM.Benskin&M.L.Samuels(eds.),Somenypeople,longages
andtonges.Edinburgh:MiddleEnglishDialectProject.131157.
rnason,K.(1980).Quantityinhistoricalphonology:Icelandicandrelatedcases.Cambridge:CUP.
Barnes,M.P.(1998).TheNornlanguageofOrkneyandShetland.Lerwick:ShetlandTimes.
Campbell,L.&Muntzel,M.C.(1989).Thestructuralconsequencesoflanguagedeath.InN.C.Dorian(ed.),
Investigatingobsolescence:studiesinlanguagecontractionanddeath.Cambridge:CUP.181196.
Jakobsen,J.(19081921).EtymologiskordbogoverdetnorrnesprogpaaShetland.Kbenhavn:Prior.
Leyden,K.van(2002).TherelationshipbetweenvowelandconsonantdurationinOrkneyandShetlanddialects.
Phonetica59.119.
Mather,J.Y.&Speitel,H.H.(1986).TheLinguisticAtlasofScotland.VolumeIII:Phonology.London:CroomHelm.
Sasse,H.J.(1992).Languagedecayandcontactinducedchange:similaritiesanddifferences.InM.Brenzinger(ed.),
Languagedeath:factualandtheoreticalexplorationswithspecialreferencetoEastAfrica.Berlin:Moutonde
Gruyter.5980.

Patterns in the diffusion of nomenclature systems: Australian subsections in

comparison to European days of the week
Harold Koch
Australian National University, Canberra, Australia
The languages of north-central Australia share a system of eight social
category terms called subsections (or skins). The terms are tightly structured into
relations of mother-child, father-child, and preferred spouse. It is assumed that the
terminological system has spread from language to language over a large area;
linguists and anthropologists have tried to reconstruct the direction and chronology of
this diffusion (e.g. McConvell 1985, Harvey 2008, McConvell et al. in progress).
Methods include the normal approach to loanwords analysis, including considering
the distribution of forms across languages, their morphological analysis in the
respective languages, principles of phonological adaptation, and phonological
changes before and after borrowing which provide linguistic stratigraphy.
In working out finer details of the spread of forms, I suggest that some lessons
can be learned from the spread of nomenclature systems elsewhere in the world. The
diffusion of the names of days of the week throughout Western Europe across
linguistic and cultural boundaries provides a partial parallel. This system of seven
names differs, however, in that: the terms are ordered in only a single dimension, they
are more meaningful (including day and the name of a deity or celestial body), their
diffusion was aided by an over-arching, literate (Latin) culture .
This paper describes some of the principles that apply to the borrowing of
naming systems and illustrates them with examples from both the European names of
days of the week and Australian social category terms.
1) Terms may be calqued rather than directly borrowed (i.e. be partially translated
into native equivalents
2) Terms, once borrowed, may be joined by alternatives terms, which in time may
replace the original form.
3) Terms replaced in central areas may persist in more peripheral areas.
4) A language may borrow different terms from different languages.
5) Terms may receive explicit marking of their common semantic base (day) or
may drop it (Italian -di vs. Catalan di- vs. no reflex of day in Spanish); for
subsection names the common basis is masculine or feminine gender.
6) Forms within the set may be formally changed due to system-internal analogies.
7) Forms within the set may reflect particular semantic relations between the terms.
The last three principles reflect the fact that nomenclature systems, like
morphological paradigms, consist of a set of terms bearing a common semantic
denominator that are distinguished along certain parameters.
References
Harvey, Mark. 2008. Proto-Mirndi: a discontinuous language family in northern
Australia (Pacific Linguistics 593). Canberra: Pacific Linguistics.
McConvell, Patrick. 1985. The origin of subsections in northern Australia. Oceania
56: 1-33.
Patrick McConvell, Harold Koch, Jane Simpson, and Laurent Dousset. In progress.
Skin and Kin in Aboriginal Australia: linguistic and historical perspectives on
the dynamics of social categories. Research project funded by the Australian
Research Council through the Australian National University, Canberra, in
partnership with CREDO, Marseille, France.

The historical development of sentence-internal capitalization of words in German

Klaus-Michael Kpcke (University of Mnster)
Renata Szczepaniak (University of Hamburg)

The topic of our talk is the development of sentence-internal capitalization of words (SIC) in
German. We report from our recently started project, funded by the Deutsche
Forschungsgemeinschaft (KO 909/12-1 and SZ 280/2-1).
From earlier studies (among others Weber 1958, Kmpfert 1980, Risse 1980, Moulin 1990,
Bergmann/Nerius 1998) we know that the decisive period for the increase in the use of capital
letters falls into Early New High German (ENHG) from the 14th to the 17th century. We
explain this phenomenon by the interaction of cognitive-semantic and syntactic factors. The
goal of the project is to show that the increased use of capital letters was motivated by
cognitive-semantic categories such as animacy and individualization and was sensitive to
semantic roles (degree of agentivity) and syntactic functions (subject, object, etc.). For
example, we hypothesize that a specific animate noun in subject position is more likely
written with a capital letter than the same noun in object function. Although the development
of SIC took place in different areas at different times, we assume that the same cognitivesemantic and syntactic principles were at work, implying that the order of extensional phases
is roughly the same. Hence, our approach does not necessarily rely on the standardizing role
of individual chanceries, but rather foregrounds cognitive factors. We thus interpret the SIC as
a process of a grammaticalization. More specifically, we assume that initially capitalization
has the pragmatic function of emphasizing relevant information. It spreads over scalar
semantic categories including animacy (animated > unanimated), individuation (definite >
indefinite; singular > plural), agentivity (agent > patient), and syntactic categories (subject >
object > adverbial). In the course of time, these factors grammaticalize, i.e., the syntactic head
of the noun phrase is capitalized irrespective of its meaning. Our analysis is based on a
database consisting of 92 ENHG protocols of interrogation of witches dating from 1570 to
1665 (edited by Macha et al. 2005).
In our talk, we present the annotational method used in our project. In total, 65,00070,000
words are annotated and analyzed. For the computer-based text annotation we use the tools
provided by GATE 5.2.1 (General Architecture for Text Engineering).
Preliminary results of our project are presented on the basis of four pre-annotated protocols:
(i) The high degree of animacy is crucial for the use of sentence-initial capital letters,
independent of the paramaters of geographical area and syntactic position. The amount of
capitalized nouns is up to three times higher with animate nouns than with inanimate
ones.
(ii) The combination of animacy and the semantic role of agent leads to an even higher
frequency of capitalized nouns.
(iii) The syntactic function of nouns plays a role in the increased use of capital letters; i.e.,
capitalized nouns in subject function occur significantly earlier and more frequently than
the same nouns in object function.
(iv) Finally, the combination of animacy and subject function leads to an even higher number
of capitalized nouns.

Diachronic text charting

Erwin R. Komen, Radboud Universiteit Nijmegen & SIL-International
E.Komen@Let.ru.nl

This paper describes an approach to capture changes in the structure of the English language
by making use of a field-linguistics technique that has been in use for several decades: textcharting. I demonstrate the strength of manual text-charting as applied to an Old English (OE)
and late Modern English (LmodE) text, and I then propose an objectivized version of charting
that can automatically generates charts from syntactically encoded texts (Komen, 2013).
Application of this technique to several narratives visualizes the structural changes that took
place in English.
Charting methods have their roots in Longacre & Levinsohn (1978), and continue to be
updated (Dooley and Levinsohn, 2001, Huttar, 2003, Levinsohn, 2009). When charting a text,
a linguist divides the constituents in each clause over different columns, which may be
distinguished by syntactic function (subject, object, finite verb), by pragmatics (established
versus non-established information) or by position characteristics (preposed, postposed), as
exemplified in Figure 1.
Figure 1 Charted representation of two sentences from OE Euphrosyne
#

Intro

PreCore
PreAP

25 a

then
88

PreSbj Vb1
wurdon
were
Ongemang issum
com
Meanwhile
came

Core
Sbj
Est Nest AP
Vb2
hire yldran
swilice
geblissode
her parents
exceedingly made-joyful
[Postposed]
ham
home

PostCore
urh hi
through her
Pafnuntius
Paphnutius

The number and purpose of each of the columns is chosen such, that the result is economical
and informative: only a definable fraction of constituents may appear in columns where they
do not normally occur, which is marked by tagging their normal column with labels such as
preposed or postposed. One result of charting a text is that we end up with a set of
columns (a slot-structure) that represent the structure of the texts language.
Since the slot-structures based on the manually charted texts could be argued to be
error-prone and subjective, I present an automatic charting approach: an algorithm that takes a
text from the available syntactically parsed English texts as input, and chooses the best slotstructure based on a few assumptions (such as: take a fixed slot for the finite verb). Applying
this approach to 17 narratives from OE to LmodE shows: (a) a gradual reduction of the total
number of slots needed from 8 to 6, (b) reduction from two to one dedicated subject slot by
the second half of early Moder English, and (c) a fixed number of 3 slots preceding the finite
verb.
References
Dooley, Robert A., and Stephen H. Levinsohn. 2001. Analyzing discourse: basic concepts:
Summer Institute of Linguistics.
Huttar, Lars Andrew. 2003. Constituent Charting for Discourse Analysis: Information Model
and Presentation Model. MA thesis, Graduate Institute of Applied Linguistics
Komen, Erwin R. 2013. Finding focus: a study of the historical development of focus in
English. PhD dissertation, Radboud University Nijmegen
Levinsohn, Stephen. 2009. Self-instruction materials on narrative discourse analysis: SILInternational, <www.sil.org/~levinsohns/narr.pdf >.
Longacre, Robert, and Stephen Levinsohn. 1978. Field analysis of discourse. Current
trends in textlinguistics. ed. by Wolfgang U. Dressler, 103-122. Berlin, New York,
Walter de Gruyter.

Margot Kraaikamp
University of Amsterdam

Variation in pronominal gender agreement: semantic versus lexical gender in the history of Dutch

This talk addresses the question of what determines the diachronic variation between two types of
pronominal gender agreement: lexical gender agreement, that is agreement with the lexically stored
gender of the antecedent noun, and semantic gender agreement, that is agreement with certain
properties of the referent. Present-day Dutch shows variation between these two types of
agreement. Semantic agreement in Dutch is based on the degree of individuation of the referent:
masculine/common gender pronouns are used for referents with a high degree of individuation,
while neuter gender pronouns are used for referents with a low degree of individuation. In spoken
language, this semantic agreement occurs in more than half of the pronominal references, a number
that seems to have increased over time (Audring 2009).
It has been suggested in the literature that semantic agreement based on individuation has
developed in response to the change from the original three-way nominal gender system (i.e.
masculine, feminine, neuter) to a two-way nominal gender system (i.e. common and neuter). In this
view, the uncertainty about the gender of formerly masculine and feminine nouns has led to a new,
semantic interpretation of the pronouns (e.g. Audring 2009; De Vogelaer & De Sutter 2011; De Vos &
De Vogelaer 2011).
I will present data, however, that show that semantic agreement based on individuation
occurred in 16th Century Dutch as well (before the conflation of masculine and feminine nominal
gender), although on a smaller scale than in present-day Dutch. This finding suggests that semantic
agreement based on individuation is not new, but something that has always existed beside lexical
gender agreement. It seems therefore that lexical and semantic gender agreement have been in a
long-standing competition with each other, with different outcomes over the centuries. The
question that arises is what determines the preference for one or the other type of agreement in
different time periods. I propose that an important factor is the general visibility of lexical gender in
the language, in the form of lexical gender marking on adnominal elements.

References
Audring, Jenny. 2009. Reinventing pronoun gender. Amsterdam: Vrije Universiteit Amsterdam
dissertation.
De Vogelaer, Gunther & Gert De Sutter. 2011. The geography of gender change: Pronominal and
adnominal gender in Flemish dialects of Dutch. Language Sciences 33. 192205.
De Vos, Lien & Gunther De Vogelaer. 2011. Dutch gender and the locus of morphological
regularization. Folia Linguistica 45.2. 245-281.

Diachrony of relative clauses: postnominal and prenominal relatives in Basque

Dorota Krajewska
University of the Basque Country
Typological properties of Basque are usually summarized along these lines: Its basic word order is SOV, and it exhibits
virtually all the typological characteristics commonly associated with SOV word order. Apart from lexical adjectives, all
modifiers are preposed, and this includes large and syntactically complex modifiers like genitives and finite relative
clauses (Trask 1998:320). While accurate for the modern Basque, this view of the language has been questioned for
earlier stages of Basque (e.g. Lakarra 1995, 2006). Among the features that are in conflict with the properties of a wellbehaved SOV language, there are postnominal relative clauses, illustrated in (1). They are attested in all dialects (and
th
account for as many as 40% of relative clauses in some 17 centurys texts), but have only been retained in northern
varieties (as a marked and infrequent form). (2) is an example of a prenominal relative, which prevails in modern
Basque. Both constructions are head-external, embedded and marked with a clause final complementiser. They differ
only in the order of the constituents.
(1) gizon [lur
asko
dauka-n]-a
man
land
a.lot
have.3sg-COMP-DET
(2) [lur
asko
dauka-n]
gizon-a
land
a.lot
have.3SG-COMP man-DET
a man who has a lot of land
Even though the existence of postnominal relatives has not passed unnoticed in Basque linguistics (e.g. de Rijk 1980,
Oyharabal 1987), their history has not been examined in detail. The goal of the present paper is thus to study the
th
th
diachrony of relative clause constructions in Basque between the 16 and 20 centuries.
First, the paper addresses the diachronic changes in frequencies of postnominal and prenominal constructions.
Furthermore, their syntactic properties are discussed, as in old texts there seem to be some differences as to which
syntactic positions are relativized in each construction: while object relatives and relatives from locative phrases are
typically prenominal, transitive subjects tend to be relativized by means of the postnominal construction (intransitive
subjects are equally common with both variants).
Factors that might have contributed to the loss of the postnominal construction will be discussed as well. Hendery
(2012) argues that language contact is the most frequent reason for a change in relative clauses. In the case of the
Basque postnominal construction, it could not play a role, as Romance languages have postnominal constructions.
Instead, the loss seems to be related to other changes in the word order, as several constructions which nowadays have
preposed modifiers, once were also used with postposed modifiers: non-finite relative clauses, genitives and phrases
with the relational suffix (derived adjectival modifiers). Basque provides thus further evidence for the generalization
proposed by Lehmann (1984:394) and modified by Hendery (2012:207), namely that the gain of a new prenominal and
loss of postnominal relative happens when the language undergoes significant changes in the word order.
References:
Hendery, R. (2012). Relative Clauses in Time and Space: A Case Study in the Methods of Diachronic Typology.
Amsterdam: John Benjamins.
Lakarra, J. A. (1995). Reconstructing the Pre-Proto-Basque Root. In J. I. Hualde, J. A. Lakarra, & R. L. Trask (Eds.),
Towards a History of the Basque Language, 189206. Amsterdam: John Benjamins.
Lakarra, J. A. (2006). Notas sobre iniciales, cambio tipolgico y prehistoria del verbo vasco. Anuario del Seminario de
Filologa Vasca Julio de Urquijo, 40(1-2):561622.
Lehmann, C. (1984). Der Relativsatz: Typologie seiner Strukturen, Theorie seiner Funktionen, Kompendium seiner
Grammatik. Tbingen: Gunter Narr.
Oyharabal, B. (1987). Etude descriptive de constructions complexes en basque : propositions relatives, temporelles,
conditionelles et consessives. PhD thesis, Universit de Paris VII.
Rijk, R. P. G., de, (1980). Erlatiboak idazle zaharrengan [Relative clauses in old writers]. Euskera, 25(2):525536.
Trask, R. L. (1998). The Typological Position of Basque: Then and Now. Language Sciences, 20(3):31324.

On word order in infinitival complement constructions in Early New High German

Julia Krasselt (Ruhr-University Bochum, Germany)
Word order is the subject of controversial discussions within the field of historical linguistics. In my presentation,
I will talk about constituent order in Early New High German (ENHG) verb-second main clauses containing an
infinitival complement introduced by the particle zu. For this purpose, I will present a corpus study based on
historical German texts compiled in the 15th century in two different dialectal regions of Germany. The corpus
has a size of approximately 12,500 sentences and is annotated with part-of-speech information.
This study is concerned with two aspects of word and constituent order: first, the relative serialization of the
subcategorizing matrix verb and the embedded zu-infinitive; second, the position of arguments subcategorized by
the embedded zu-infinitive. The placement of constituents will be analyzed within the framework of topological
fields (Hohle 1986), which is considered a well-established model for German sentence structure. Previous studies
have shown that there is much more variation and flexibility regarding verb and constituent order in clauses
containing an infinitival complement with zu in ENHG than in Modern Standard German (Ebert 1976).
(1)

Both parties asked to write this indicated contract into the book.
ACCUSATIVE OBJECT

INFINITIVAL COMPLEMENT
AUX
PREP OBJECT
SUBJECT

Sollichen obgeschribn handel

haben
beyde partey
gebeten
ins ratsbuch
zu
schreyben
Such
indicated contract have:AUX both parties asked:PTCP in.the book
INF. PARTICLE write: INF
LEFT SENTENCE BRACKET

RIGHT SENTENCE BRACKET

The example sentence in (1) illustrates the above-mentioned word order aspects: the perfect auxiliary haben
have and the participle gebeten asked form the sentence frame as it is typical for German V2 main clauses.1
The participle gebeten asked subcategorizes the infinitival complement zu schreyben to write. They are nonadjacent because the prepositional object of the infinitival complement (ins ratsbuch in the book) stands between
them. Both, the infinitival complement and its prepositional object, are extraposed to the right of the sentence
frame. However, the accusative object of the subcategorizing zu-infinitive (sollichen obgeschribn handel such
indicated contract) stands in topicalized position, i.e. to the left of the sentence frame. In contrast, this word
order would be highly marked in contemporary Standard German which does not easily allow for topicalization
of embedded objects. In sum, ENHG shows more variation concerning the positioning of constituents either in
topicalized or extraposed position or between the left and right sentence bracket.
Despite the fact that ENHG word order variations in this domain are long known to exist, there is a lack of
comprehensive studies on this topic as well as linguistically motivated explanations. In this paper, it shall be investigated which factors influence constituent placement in ENHG infinitival complement constructions. Special
attention will be paid to constituent weight (Behaghel 1909, Wasow 2002, Tily 2011) and dialectal differences as
factors influencing constituent serialization.
References
Behaghel, O. (1909). Beziehungen zwischen Umfang und Reihenfolge von Satzgliedern. Indogermanische Forschungen, (25):110142.
Ebert, R. P. (1976). Infinitival Complement Constructions in Early New High German. Niemeyer, Tubingen.
Hohle, T. N. (1986). Der Begriff Mittelfeld. Anmerkungen u ber die Theorie der topologischen Felder. In Weiss, W., editor, Kontroversen
alte und neue. Akten des VII. Internationalen Germanistenkongresses Gottingen 1985, pages 329340. Niemeyer, Tubingen.
Tily, H. J. (2011). Weight and Word Order in Historical English. In Bender, E. M. and Arnold, J. E., editors, Language from a Cognitive
Perspective: Grammar, Usage, and Processing, pages 223246. CSLI Stanford, Stanford.
Wasow, T. (2002). Postverbal Behavior. University of Chicago Press, Chicago.

1 The

finite verb occupies the second position in the clause, the rest of the predicate is positioned later with some intervening constituents
between both verbs.

Labile verbs and transitivity in Indo-European and beyond:

A diachronic typological perspective
Abstract
Leonid Kulikov
Ghent University
This paper deals with the history of transitivity oppositions in several groups of IndoEuropean languages. It focuses on the evolution of the system of labile verbs (= verbs that
can show changes in syntactic pattern with no formal change in the verb, such as English
The door opened ~ John opened the door). One of the most difficult issues of the historical
Indo-European syntax is the amazing diversity of scenarios of the evolution of labile verbs:
in several languages the class of labile verbs is constantly increasing (as in English or
Greek), while in other languages this class is decreasing (as in Vedic Sanskrit and many
Iranian languages). I will concentrate on this issue; particularly, I will subject to critical
analysis the hypothesis on the high degree of lability of the ancient Indo-European syntax
(see, e.g., Hirt 1937), elucidate the position of Indo-European from the point of view of the
history of transitivity.
Integrating data from Indo-European with evidence from non-Indo-European languages
which attest drastic changes in the system of labile verbs (such as Estonian or East
Caucasian), I will concentrate on mechanisms responsible for the emergence and
disappearance of lability. The scenarios of the rise of lability include, for instance,
phonological merger of the originally distinct members of a causative pair and content
accusative extension (reanalysis of content accusative constructions, which, normally,
should be qualified as a specific variety of intransitive pattern, as transitive-causative). The
most important diachronic process responsible for decline of labile syntax is the
development of the overt causative (or anticausative) morphology and the expansion of
productive morphological (anti)causatives.
I will argue that we observe two basic types of evolution (evolutionary types), attested in
the history of the system of transitivity oppositions and valency-changing categories in
Indo-European: the syncretic type, found in many Western branches, and the antisyncretic type (the syncretic strategy of encoding several intransitive derivations by means
of the middle morphology is largely abandoned; the Proto-Indo-European middle diathesis
is degrammaticalized and eventually disappears; the labile syntax, even if attested in some
ancient languages, tends to disappear), attested at least in some Eastern branches, in
particular, in Indo-Aryan. Typologically, the Eastern type shares more features with some
non-Indo-European families, such as Turkic.
References
Hirt, H. 1937. Indogermanische Grammatik. Teil VII. Syntax II. Heidelberg: Winter.

Disparity in loanword adaptation: Portuguese and English loans in Bengali

Aditi Lahiri1, Stephen Parkinson1, Ballari Ray Choudhury2
1
University of Oxford, 2Gokhale Memorial Girls College
Loan word phonology research continues to flourish and is of particular interest within
European and Indian contexts, which are melting pots of linguistic and cultural backgrounds.
Countless Portuguese and English words have been adopted in Indian languages and are part
of their fundamental vocabulary. This research examines the way in which Bengali phonology
was influenced by the Portuguese (around 16th C) and English (from mid 18th C). Although
the main activity of the Portuguese was on the west coast of India, Bengali has had a large
share of cultural and religious influence and the linguistic influence must have been quite
extensive as well, although we know surprisingly very little about it. Although substantial
literature exists on English influence on Bengali, little is known about Portuguese.
The Portuguese loans affected Middle Bengali, while English loans appear in the early
modern texts at a later period. Our research questions are as follows: (a) Was the influence of
Portuguese and English on the Bengali language restricted to additions to the vocabulary, or
did it effect the phonological system? (b) Did Portuguese loans have a differential effect on
Bengali compared to English loans? Popular literature suggests that English has had a larger
influence, but the languages have different linguistic systems and the way in which they have
been borrowed must differ.
In this talk we will focus on the effect the loans have had for the coronal stops in
Bengali. Although Indo-European did not have retroflex consonants, Sanskrit and its
descendants have two series of coronal stops, dental and retroflex. Although retroflex stops
are considered to be more marked in the languages of the world, Bengali uses these
consonants quite prolifically in its grammatical system. For example, a productive
reduplication process adds plurality by reduplicating a word and replacing the initial
consonant with []: [kal] 'day' ~ [kal-al] 'day-and such'. The particle indicating specificity
also begins with a retroflex stop: [boi] 'book'~ [boi-a] 'this specific book'.
Where loans are concerned, the following problem arises. Portuguese has a dental
] while English has an alveolar stop [t] and a dental fricative []. Neither have
retroflex consonants. How did Bengali listeners perceive and adapt these consonants into their
own language? Did they perceive them as different or were they indiscriminately borrowed as
some coronal variant? The typological literature on dental and alveolar consonants show that
these sounds do not contrast in many languages of the world and indeed, they are quite
difficult to distinguish (cf. Mohanan & Mohanan 1993). Initial study of loans suggests that the
Bengalis, however, could distinguish between the Portuguese dental stops and the English
alveolar phonemes. Compare the following words:
am] 'button' Portuguese boto
iear] English theatre [t]
As we can see, Portuguese dental stops appear to be borrowed as dental in Bengali, while the
English dental fricatives were adapted as aspirated dental stops and the alveolar stops as
retroflex consonants. This supports the acoustic distinction drawn by Stevens & Blumstein
(1979) in their examination of the contrast between the two sets of consonants. Thus, in terms
of phonological features, [CORONAL] consonants remained coronal. The fricatives were
borrowed as [SPREAD GLOTTIS] while the dental/alveolar distinction was maintained as
dental/retroflex. Consequently, one may conclude that not only did the Bengali listeners
distinguish dentals and alveolars, the alveolar consonant with higher frequency acoustic
energy were perceived as non-dental and borrowed as retroflexes.

A multidisciplinary approach to diachronic variation in Sardinian

The case of stop-plus-liquid sequences
Rosangela Lai
rosangela.lai@unifi.it

University of Florence
Since the Middle Ages, liquids in stop-plus-liquid clusters have been particularly unstable in
Southern Sardinian (Wagner 1941). They have systematically been cancelled by different kinds of
metathesis. Our database concentrates on both philological and dialectological data. The database
contains 92 items which had a stop-plus-liquid sequence in their etymological form. Each item was
compared with all the forms attested in the medieval Sardinian documents (references omitted) plus
the modern form in the main Sardinian dialects.
When comparing philological with dialectological data, it can be seen that, in Southern Sardinian,
metathesis is applied to all stop-plus-liquid sequences, except for a small group. An interesting fact
is that the same small group patterned differently under another respect: the obstruent of these
clusters did not undergo lenition. We expect that in a language with lenition processes, both simplex
stops and stops in stops-plus-liquid clusters would be affected. The crucial condition is being in the
intervocalic position. However, in this group, despite being in a seemingly suitable condition (i.e.
these stop-plus-liquid clusters were all in intervocalic position) lenition did not apply. Two puzzles
are thus to be solved: a) Why did only few stop-plus-liquid clusters not undergo metathesis? b)
Why, in clusters in which metathesis did not take place, did lenition not apply?
We believe that the absence of lenition is related to the absence of metathesis, and that the above
questions may have a common answer. We argue that a theoretical model, which offers a
phonological explication in terms of positional effects (Lowenstamm, 1996; Scheer, 2004), might
be a viable approach to address the issue. We believe that the reaction of obstruents and liquids to
the processes at hand is the expression of their structural condition. In our view, stop-plus-liquid
clusters in Old Sardinian were of two kinds: tautosyllabic (i.e. branching onset) and heterosyllabic
(i.e. coda-onset cluster). Only the former was the target of both metathesis and lenition. In a
heterosyllabic cluster, stops avoided lenition because the stop in this kind of cluster is a coda
consonant. Lenition applied only to intervocalic stops. Similarly, the liquid was not affected by
metathesis because it was in strong position: in a heterosyllabic cluster, the liquid sits in postcoda position.
Table (1) summarises the Southern Sardinian reflexes for Latin -BR- sequences, with a subdivision
between tautosyllabic and heterosyllabic configuration. Note that lenition for voiced stops involves
complete deletion (Wagner, 1941). To conclude, we argue that the contrasting behaviour of stops
and liquids is the result of positional factors related to the syllabic structure in which they occur.
(1) Stop-plus-liquid Sequences
Tautosyllabic
FEBRUARIU> *fe.bra.riu > friaru
Heterosyllabic
CALABRICE> *calab.ricu > kalavriu

Lenition and Metathesis

No Lenition, No Metathesis

Selected References
Lowenstamm, J., 1996. CV as the Only Syllable Type, Current Trends in Phonology Models
and Methods, J., Durand (ed.). Scheer, T., 2004. A Lateral Theory of Phonology. Berlin, Mouton de
Gruyter. Wagner, M., L., 1941. Historische Lautlehre des Sardischen, Halle (Saale), Max
Niemeyer Verlag.

The material presented here is from Lai, R., 2013. Positional Effects in Sardinian Muta cum Liquida. Lenition,
Metathesis, and Liquid Deletion, Ph.D. dissertation. University of Florence.

On the Development of Contrastive Voice in English Fricatives:

A Comparative Perspective
Stephen Laker, Kyushu University
Among Germanic languages, English is unique in developing a robust voicing contrast in fricatives (/f s / vs.
/v z /) in initial, medial and final positions. Various proposals have been put forward to account for this
situation (e.g. Minkova 2011), but they usually deal with English in isolation and do not question why English
evolved differently from its Continental relatives. This paper attempts to redress some of the weaknesses of
such English-centred research by comparing the changes in English with those in other Germanic languages
and reconsidering why English took a different path.
First, I look at early evidence for contrastive voice in fricatives. It can be shown that the first signs of
any such contrast are found in Medieval English. The Old English Corpus provides some tantalizing evidence
from spellings, but the codified nature of the language means that examples are few in number and difficult to
interpret. However, such spelling evidence can be shown to increase incrementally during the early medieval
period, which lends credibility to the contested Old English evidence and suggests a change from below.
Next, evidence is found in Early Middle Dutch, though to a far limited extent than English. Finally, Frisian
shows no real, or at best extremely limited, signs of a fricative voicing contrast even today.
The next question that I ask is whether intensity of French borrowing was responsible for a stronger
influence on English phonology than Dutch. To answer this question, I draw on results from Early Medieval
English and Early Middle Dutch. My results indicate that Early Middle Dutch in fact borrowed roughly the
same amount of French loans with contrasting fricatives as Early Middle English. Consequently, we must call
into question ideas floated by Nielsen (1994) and Minkova (2011) that larger numbers of French loans in
English helped create a more salient voicing contrast in English.
Assuming French influence played a slightly lesser role, more prominence must be given to reductive
processes and prosodic changes influencing fricatives in English. But why did weakening such as apocope and
degemination, which also occurred in Dutch and Frisian, affect English quite differently? Rather than ascribe
these knock-on effects in English to mere chance, I consider whether British Celtic influence (see Laker 2009)
may yet provide part of an explanation for priming later developments in English. Various theoretical
problems against this proposal were highlighted in Minkova 2011, but they result from misunderstandings
based on research that does not consider change resulting from language contact and second language
acquisition. Reconsideration of theoretical insights and empirical evidence from these key research areas
indicates that British Celtic influence may yet go some way towards explaining why English developed so
differently from Dutch, Frisian and other Germanic languages.
References:
Laker, Stephen. 2009a. An explanation for the early phonemicisation of a voice contrast in English fricatives. English
Language and Linguistics 13, 21326.
Minkova, Donka. 2011. Phonemically contrastive fricatives in Old English? English Language and Linguistics 15, 31
59.
Nielsen, Hans F. 1994. On the origin and spread of initial voiced fricatives and the phonemic split of fricatives in English
and Dutch. In Margaret Laing & Keith Williamson (eds.), Speaking our Tongues: proceedings of a colloquium on
medieval dialectology and related disciplines, 1930 and discussion 5166. Cambridge: Brewer.

Reorganizing Voice in the Diachrony of Greek:

The Role of Prescriptivism and Frequency in Language Change
Nikolaos Lavidas
Aristotle University of Thessaloniki
nlavidas@auth.gr
The aim of this paper is to examine a violation (for centuries of time) of the 1:1-formto-function-correspondence in the area of Voice. Intransitive change-of-state verbs
that have a causative counterpart (anticausative verbs) in Modern Greek can bear
active (alakse changed) or mediopassive (miothike reduced) or both active and
mediopassive endings (lerose/lerothike got dirtied). This voice alternation is not
only a synchronic state (cf., among others, Theophanopoulou-Kontou 2004;
Alexiadou & Anagnostopoulou 2004; Tsimpli 2006) but is a variation that started in
the Early Medieval Greek period and continues today. The variation was
refunctionalized: mediopassive endings marked all valency-reducing derivations (in
pre-Medieval Greek), and, therefore, could also appear with an anticausative reading.
Mediopassive endings mark only the promotion of arguments in the accusative case to
the subject position (in Medieval and Modern Greek), and, therefore, anticausatives
can again bear mediopassive endings. Active endings, on the other hand, result in
labile forms (verbs without a change in their form for both causative and anticausative
constructions), something that emphasizes the derivational relationship between the
causative and the anticausative construction (for instance, anticausative: alakse i
katastasi the situation changed causative: alakse tin katastasi (s/he) changed the
situation).
Verbs that can mark with the mediopassive voice not only the early valencyreducing derivation but also the late derivation of the promotion of the accusative
argument to the subject position appear in all periods with voice morphology variation
(for instance, skorpizo scatter; ex. 1a-c). Verbs that do not have an accusative
argument to promote lost this variation when voice in Greek was reorganized and can
bear only active endings in Modern Greek (for instance, teliono end; ex. 2a-c).
We will demonstrate that the stability over centuries and the reorganization of this
variation can be related to (a) prescriptivism and frequency and (b) processing factors,
respectively. Prescriptivism plays a role in the stability or re-emergence of the voice
morphology variation. On the other hand, the reorganization of the variation in Voice
can be connected to principles of language processing that can account for variation
patterns and their reorganization (cf., among others, Rohdenburg 1996; Hawkins
2011; Author et al. 2012). A corpus-based analysis shows the complex interaction of
prescriptivism and processing. Frequency (both lexical and categorical; cf., among
others, Krug 2003) appears to be influenced by prescriptivism but also to determine
the path and the outcome of the grammatical reorganization.

(1) skorpizo scatter

anticausative in Late Medieval and Early Modern Greek: mediopassive and active voice morphology
a. ke ta
nera
skorpizunde
and the.NOM.PL water.NOM.PL scatter.MP.PRS.3PL
And the water is scattering. (Florios and Platziaflora, 1201; AD 15)
b. na skorpisi
i
kakosini
tus
to scatter.ACT.PFV.3SG the.NOM

malevolence.NOM

their

...that their malevolence scatters (ceases). (Agapios Landos, Dietetics, 62, 12; AD 17)
anticausative in Modern Greek: mediopassive and active voice morphology
c. nera
ke fagita
skorpisan /
skorpistikan
sto
water.NOM.PL and food.NOM.PL

scatter.ACT.PST.PFV.3PL /

scatter.MP.PST.PFV.3PL

on-the

patoma
floor

Liquids and foods scattered on the floor.

(2) teliono finish
anticausative in Late Medieval and Early Modern Greek: mediopassive and active voice morphology
a. otan akomi o
polemos
dhen eteliothi
when yet
the.NOM war.NOM not
finish.MP.PST.PFV.3SG
When the war was not yet finished. (GV I, 343, 28-29; AD 17)
b. opu eteliosen
i
mahi
that finish.ACT.PST.PFV.3SG the.NOM battle.NOM
that the battle finished. (Agapios Landos, Geoponikon, 42, 17; AD 17)
anticausative in Modern Greek: *mediopassive, active voice morphology
c. *teliothike /
teliose
o
polemos
finish.MP.PST.PFV.3SG / finish.ACT.PST.PFV.3SG the.NOM war.NOM
The war finished.
References
Alexiadou, Artemis & Elena Anagnostopoulou. (2004) Voice morphology in the causative-inchoative
alternation: evidence for a non unified structural analysis of unaccusatives. In: Artemis Alexiadou,
Elena Anagnostopoulou & Martin Everaert (eds.), The Unaccusativity Puzzle: Explorations of the
Syntax-Lexicon Interface. Oxford: Oxford University Press, 114-136.
Hawkins, John A. (2011) Processing efficiency and complexity in typological patterns. In: Jae-Jung
Song (ed.), Oxford Handbook of Language Typology. Oxford: Oxford University Press, 206-226.
Krug, Manfred (2003) Frequency as a determinant in grammatical variation and change. In: Gnter
Rohdenburg & Britta Mondorf (eds.), Determinants of Grammatical Variation in English. Berlin/New
York: Mouton de Gruyter, 7-68.
Rohdenburg, Gnter (1996) Cognitive Complexity and Increased Grammatical Explicitness in
English. Cognitive Linguistics 7 (2): 149-182.
Theophanopoulou-Kontou, Dimitra (2004) The structure of the VP and the mediopassive morphology.
The passives and anticausatives in Modern Greek. Parousia 15-16, 173-206.
Tsimpli, Ianthi Maria (2006) The acquisition of voice and transitivity alternations in Greek as native
and second language. In: Sharon Unsworth, Teresa Parodi, Antonella Sorace & Martha YoungScholten (eds.), Paths of Development in L1 and L2 Acquisition. Amsterdam: John Benjamins, 1555.

Partial (de)grammaticalization and its consequences:

the English possessive
John J. Lowe
University of Oxford
The synchronic syntactic analysis of the Present-Day English (PDE) possessive s has been the subject of
a considerable volume of work (recently, for example Anderson, 2008; Payne, 2009; Brjars et al., 2013).
There have long been two widely-held and competing analyses of its synchronic status: one that it is a
clitic, the other that it is an (edge-)affix. However, there are few signs of progress towards a consensus as
to which analysis is correct; this is because, I argue, both approaches are well supported by the evidence.
In order to understand the synchronic status of the English possessive, I reconsider the evidence of Late
Middle English / Early Modern English, the period during which, according to advocates of the clitic
analysis, the originally affixal possessive degrammaticalized to a clitic. In many respects the evidence
is difficult to interpret, due to the imperfect attestation of different dialects in texts of different dates.
Nevertheless two facts previously noted in the literature are highly significant: firstly, there was a period
when a variety of lexically idiosyncratic possessive markers (i.e. affixes) were still found but invariant -s
(i.e. potentially a clitic) was a competing possibility in many of the same contexts; secondly, the so-called
phrasal possessive construction, widely agreed to be the crucial evidence in favour of a clitic analysis of
the PDE possessive, developed gradually, originating in certain sorts of complex NP and then extending
its reach (Rosenbach, 2004).
These facts, particularly the second, necessitate the conclusion that, if indeed degrammaticalization of the
possessive affix took place in LMidE/EModE, this degrammaticalization must have proceeded gradually.
That is, we cannot reckon with a sudden, catastrophic, reanalysis of affix to clitic; rather there must have
been a time when both a clitic and affixal possessive construction existed side by side.
Gradual syntactic change is well attested; again in English, the development of the category of modals
out of the Old English preterite-presents took many centuries to go to completion (pace Lightfoot, 1979).
However, the potential consequences of gradual syntactic development (e.g. the partial evolution of a
new syntactic category, or the partial grammaticalization of a clitic as an affix or vice versa) are highly
problematic for synchronic analyses, where formal modelling is considerably simpler and more elegant if
we can say that a category clearly does or does not exist, or that a certain morphosyntactic element is
clearly one thing (a clitic) or another (an affix), but not both.
I argue that, given the clear evidence for a period in LMidE/EModE when the English possessive can
be characterized as both clitic and affix, and given the good evidence for both clitic and affixal analyses
of the PDE possessive, it is reasonable to assume that even in PDE the possessive is indeed both. That
is, the degrammaticalization of the possessive from affix to clitic has not (yet) been carried through to
completion.
A synchronic analysis of the PDE English possessive should therefore reflect its complex nature. I show
how it might be possible to provide a coherent synchronic analysis of the PDE possessive as both clitic
and affix, and argue that synchronic syntax must be prepared to deal with phenomena such as partial
(de)grammaticalizations.

References
Anderson, Stephen R. (2008). The English Group Genitive is a Special Clitic. English Linguistics 25 , pp. 120.
Brjars, Kersti, David Denison, and Alan Scott (eds.) (2013). Morphosyntactic Categories and the Expression of Possession.
Amsterdam: J. Benjamins.
Lightfoot, David (1979). Principles of Diachronic Syntax . Cambridge: Cambridge University Press.
Payne, John (2009). The English genitive and double case. Transactions of the Philological Society 107 (3), pp. 322357.
Rosenbach, Anette (2004). The English s-genitive: A case of degrammaticalization? In Olga Fischer, Muriel Norde, and Harry
Perridon (eds.), Up and Down the Cline: The Nature of Grammaticalization, Amsterdam: Benjamins, pp. 7396.

Ordinals and superlatives in Proto-Indo-European

Eugenio R. Lujn
Universidad Complutense de Madrid
Since the beginnings of Indo-European comparative linguistics (e.g., Brugmann
1911: 50ff) it has been noticed that ordinals and superlatives share the same suffixes in
the Old Indo-European languages, basically the thematic suffixes *-mo- and *-t(h2)o-.
This morphological coincidence was explained as due to the semantic connection
between the two categories. Benveniste (1948), elaborating on previous works, insisted
on the idea that both ordinals and superlatives had basically the function of signalling the
end of a series, and this idea is still found in recent books (e.g., Meier-Brgger 2003:
236).
In this paper I will revisit the alleged semantic connection between ordinal
numerals and superlatives in the Old Indo-European languages and its implications for
the reconstruction of the ordinal and superlative suffixes in Proto-Indo-European. From
the semantic point of view, it will be argued that ordinals do not necessarily mark the end
of a series as Cowgill (1970: 118) remarked, in Old Indo-European languages (like in
modern languages) we can expect instances of a fourth followed by a fifth and a
sixth, which proves that that is not an inherent semantic property of ordinals.
In fact, the link between ordinals and superlatives lies in the ordinal first, which
is the only ordinal numeral that shares a series of semantic and combinatory properties
with the superlatives (cf. Sleeman 1984: 62-64 for modern French). However, in standard
explanations of ordinals in Proto-Indo-European (e.g., Szemrenyi 1960: 67-114), the
ordinals first and second were left apart since they are usually exceptions to ordinal
formation rules. Another piece of neglected evidence has generally been the morphology
of the words for last, the natural opposite to first and a superlative in nature, too.
When we have a closer look at the relationship between the morphology of
ordinals and superlatives in the Old Indo-European languages, an interesting pattern
emerges: the extension of a suffix from the ordinals to the superlatives has been generally
achieved through the intermediate of the ordinal first (and the adjective last). This has
interesting morphological implications concerning the restrictions operated on the
extension of the ordinal suffixes. Apparent exceptions, like Skt. pratham- or Goth.
frumist, can also be accounted for if we analyse the history of the formations.
REFERENCES
Benveniste, mil 1948: Noms dagent et noms daction en indo-europen. Paris: A.
Maisonneuve.
Brugmann, Karl 1911: Grundriss der vergleichenden Grammatik der indogermanischen
Sprachen, vol. 2.2, Strasbourg (2nd ed.).
Cowgill, Warren 1970: Italic and Celtic superlatives and the dialects of Indo-European,
in G. R. Cardona Henry M. Hoenigswald Alfred Senn (eds.), Indo-European
and Indo-Europeans, 113-153. Philadelphia, University of Pennsylvania.
Hurford, James 1987: Language and Number: the emergence of a cognitive system.
Oxford New York, Blackwell.
Meier-Brgger, Michael 2003: Indo-European Linguistics. Berlin New York, Walter de
Gruyter.
Sleeman, P. 1984: Description smantico-syntaxique du numral ordinal", Travaux de
linguistique (Ghent) 11 : 55-69.

Two ways to conceptualize emotions: competing constructions in Ancient Greek

Silvia Luraghi and Eleonora Sausa, University of Pavia

Experiential situations typically involve the occurrence of an experiencer and a stimulus. When both
participants are present, experiential situations are encoded by bivalent verbs of various type: in Ancient
Greek, in particular, either the stimulus or the experiencer can function as subjects (mostly in the nominative,
with a small amount of non-canonical dative subjects, see Conti 2010). Experiential situations may be of
different types. In this paper, we concentrate on verbs of emotion, and discuss two constructions: type (i)
Nom/Dat and type (ii) Nom/Gen (possibly alternating with type (iii) Nom/Acc). In these constructions, the
nominative encodes the experiencer (experiencer-actor type, see Verhoeven 2007: 79), and the dative or
genitive the stimulus. Our study is based on an exhaustive analysis of all verbs in a diachronic corpus,
constituted by the Homeric poems, Herodotus Histories, a choice from Platos Dialogues, and the New
Testament. We show that type (i) construction, with dative stimuli, typically occurs with verbs that mean be
indignant toward, be angry at, be enraged with (e.g. nemeszomai, aposkudman, epimn, kot,
skudman, khol, khmai), and usually do not exhibit case variation (in particular, accusative stimuli do
not occur, i.e. type (iii) Nom/Acc construction is not a possible option). Type (ii) construction, with genitive
stimuli, on the other hand, most often occurs with verbs that mean love, desire, yearn (e.g. ldomai,
epimaomai, ramai, lilaomai). The same verbs occasionally can take type (iii) construction, Nom/Acc,
featuring accusative stimuli, and some semantically similar verbs always occur with Nom/Acc (e.g. poth).
Remarkably, all the above verbs typically take human stimuli, so animacy is not relevant for the choice
between either construction. A wider analysis shows that type (i) construction usually occurs with verbs of
communication, fighting, and other types of human interaction, which do not feature Nom/Acc as a possible
alternative. Type (ii) construction frequently occurs with verbs of hitting, touching, striving for and, in
Homeric Greek, directional motion, and may alternate with type (iii) Nom/Acc (a quantitative
correspondence analysis and an association plot will be provided). It is argued that the two types of emotion
verbs are conceptualized as representing different situations, i.e. active interaction of the experiencer with
another human being or direction toward a (non-interacting) goal. These different conceptualizations explain
the extension of either construction to one of the two specific sets of verbs of emotion, as well as possible
alternation of Nom/Acc with Nom/Gen but not with Nom/Dat. The diachronic analysis shows further
tendencies in the distribution of constructions (i) and (ii), as well as possible development of transitivity
features (e.g. possible passivization, see Conti 1998, Luraghi 2010).
References
Conti, Luz 1998. Zum Passiv von griechischen Verben mit Gen. bzw. Dat. als zweitem Komplement,
Mnchener Studien zur Sprachwissenschaft (MSS), 58:13-50.
Conti, Luz 2010. Weiteres zum Genitiv als Semisubjekt im Altgriechischen: Analyse des Kasus bei
impersonalen Konstruktionen, Historische Sprachforschungen 122, 2009 (2010):182-207.
Luraghi, Silvia 2010 The extension of the passive construction in Ancient Greek. In Acta Linguistica
Hafniensia 42/1: 60-74
Verhoeven, Elisabeth 2007. Experiential Constructions in Yucatec Maya. Amsterdam/Philadelphia:
Benjamins.

Contact-induced grammaticalization and the sociolinguistics of contact

Theodore Markopoulos
University of Patras
This paper discusses the relationship between grammaticalization and sociolinguistics
in language-contact settings. Although the study of grammaticalization phenomena in
situations of language contact has gained considerable momentum since Heine &
Kutevas (2005) book, there remain a variety of relevant issues to be determined, as
Heine & Kuteva (2010) readily admit. This paper aims to tackle one of these
debatable topics, namely the significance of sociolinguistic factors for the
understanding of contact-induced grammaticalization.
Heine & Kuteva (2010: 101) claim that sociolinguistic variables only play
a minor role and can largely be ignored in the study of grammatical replication. I will
try to show that this belief cannot do justice to the complexity of the phenomenon
involved on the following grounds:
a) It is built on rather unstable premises, as the evidence given to support it (e.g.
Spanish in the Basque country, the variety of English in the Channel Islands)
will be shown to be inadequate
b) It is rather erroneous to claim, as Heine & Kuteva (2010) do, that the
dominant vs. non-dominant distinction is of no relevance for grammatical
replication; there is widespread evidence that speaks to the contrary (cf. also
Matras, 2012)
c) On purely theoretical grounds, to claim that a contact-related phenomenon can
be studied largely without reference to sociolinguistic variables (such as extent
of bilingualism in relevant linguistic communities, register of interference etc.)
is highly dubious.
To further strengthen this view, two possible cases of contact-induced
grammaticalization in Medieval Greek will be discussed (the development of the
relative pronoun o opoios and the emergence of the periphrastic adjectival
comparatives), whose understanding seems to depend on the knowledge of their
sociolinguistic setting.
Finally, the paper discusses whether the mere occurrence of instances of
contact-induced grammaticalization can provide us with clues with regard to the
overall language contact situation they occur in and, more precisely, how it correlates
with two known types of language contact settings, namely adult bilingualism leading
to simplification and extensive, long-lasting bilingualism leading to complexification
(cf. e.g. Trudgill, 2010).
References
Heine, B. & T. Kuteva (2005) Language contact and grammatical change. Cambridge:
Cambridge University Press.
Heine, B. & T. Kuteva (2010) Contact and Grammaticalization. In: R. Hickey (ed.), The
Handbook of Language Contact. Oxford: Wiley-Blackwell.
Trudgill, P. (2010) Contact and Sociolinguistic Typology. In: R. Hickey (ed.), The
Handbook of Language Contact. Oxford: Wiley-Blackwell.
Matras, Y. (2012) An activity-oriented approach to contact-induced language change. In: I.
Leglise & C. Chamoreau (eds.), Dynamics of contact-induced change. Berlin: Mouton de
Gruyter.

The Ongoing Historical Developments of the Come-V and the Go-V Sequences in English
Noriko Matsumoto (noriko-mtmt@pop02.odn.ne.jp)
Kobe University
This paper demonstrates the ongoing historical developments of the come-V and the go-V sequences in
Present-Day English, relying upon the Corpus of Historical American English (COHA). The developments are
discussed in comparison with the uses of the come/go-and-V sequences, which are in many ways semantically
similar to the come/go-V sequences. In this paper, the come-V sequence is semantically divided into two types,
the motion type and the hortative type (Visser 1969), as in (1). The go-V sequence is divided into three types,
the motion type, the aspect type where go functions as a marker of aspect, and the modality type which signals
the modal notion of counter-expectation, as in (2).
(1) a. motion type:
She will manage to come see us.
b. hortative type: Come join us. (We would very much like to welcome you.)
(2) a. motion type:
You can go buy food somewhere else.
b. aspect type:
Go shut the door. (used to express to start to act so as to V2)
c. modality type: He didnt even leave a message. Go figure.
(used to show that you think something is strange to explain)
Both the come-V and the go-V sequences have one common syntactic feature: Only non-past forms of the verbs
come and go are acceptable, and V2 is always in a bare form. In order to compare with such come/go-V
sequences, the come/go-and-V sequences to be discussed in this paper are restricted to those in which come/go
and V2 are in the non-past forms.
This paper provides clear evidence for the frequency of the come/go-V sequences. Figures 1 and 2 show
instances of the come/go-V and the come/go-and-V sequences, respectively, per million words in COHA. Both
Figures also show that the quantitative use patterns of the come-V and the go-V sequences are different. This
raises two questions to be resolved. First, why does only the frequency of the go-V sequence show a fairly
marked increase from the 1940s onwards? The answer is that the go-V sequence is replacing the go-and-V
sequence. The movement towards the replacement has continued growing since the 1940s. In most cases, the
same replacement movement happens with V2 used in the go-V sequence. Especially, such replacement
movements are perfectly obvious in the modality type and the top five V2s (see, get, find, tell, and take) used
most frequently in the go-V sequence in COHA. Whether the go-V sequence expresses motion or functions as a
marker of aspect necessitates an understanding of contexts. It is tough to determine which types of the go-V
sequence involve an increase in frequency. However, it is fair to state that the go-V sequence is currently
undergoing historical development through replacing the go-and-V sequence.
Second, why does the come-V sequence not show as marked increase in frequency as the go-V sequence?
The answer is that the process of the come-V sequence replacing the come-and-V sequence comes later than the
go-V sequence. The same replacement movement as the go-V sequence happens with get and tell as V2, among
the top ten V2s used most frequently in the come-V sequence in COHA. The more complicated replacement
movement than the go-V sequence happens with some of the top ten V2s (visit, help, sit, and join). The
hortative type is more likely to replace the come-and-V sequence than the motion type. By contrast, the
replacement movement does not occur with see and take. Their quantitative use patterns are quite similar to
those of the come-V and the come-and-V sequences in Figure 1. It is thus fair to state that the come-V sequence
is currently in transition, and that there is a reasonable possibility that the come-V sequence will replace the
come-and-V sequence in the future.
In conclusion, the go-V sequence takes a different path from the come-V sequence with respect to historical
development, though both are currently undergoing changes.
40
30
20
10
0

60
40
0
1990s

1970s

1950s

1930s

1910s

1890s

1870s

1850s

go-&-V
1830s

1990s

1970s

1950s

1930s

1910s

1890s

1870s

1850s

1830s

1810s

come-&-V

go-V

20
1810s

come-V

Figure 1.
Figure 2.
Reference
Visser, Frederik Theodoor. 1969. An Historical Syntax of the English Language. Part Three, First Half:
Syntactical Units with Two Verbs. Leiden: E.J. Brill

Productivity in syntactic change: Polish and Czech impersonals

Roland Meyer
Institut fr Slawistik, Humboldt-Universitt zu Berlin meyerrol@hu-berlin.de
Although Polish (Pol) and Czech (Cz) impersonal constructions with an anonymous human agent dier in
some synchronic features (cf. Fehrmann et al. 2010 and references therein) e.g., Pol has a reflexive (SEtype) and a participial subtype (-no/-to-type), Cz only has the former , they also share important basic
properties: Both disallow the syntactic realization of the agent as a by-adjunct, and both can be formed
from transitive and intransitive verbs, including unaccusatives. Despite many treatments in the literature,
the diachronic development of these impersonals has not been analyzed satisfactorily yet: There is a general
consensus that impersonals somehow developed from true passives, expanding from transitives to unergative
intransitives, and later to unaccusatives. However, datings of changes dier by 200-300 years (cf. Pisarkowa
1984 vs. Shevelov 1968), argument-structural distinctions are being glossed over (cf. tcha 1988), and
the quantitative picture still remains quite unclear. Instances of clear impersonals can be found almost
throughout (1), and the alleged spread through verbal subtypes has not been proven yet.
(1)

a.

Lev Tet [zaal], aby se

kadvalo,
Innocentius, aby se
Pacem lbalo. [Cz]
L. third began that refl use-incense, I.
that refl peace kiss-3sg
Leo III. introduced the use of incense, Innocentius the kiss of peace. (Glat., 1565, Cz Nat. Corpus)

30
0

10

20

E[V(N)]

40

50

The paper analyzes the diachronic development of Pol and Cz impersonals (SE- and participial subtype)
on the basis of their changing productivity over time. We employ the current corpus-linguistic concept of
potential productivity (Baayen 2009, 902): According to a model based on the observed evidence, the amount
of newly occurring types (in our case, verb lexems) for various numbers of tokens (in our case, impersonals) is
estimated and compared; the slope of the growth curve indicates productivity. The use of a predictive model
remedies the dierences in size between diachronic samples, and
avoids the pitfalls of simpler productivity measures (Baayen
2009). Similar productivity models have been advanced for
diachronic morphology and syntax (Ldeling and Evert 2005;
Zeldes 2009); but cf. also Bybee (2003, 604) with a radically
dierent view based on pure token frequency. The data for the
present study stem from a diachronic corpus of Pol (ca. 2.8 mio
tokens), the Cz National Corpus and the Cz Academy corpus
(Vokabul webov) (2.9 mio tokens); query output (8916 Cz
XV
SE; 7322 Pol SE, 2217 participial) was grouped for construcXVIa
XVIb
tional subtypes, leaving, in the case of reflexives, 641 Pol and
XVIIa
XVIIb
805 Cz relevant examples, which were then marked up for verb
lemma and transitivity. A sample result (cf. fig. 1) shows
0
10
20
30
40
50
60
70
a productivity jump of SE-passives/impersonals between 16th
N
and 17th c. in Pol. For Cz SE-passives/impersonals, we deFig.1: Productivity of SE-impersonals in Pol
tected a similar non-continuous development in mid 16th c. In
N: No. of tokens, E[V(N)]: No. of types,
XVIa/b: 1st/2nd half of XVIth c. etc.
both cases, however, instances of intransitives were too rare to
fully account for the productivity jump.
Higher productivity of a pattern obviously points to its stronger anchorage in the linguistic system. According to Roberts (2007, 153), argument-structural change either proceeds as slow diusion of a pattern
through the verbal lexicon, or it aects all of a verb class suddenly, as an instance of reanalysis. Likewise,
we should expect slow, continous increases in productivity in the former case vs. short-term jumps in the
latter. Our data analysis clearly favors the reanalysis view. Moreover, the jumps in productivity point to a
reanalysis of the functioning of SE for all verbs, i.e., a switch from a passive to an impersonal marker.
Baayen, H. (2009): Corpus linguistics in morphology: Morphological productivity. In: Corpus Linguistics. An International Handbook,
eds. Anke Ldeling and Merja Kyt, Mouton de Gruyter, Berlin, New York, vol. 2, pp. 899919.
Bybee, Joan (2003): Mechanisms of Change in Grammaticization: The Role of Frequency. In: The Handbook of Historical Linguistics,
eds. Brian D. Joseph and Richard D. Janda, Blackwell, Malden, MA, pp. 602623.
Fehrmann, D.; U. Junghanns; and D. Lenertov (2010): Two reflexive markers in Slavic. Russian Linguistics, 34:203238.
Ldeling, Anke and Stefan Evert (2005): The emergence of productive non-medical -itis: corpus evidence and qualitative analysis. In:
Linguistic Evidence. Empirical, Theoretical, and Computational Perspectives, eds. Stephan Kepser and Marga Reis, Mouton de Gruyter,
Berlin, pp. 351370.
Pisarkowa, Krystyna (1984): Historia skadni jzyka polskiego. Ossolineum, Wrocaw.
Roberts, Ian (2007): Diachronic Syntax, vol. 8 of Oxford Textbooks in Linguistics. Oxford University Press, Oxford, UK.
Shevelov, George Y. (1968): Orzeczenia bezpodmiotowe odimieslowowe na -no/to w jzyku polskim przed rokiem 1450. Slavia Orientalis,
17(3):387393.
tcha, F. (1988): K vvoji zvratnho pasva ve spisovn etin. Listy filologick, 111:2229.
Zeldes, Amir (2009): Quantifying Constructional Productivity with Unseen Slot Members. In: Proceedings of the NAACL HLT Workshop
on Computational Approaches to Linguistic Creativity, June 5, Boulder CO. pp. 4754.

Roland Mittmann Goethe-Universitt Frankfurt am Main mittmann@em.uni-frankfurt.de

From the digitized glossary to the automatically pre-annotated text:

Pre-processing the grammatical data for the Old German Reference Corpus
The approach used by the DFG-funded research project Referenzkorpus Altdeutsch1 (Old German Reference Corpus)
shows how the pre-annotation of a historical plain text corpus can be largely automated: The respective glossaries are
digitized, the required information is retrieved, enriched and adapted to the standards used for the corpus, so that the
word tokens in the corpus can be automatically assigned the appropriate annotation datasets.
The project aims to produce a deeply-annotated corpus of all preserved texts from the oldest stages of German (Old
High German and Old Saxon), which date from ca. 750 to 1050 CE and comprise a total of 650,000 word tokens. The
largest coherent subcorpora are the Old High German works of Notker Labeo and Otfrid of Weissenburg, an Old High
German translation of the gospel harmony of Tatian the Assyrian and the Old Saxon gospel harmony now known as the
Heliand. Edited versions of all texts exist in print; they have been digitized by the TITUS2 project. The respective
glossaries, many of which are themselves over a hundred years old, typically give the lemmata together with their
translations and at least some of the morphological features specific to each lemma. The entries are completed by a list
of attestations, followed by a reference to their location within the text. The glossaries have been digitized into an XML
format (cf. Mittmann 2013)3.
All attested records of word tokens contained in a glossary are extracted with their corresponding lemma, part of
speech, inflectional information and their location within the text. To this end, the glossary file is scanned line by line
and the values are stored, and whenever a record appears, it is output into a file together with its corresponding
qualities. If a record is given within its context, it first has to be identified. Whilst a scholar can easily recognize the
word concerned within the context, the computer cannot unless the record is identical to the lemma. Otherwise, all
words of the phrase are checked whether their first letter is identical to the one of the lemma. If there are several of
them, the same is done with the first two letters, and so on. For the contingency that words contain a past participle
prefix, these are tentatively deleted and this option checked as well. If several possibilities remain, all are output.
Failing this, the same process is repeated numerous times, using a pair of lists of graphemes or grapheme clusters that
frequently correspond to each other within lemmata and records: All graphemes within the phrase covered by an entry
in the second list are tentatively replaced by the corresponding ones in the first. In case this still does not yield a result,
the same is performed again with another pair of lists covering rarer correspondences, including, for example, verbal
suppletion or inflectional forms of pronouns. The two pairs of lists are kept separate as the results of the application of
the frequent possible correspondences should be checked first to avoid rare possible correspondences being incorrectly
applied. The lists are set up manually by examining missing or incorrect results.
<lem>uusan</lem><pos>an. v.</pos> [...]
<case><form>imp. sg.</form><inst><expr>ouh thu uuis obar fimf burgi</expr>
<rec>151, 6</rec>
Figure 1: Excerpt from digitized glossary file (cf. Sievers 1892: 491 and 495)4
In the case of Figure 1, the record uuis can easily be recognized automatically as it is the only word token beginning
with u. For the eventuality of, for instance, a word token uuela also being contained in the phrase,5 the following
thus not only the third, but also the fourth grapheme is checked as well: uuel cannot be matched to uus, but uuis
can be matched to it, as the replacement of i by e is contained in the lists. The lists would also help to attribute a wordinitial vu or an inflected form ist to uusan.
Within the file, all part-of-speech and inflectional information is transferred into the standard of the Deutsch-DiachronDigital-Tagset (DDDTS), a tagset developed by the project and built on the basis of the Stuttgart-Tbingen-Tagset
(STTS) for modern German. The transfer is done by applying regular expressions. Automatically produced lists of all
part-of-speech and inflectional information occurring in the glossary facilitate this task, although in the digitized
glossaries these two types of information are not always clearly separated. The information from the glossaries is
1

http://www.deutschdiachrondigital.de
Thesaurus Indogermanischer Text- und Sprachmaterialien (Thesaurus of Indo-European Text and Language Materials), http://titus.uni-frankfurt.de
3
Mittmann, Roland (2013): Digitalisierung historischer Glossare zur automatisierten Vorannotation von Textkorpora am Beispiel des Altdeutschen.
In: Hoenen, Armin/Jgel, Thomas (eds.): Altberlieferte Sprachen als Gegenstand der Texttechnologie / Ancient Languages as the Object of Text
Technology (= Journal for Language Technology and Computational Linguistics JLCL, Vol. 27 2/2012). Berlin: Gesellschaft fr Sprachtechnologie und Computerlinguistik (GSCL). http://www.jlcl.org/2012_Heft2/3Mittmann.pdf
4
Sievers, Eduard (1892): Tatian. Lateinisch und althochdeutsch mit ausfhrlichem Glossar. 2nd edition. Paderborn: Schningh.
5
The difference between e and is disregarded here.
2

enriched by manually added rules developed using the relevant grammars that identify, for example, exact inflectional
classes of verbs and nouns from the shape of the lemma, whereas most glossaries only indicate a strong or a weak
inflection. Finally, all records and their corresponding information are stored in a file, depicted in extracts in Figure 2.
Lem | Lem2 | Lem3 | PoS | Flex | Form | Expr | Expr2 | Rec
Lemma DDDTS Lemmabezug Belegbezug Flexion
[...]
uu&euml;san | uusan | uuesan | an. v. | imp. sg. | uuis | uuis | 151, 6
VA
VAIMP irrst5
irrst5
Imp_Pres_Sg_2
Figure 2: Title and sample line from glossary data file
Figure 2 shows the data extracted from Figure 1, converted and enriched. The part-of-speech DDDTS tags VA/VAIMP
(verb, auxiliary, imperative) reflect the information v. and imp., the consideration as an auxiliary has been added
lemma-specifically. The lemma-specific inflectional information irrst5 also goes back to a manual amendment:
The glossary lemma uusan combines an irregular and a strong class 5 verb (cf. also Figure 3), the broken bar denotes
alternatives. The declaration an. (anomalous) which would only yield irr is thus ignored in this case. The second
irrst5 denotes the lemma-specific inflectional information of the record: Here, only st5 would be correct, but this
selection is left to the manual annotation. The record-specific inflectional information Imp_Pres_Sg_2 is generated
completely from imp. sg.; Pres and 2 are added automatically as there is an imperative only of the present tense,
and in the singular, there is only one of the second person.
In the case of the Old High German texts, the various forms of each lemma and its translations are given in a unified
form corresponding to the entries in Splett (1993)6, which covers the whole Old High German lexicon using a
standardized orthography. Automatically generated lists of lemmata from each of the Old High German glossaries listed
are expanded by giving the form and the translation found in Splett (cf. Figure 3). To map the glossary lemmata to the
Splett lemmata, again, two pairs of lists are assembled. The first pair contains all replacement rules from the glossary
lemmata to the Splett lemmata that apply mostly or always. The second one contains rules that have to be tentatively
applied if the lemmata do not match or to enable exceptions from the former rules. The composition of the lists is
controlled by checking the alteration of the number of overall concordances when applying a rule. By this, a weighted
total average of 84 % of all lemma concordances can be calculated for the seven Old High German glossaries. If there
are several possible results, they are all output. In any case, the concordance list finally has to be precisely checked by
hand to detect mistakes, especially false friends that look alike, but actually equate to different lemmata. When the
lemma matching is done, the Splett translations can be added automatically.
We deviate from Spletts practice in that e unaffected by umlaut is marked as and fricative z as in order to
separate these pairs of phonemes according to the orthography used in, for instance, Braune (2004)7. To this, rules are
set up for the application of the different graphemes that can be determined from the history of Old High German. The
rules cover a weighted total average of 90 % (94 % for e/ and 77 % for z/) of all cases, and in the course of a
manual check of all concerned lemmata, an adaptation of the undecidable cases has to be performed.
uusan snwsan 'sein, werden, geschehen, [...]sein, werden, kommen, [...]'
Figure 3: Sample line from lemma concordance file (cf. Splett 1993: 815 and 1111)6
Figure 3 shows the attribution of the glossary lemma uusan to the Splett lemmata snwsan. In the first stage, the
Splett lemma wesan is automatically retrieved, before it is altered to wsan as the e stands before a vowel a in the
next syllable.8 sn is then added manually, for the forms of both verbs are subsumed under the same glossary lemma.
After the translations are added, ermattet, kraftlos and Sein, Grundlage are deleted manually, as there
are no equivalents in the glossary to the adjectival and substantival homographic lemmata wesan given by Splett
(1993: 1111 and 1113)6.
A subsequent program then links the pre-processed glossary data file and the lemma concordance file to the TITUS text.
This program matches every word of the text with the records in the glossary data file. If the numbering of the record
locations is identical in TITUS and in the glossary, a one-to-one assignment is possible. Otherwise, all corresponding
datasets are assigned, and all but one will be later discarded manually. Thus, the word token uuis in the phrase Themo
quad her: ouh thu uuis obar fimf burgi. (Tatian 151, 6; cf. Sievers 1892: 227)4 will be correctly pre-annotated solely
as shown in Figures 2 and 3 and not also as an uninflected form of the adjective ws wise (cf. Sievers 1892: 503)4.
6
7
8

Splett, Jochen (1993): Althochdeutsches Wrterbuch. Berlin: de Gruyter.

Braune, Wilhelm (2004): Althochdeutsche Grammatik. Band I: Laut- und Formenlehre. 15th edition, ed. by Ingo Reifenstein. Tbingen: Niemeyer.
The rare cases of e instead of before a, as in reda speech, are most effortlessly re-adapted by hand.

Retrieval and evaluationonthe basisof bilingualhistoricalcorporain referencetothe

translationofLatinparticipleconstructionsinOldHighGermantexts
NataliaMull(Jena)
The aim of the DFGfunded project Old German Reference Corpus (Referenzkorpus Altdeutsch) is to provide a
linguistically parsed digital database of the entire text corpusof writtenGermandatingbacktotheearlystages(around
7501050AD,thusincludingOHGandOS).
Due to the fact that a part of the Old German record depends more or less on Latin archetypes, it is necessary to
includeLatinparallelrecordingsinthedatabasetoprovideabetteropportunityforidentifyinginterferences.
In general, the parsing of the Latin corpus is based on the same principles as the annotation of the Old Germantexts.
That means, the same levels are annotated, such as lemma, translation,language,and furtherPOSforbothlemma and
form as well as morphological information regarding flectional classification and grammatical characteristics in the
particularcontext.
In thispaper wewould liketodiscusssomepossibilities ofthedataevaluationusingthe example ofthetranslation ofthe
Latin participle constructions in Tatian,whose LatinandOldHighGerman versions arerecorded inparallelcolumns, as
wellassomepassagesofIsidorandNotker.
As most of the works on syntactic characteristics ofthesetextsgoback tothe 19thcenturyandconcentratemainly on
the Germanic version and/or justone author, wewillattempttoconsiderthespecifics ofthe relevantLatin versions and
to combine the results found in different texts in order to better understand the way of the participle construction
translations in Old High German. The methods used here can then be used on other syntactic and morphological
phenomena,suchasusageofdifferentcases,subordinateandinfinitiveclauses,etc.
We will focus on the participleconstructionsgenerally known asparticipiumconiunctum. Amongst others,thespecial
case of the so called ablativus absolutus willalsobediscussed.Wewould liketoshow theways ofdataretrievaldue
totagging,statisticalevaluationofit,andsomepreliminaryremarksonitspossibleinterpretationandlinguisticanalysis.
At the beginning, some remarks on the parsing of the texts in consideration have to be made. To ensure the
comparability of diachronic data, a special tagset (DDDTS) for POS parsing has been developed based on the well
established StuttgartTbingen Tagset (STTS). The DDDTS tagset has then been adapted for the annotation and
parsing of the Latin version. During this process some POScategories of the traditional Latin grammar had to be
reconsidered depending on their synchronic contextual function so that some minor changes were introduced into the
POSclassificationofLatin.
Since the syntactic information is very important and since its tagging should beeasilyintegrated intothepreparation of
the other data, a system of syntactic annotation has been developed and tried out according to our approach to
annotating of the morphological and partofspeech information. It can be described as an elementary systemof linear
syntax annotation based on the tool ELAN (which is also used for morphological and partofspeech analysis). Thus,
clausesaremarkedinrespectoffiniteness,introduction,syntacticstatusandsemantics.
References
Denecke, A. (1886): Der Gebrauch des Infinitivs bei den althochdeutschen bersetzern des8. Und 9.
Jahrhunderts.Leipzig.
Gering, Hugo (1876): Die Causalstze und ihre Partikeln bei den althochdeutschen bersetzern desachtenund
neuntenJahrhunderts.Halle.
Eilers,Helge(2003):DieSyntaxNotkersdesDeutscheninseinenbersetzungen.Berlin:deGruyter.
Lippert, Jrg (1974): Beitrge zu Technik und Syntax althochdeutscher bersetzungen. Unter bes. Bercks. d.
Isidorgruppeu.d.althochdt.Tatian.Mnchen:Fink.
Raposo, Berta (1982): Die Wedergabe des lateinischen Ablativs in der althochdeutschen bersetzungsliteratur.
Gppingen:Kmmerle.
Schiller, A./S. Teufel./ C. Stckert/C. Thielen. (1999):GuidelinesfrdasTaggingdeutscher Textcorpora mitSTTS
(KleinesundgroesTagset).IMS,UniversittStuttgart,August1999.
NataliaMull,M.A.
LehrstuhlfrIndogermanistik
FriedrichSchillerUniversittJena
Zwtzengasse12
D07743Jena
natalia.mull@unijena.de

Evidence for a Scrambling Position in Old and Middle French

Synnve Midtb Myking (M.A.), University of Oslo
This paper argues the existence of a scrambling position between auxiliary and main verb in Old
French. This position favoured constituents both representing given knowledge and being close to
the verb in terms of argument structure, meaning that constituents whose base position would be to
the right of the verb, and which filled these criteria, would be able to move there. This is illustrated
by the prepositional phrase (PP) avec li in 1).
1) Quant li troi chevalier qui estoient avec li venu virent ceste chose, il en furent trop dolent [...]
when the three knights who were with him come saw this thing they of-this were too pained
"When the three knights who'd come with him saw this, they were most upset"
(Le roman de Tristan en prose, 409, 13th Century)

The paper is based upon a study of 552 sentences taken from 13th and 15th Century prose texts
containing a PP linked to an unaccusative verb composed with an auxiliary. This allowed for the PP
to be studied in terms of several factors: its placement with the regards to the subject, the auxiliary,
and the main verb; the type of sentence; the degree of given/new information represented by the PP;
and the era from which the data originated. Using as its basis the unaccusativity hypothesis and
distinguishing between PPs functioning as "locative objects" and PPs functioning as regular
adverbials, the study yielded several interesting findings:

The order Auxiliary PP Main Verb is well-represented in Old French, to be found in 27

% of the 13th Century sentences

The great majority of these "interverbal" PPs represent given information

An even greater majority of the interverbal PPs are locative objects, rather than regular
adverbials

The interverbal position is much less frequent in Middle French, found only in 7 % of the
15th Century material

To explain the loss of the scrambling position, the paper proposes a connection with the
phonological changes taking place at the time: as French developed a stronger phrase-final accent,
the placement of PPs demanding a certain amount of accentuation in our case the PPs functioning
as locative objects became less flexible, since the moving of such "heavy" constituents now
entailed going against the accentuation of the phrase.

Generic agents in Reefs-Santa Cruz: Proto Oceanic *<in> revisited

shild Nss, University of Newcastle, Australia
Aashild.Naess@newcastle.edu.au
van den Berg and Boerger (2011) argue, on the basis of evidence from Bola (Papua New
Guinea) and Natgu (Solomon Islands), that Proto Oceanic *<in>/ni- had a double function
as a nominaliser and a marker of an agentless passive. In Natgu, the prefix n-, which
presumably reflects *<in>/ni- , is described as having three functions: firstly, marking a 3rd
person augmented subject, usually in combination with a suffix; this is the only instance of
prefixal person marking in Natgu. Second, forming action nominalisations in combination
with the suffix ng; and third, forming an alleged passive. van den Berg and Boerger
suggest that the passive and nominalising functions are primary, and that the person-marking
function is a later development.
Natgu belongs to the Reefs-Santa Cruz (RSC) subgroup of the Temotu first-order subgroup
of Oceanic. Examination of the two other RSC languages for which data is available, iwoo
and Engdewu (Nagu), strongly suggests that the original function of the morpheme in
question is not marking a passive in the syntactic sense, but indicating a generic agent. This
function is not only closer to the Proto Malayo-Polynesian function of *<in> as marking
object voice, but also gives a more plausible path of grammaticalisation to the 3AUG marker:
from generic they to referential they rather than from passive to person marker. In
Engdewu, the prefix on its own still indicates a generic agent, and is only used for referential
3AUG subjects in combination with a suffix. The generic-agent prefix in iwoo, which also
marks referential 3AUG subjects of intransitive verbs, is a distinct morpheme reflecting a
3PL prefix *ri- or *ra-. This prefix seems to have replaced the reflex of *<in>/ni-, but such an
analysis requires the function of the latter to have been marking a generic agent rather than a
passive, since the morpheme in question has no passive function in iwoo.
It is entirely possible that the reflex of *<in>/ni- in Natgu has developed into a passive
proper, a plausible development from a generic-agent marker. However, a reexamination of
available data in light of the constructions found in the other RSC languages suggest that a
generic-agent analysis, despite being explicitly rejected by van den Berg and Boerger, may be
more appropriate in this language too; in particular, the behaviour of n- in participant
nominalisations, a construction not discussed by van den Berg and Boerger, can only be
readily explained through an assumption that it marks a generic agent. The evidence from the
RSC languages therefore points to POc *<in>/ni- as a marker of a generic agent rather than a
passive morpheme.
References
van den Berg, Ren, and Brenda H. Boerger. 2011. A Proto-Oceanic passive? Evidence from
Bola and Natgu. Oceanic Linguistics 50:1, 221-246.

Larissa Naiditch
The Hebrew University of Jerusalem
Quotative modality in Early New High German.
Evidence from texts of the 16th century
There are grammatical means in modern German that are used in cases, when the speaker
points to a proposition claimed by another person. Thus the sentence beginning with Er
soll gesagt haben, etc. is understood as he is quoted to say, he is alleged to say, he is
alleged to have said. These constructions consist of the modal verb sollen + infinitive,
predominantly infinitive II. Previous investigations demonstrate that the epistemic
modality, including the quotative one, was established in German as late as the end of the
Middle High German period, but was probably not common then (Mller 2001, Mach
2008, 2011). In the beginning of the 16th century (Early New High German) the quotative
use of modals has become frequent in certain type of text, e.g. in newspapers (Fritz
1991). The grammaticalization of the corresponding construction has been completed.
My research deals with the analysis of two travel books of the 16th century: 1) Daniel
Ecklins Rei zum heiligen Grab (1575); 2) Sigismund von Herberstein. Moscovia.
Der Hauptstat in Reissen (1557). The quotative modality in these texts is examined
by me. It is demonstrated that the corresponding constructions are very frequent in the
book by Ecklin, which can be explained by its genre and by the individual style of the
author. Those models are used to stress that the historical background of the places of
interest observed by the author has been told to him by somebody else. E.g.: 1) darinn soll
die schwester des heiligen Jeronimi gewesen sein. 2) dz soll von S. Helena auffgehaben vnd behalten
sein worden. [Vom Kreuz L.N.]. 3) Zu Antiochia soll auch Petrus der heilig Apostel das
Bischofflich ampt lange jar verwaltet haben. The verb sollen +infinive I can have a quotative
meaning, as well as an evidential dimension: 1) Bey diser Statt / in jrer gegne sollerschaffen
sein vnsere ersten Elteren / Adam vnd Eua.

In the book by Herberstein the quotative modals are not so frequent, but are also
observed: 1) Vor etlich wenig Jarn geschehen sein soldt / Ain Moscovitischer ansehenlicher
Khriegsman Michael Chisaletzkhj genant / der hat in ainer schlacht ainen nambhafften Tattern in die
flucht bracht. 2) die Littischen sollen nit meer dann 35000 darneben etlichs Veldgeschtz gehabt
haben

In all these cases it is stressed that the information conveyed is not based on the speakers
direct experience. In the examined texts the quotative constructions deal with historical
past.

The intrusive nasal in Ved. pmn, svadyn etc.: sound law or analogy?
Benedicte Nielsen Whitehead, University of Copenhagen
This paper aims to explain as phonologically regular the unexpected occurrence of what may be
described as an intrusive nasal in Vedic nominal suffixes in final -s.
The Vedic suffixes in -s are: -ms-/-ms- (of pms- m. man); -ys-/-yas- (comparative); vs-/-vas-/-vat-/-u- (perfect active participle); s/-s-/-s- (adjectives; neuter action-nouns).
Certain allomorphs of these suffixes display a nasal before final -s-; such allomorphs contain a long
vowel:

man m.
sweeter m./f.
having driven m.
active m/f.
work n.

Nom. sg.
p-mn1
svad-yn
cakr-van
apas
p-as

Acc. sg.
p-ms-am
svad-y-sam
cakr-vas-am
ap-s-am
p-as

Gen. sg.
pu-s-s
svad-yas-as
cakr--as
ap-s-as
p-as-as

Nom. pl.
p-ms-as
svad-ys-as
cakr-vas-as
ap-s-as
p-s-i

So far, comparative evidence has not helped to explain this peculiar ablaut pattern: no other
langugage attests a nasal in these suffixes; and we cannot exploit Vedic to posit an inherited nasal,
since no known rules of morphology or phonology would account for (1) its absence in e.g. the
genitive singular puss < *pu-ms-s or (2) nominatival lenghtenening in these stems (a feature
expected only before a single final consonant).
Scholars have looked in vain for an analogical source: AiGr 293-4 points to various suffixes in nt(vant-/-vat-; ant-/-at-) without seeming entirely convinced.2
However, if we look closer, comparative evidence does, after all, offer a clue to the problem: we
know from e.g. Greek, Latin and Vedic itself that the nasalized forms of these suffixes contained
either IE * (in the animate nom.sg.) or *o (elsewhere). Short *o merged with * in open syllables
(acc. to Brugmann's Law); * then merged with *. Thus, the specific environment for the
occurrence of our nasal was the position between * and *s. The spontaneous development of a
nasalized fricative3 under very similar circumstances is well known from Avestan, where *s
developed into a phoneme written h-, when followed and precreded by a or .4
In the paper I want to explore the hypothesis that the intrusive nasal in the Vedic suffixes is the
result of a sound law operating in Indo-Iranian and extending its scope in Avestan, but not in Indic
where, as I intend to show, analogy removed its occurrences from e.g. the causative and perfect.

1
2

Vedic allows only one .inal consonant; hence the suf.ix loses its -s- in the nom. sg.
AiGr.: Wackernagel, J. & Debrunner, A. (1929/30): Altindische Grammatik, 3: Nominal(lexion
Zahlwort Pronomen. Vandehoeck & Ruprecht, Gttingen.
On this phenomenon, see Shosted, R. K. (2006): The aeroacoustics of nasalized fricatives. PhD
Thesis. Graduate division, University of California, Berkeley.
Kellens, J. (1989): Avestique, in: Compendium Linguarum Iranicarum. R. Schmitt (ed.) Reichert,
Wiesbaden, pp. 32-55; see p. 42.

Exploiting variation to sidestep phonological problems

Mary Niepokuj
Purdue University
It has long been accepted that Proto-Indo-European originally had two ways of forming the perfect
verb. One process, attested in Greek lloipa I have left, involved reduplication, o-grade vocalism in the root in
the singular or zero-grade vocalism in the plural, and a distinctive set of person and number endings. The
second process, attested in Greek oda I have seen/have found out, is similar to the reduplicated perfect in all
respects except that it lacks the reduplicative prefix (Watkins 1998:57-58). This raises the question of exactly
how the two different morphological approaches to the perfect were distributed in PIE. Was reduplication
completely optional, as scholars such as Brugmann (1913) suggest? Or was the distribution of
reduplicated/non-reduplicated stems more systematic? Following Bader (1968) concerning Latin, I argue that
the distribution was largely systematic: when stems were unable to exhibit the o-grade/zero-grade vocalism
characteristic of the perfect, then reduplication was obligatory. In other cases, it was not required, motivating
the maintenance of two different morphological processes for the same grammatical category. The daughter
languages then differ in how they remade the inherited system.
Evidence for this claim comes from languages such as Gothic, Latin, and Old Irish, in which the
reduplicative system still maintains a relatively limited distribution. In these three languages, for example, roots
which originally had a-vocalism consistently show up reduplicated: Gothic fhan to seize, 3rd sg. pret. faifh,
from PIE * pak-; Latin can, I sing, 1st sg. perf. cecin from PIE *kan- ; OIr canid to sing, reduplicated stem
cechann- also from PIE *kan-. Another subgroup found in Latin are present stems which themselves already
exhibit o-vocalism; morde I bite, 1st sg. perf. momord (Bader 1968). Under this analysis, daughter languages
such as Gothic, Latin, and Old Irish point to the original distribution of forms; languages such as Greek and
Sanskrit, in which reduplication has been extended analogically throughout the perfect system with the
exception of a few relic forms are innovators.
The Proto-Indo-European case offers evidence that variant grammatical systems can be maintained in
a language if they offer speakers the possibility of sidestepping phonological difficulties. In this respect, the PIE
perfect verb parallels other examples such as the increase in Early Modern English of the use of unemphatic do
in non-negative non-interrogative contexts in which no emphasis can be discerned. As Rissanen (1999:241)
notes, one factor in the rise of constructions such as thou didst imagine is that it offered a way of avoiding
awkward consonant clusters; compare thou imaginedst, which, after loss of the schwa in the final syllable,
would have had the final cluster /ndst/. The increase and maintenance of constructions with unemphatic do is
explicable partly because it offers a way around phonological difficulties, as does the use of reduplication for
perfect verbs which cannot participate in the normal ablaut system. This type of functional explanation has
been criticized as teleological (for example, Lass 1997:340-352); however, as the focus of historical explanation
shifts from changes in abstract systems to choices made by individual speakers in the act of speaking, the
objection to functional explanations falls away.

Partial Bibliography
Bader, Francoise. 1968. Vocalism et Redoublement au parfait radical en latin. Bulletin de la Societ de
Linguistique de Paris 63.160-196.
Lass, Roger. 1997. Historical linguistics and language change. Cambridge studies in linguistics 81. Cambridge:
Cambridge University Press.
Rissanen, Matti. 1999. Syntax. The Cambridge history of the English language, vol. III: 1476-1776, ed. By
Roger Lass. Cambridge: Cambridge University Press.
Watkins, Calvert. 1998. Proto-Indo-European: Comparison and reconstruction. The Indo-European
Languages, ed. by Anna Giacalone Ramat and Paolo Ramat. New York: Routledge.

Latin constructions with participles in a diachronic perspective

Tatiana Nikitina (CNRS) and Dag Haug (University of Oslo)
Latin participles appear in two types of construction that are unusual from a modern European perspective: absolute
constructions, and dominant participle constructions (Ramat 1994). In the absolute constructions (which are also attested
in other Indo-European languages), the participle agrees in gender and number with a noun phrase that corresponds to its
subject, and both appear in the ablative, as in (1).
(1)

his
pugnantibus
illum
in equum
quidam
. . . intulit
them:abl fight:ptcp.pres.abl him:acc in horse:acc someone:nom
mount:perf.3s
while they were fighting, one from his [attendants] mounted him on a horse. (Caes. Gal. 6.30)

Other ancient IE languages mark absolute constructions with different case (the genitive in Greek, the locative in Sanskrit).
The absolute construction is typically interpreted as an adjunct describing a temporal setting, but it can also introduce
cause, condition, concession, circumstance, etc.
In the dominant participle construction, the participle again agrees with a noun phrase corresponding to its subject, but
the construction can appear in any nominal syntactic function (subject, object, oblique argument, a prepositions argument,
genitive modifier); the syntactic function is directly reflected in the participles case (e.g., the accusative in (2) is governed
by the preposition). Despite its superficial similarity to a nominal attributive construction, the construction displays a
number of properties suggesting that it is headed by the participle (hence the term dominant participle), including event
interpretation (Bolkestein 1980; Haug & Nikitina 2012).
(2)

ante exactam
hiemem
before expel:ptcp.perf.pass.acc winter:acc
before the winter expired (Caes. Gal. 6.1)

The two exotic uses of the Latin participle are widely regarded as related; thus, Pinkster (1990: 117-118) describes the
absolute construction as nothing more than a manifestation of the dominant use in an adjunct function. On this view, the
ablative case of the absolute use is the same as the ablative case introducing nominal adjuncts with temporal meaning. The
dominant use, on the other hand, is often regarded as a historical quirk of Latin (Horrocks 2011), and has not received any
principled historical explanation.
The idea that the absolute use of the participle is merely a dominant use in an adjunct function does not fit the comparative
evidence. In particular, other ancient Indo-European languages make use of the absolute construction, but do not feature the
dominant use of the participle (in Ancient Greek, the dominant construction is only attested sporadically, and may be due
to Latin influence, cf. Jones 1939). This distribution suggests that dominant participles are in all likelihood a Latin-specific
innovation, while the absolute use should be reconstructed to PIE (Wackernagel 1920: 293; overview in Bauer 2000), even
though the specific case form does not seem to be reconstructible (Brugmann 1892, 1903, but cf. Holland 1986: 181). If the
absolute construction developed as an instance of the dominant use, it would be hard to explain why the latter is virtually
unattested outside Latin.
We present a historical account that reconciles the intuition that Latin dominant participle constructions are closely
related to the absolute construction with the comparative evidence pointing at the historical precedence of the absolute
use. Like Greek and Sanskrit, Proto-Latin most likely had an absolute construction: participial clauses could be used in an
adjunct function, and the case of the participle reflected the constructions adjunct status.
The participial adjunct may have developed due to a reanalysis of nominal adjuncts with attributive modifiers: [with]
ending winter (winter is the syntactic head) [with] winter ending (ending is the syntactic head), cf. somewhat
similar albeit different in detail accounts in Brugmann 1904: 609, Sommer 1931, Chantraine 1953). The same reanalysis
appears to have happened more than once in the history of some languages, resulting in the emergence at certain periods of
absolute constructions in non-canonical cases (such as the accusative and the nominative in Late Latin and in Greek; the
genitive in Sanskrit).
We suggest that in Latin, the absolute use of participial clauses in the ablative case led to a subsequent reanalysis of this
construction as a subtype of a nominal adjunct in which a participial clause functioned as a noun phrase (a similar scenario
is suggested informally in Heick 1936). This reanalysis led to an introduction of a nominalization rule that allows clauses
headed by participles to appear in nominal positions; this is precisely the rule that is responsible for the dominant participle
use in Latin (Haug & Nikitina 2012).
By identifying a mechanism of change that could result in the development of a dominant participle construction, our
account accommodates two seemingly contradictory observations implicit in previous research: (i) the intuition that Latin
dominant and absolute uses are instances of the same construction at the synchronic level, and (ii) the understanding that
the development of the dominant use involved some kind of change that was specific to Latin. Thus, our analysis allows us
to explain why the absolute construction is present in so many old IE languages, while its related dominant construction is
not.

A corpus-linguistic approach to serialisation changes in the Wackernagel chain

Patrizia Noel, University of Bamberg
patrizia.noel@uni-bamberg.de
The methodological problem in studies on the Wackernagel chain is threefold:
1) The elements taking part in the Wackernagel chain have been described by Wackernagel
(1892) for classical languages, but the correspondent elements in other languages are an open
question.
2) The elements involved have been characterized to be subject to language change (e.g.
Wackernagel 1892 and Hirt 1929: V 213); a law subject to language change, however,
cannot be described adequately if the members of the class are yet to be determined.
3) The order of Wackernagel elements is fixed in the chain (e.g. Wackernagel 1892 and Luraghi
1997: 62-63 for Hittite), but what is the cross-linguistic default serialisation in a chain of yet
to be determined members which are subject to language change?
The paper focuses on the development of the Wackernagel chain in German language history. The
results are based on an SQL data base with approx. 1.900.000 words and 190.000 sentences from
various genres and regions. The data base can be queried for lexical entries, word chains, and
positions.
The corpus-based approach to question (1) is to retrieve sentences with:
-

German words related etymologically to PIE Wackernagel elements and

German words which are etymologically unrelated but display similar functions and/or
German words which occur in second position and/or
German words which are adjacent to etymologically related and/or already established
Wackernagel elements.

Question (2) is answered 'automatically' by grouping the texts according to their date of origin and by
analysing output differences.
Question (3) is answered by the data base output of word order patterns in Wackernagel position as
well as by a semantic analysis of the elements.
The result of the analysis is that elements in Wackernagel position display features with similar
information structural function. The Wackernagel chain is both stable in its principles of serialisation
and dynamic in the change of lexical elements. Corpus linguistics thus contributes to diachronic
linguistics by offering the relevant data necessary for linguistic generalisations.

References
Hirt, Hermann (1929), Indogermanische Grammatik. Der Akzent. Heidelberg: Carl Winters
Universittsbuchhandlung.
Luraghi, Silvia (1997), Hittite. Munich: Lincom.
Wackernagel, Jacob (1892), ber ein Gesetz der indogermanischen Wortstellung. Indogermanische
Forschungen 1, 333436.

Parentheticals revisited
Jan Nuyts and Karolien Janssens
University of Antwerp

Topic: It is well known that mental state predicates such as think or know feature a parenthetical use, as in (1), next to
their use with a complement clause, as in (2).
(1) He is not coming to the party, I think.
(2) I think that he is not coming to the party.
Thompson and Mulac (1991) have postulated that the parenthetical use diachronically originates in the complementing
use, via a stage of complementizer omission. In this evolution the tie between the original main clause and the
complement is loosened and ultimately the former can be moved around freely in the latter. Since then, several
authors have found support for this hypothesis (e.g. Palander-Collin 1999, Apothloz 2003, Krkkinen 2003,
Tagliamonte & Smith 2005), but others have offered counterevidence for it (e.g. Brinton 1996, 2008, Fischer 2007, Van
Bogaert 2009). Brinton has suggested other source constructions for the parentheticals, such as appositional clauses of
the kind in (3), in which the deictic element eventually got deleted.
(3) He is not coming to the party. So/That I think.
In this paper we will try to throw more light on the debate, by means of a diachronic corpus investigation of four
mental state predicates in Dutch which show the grammatical alternation in (1)-(2): denken think, dunken think
(impersonal), geloven believe and vinden find.
Method: We investigate the grammatical structures and meanings of these predicates in corpora from four stages of
the language: Old Dutch, Middle Dutch, Early New Dutch and Present Day Dutch. We use samples of 200 instances per
verb per period (selected according to criteria such as representativity, e.g. in terms of text genres, and comparability
across the periods). For Present Day Dutch we have two separate samples of 200 instances, one written and one
spoken.
Results: For Dutch, Thompson and Mulacs hypothesis is problematic because of the different word order in main and
subordinate clauses and because of the near absence of complementizer omission. In fact, our diachronic data for the
four verbs show hardly any correlation between the evolutions in their complementing pattern and in their
parenthetical pattern. Instead, the latter shows clear historical correlations, on the one hand with appositional clauses
of the kind in (3), and on the other hand with a pattern with two juxtaposed main clauses of the kind in (4), i.e. the
pattern used for introducing quotes in Present Day Dutch.
(4) Ic gheloove hy zal noch slaghen ghenieten. I believe he will get hurt (Early New Dutch)
Our investigation thus provides empirical evidence for (some variant of) the hypotheses formulated by Brinton (1996,
2008) and Fischer (2007).
Apothloz, Denis. 2003. La rection dite 'faible': Grammaticalisation ou diffrentiel de grammaticit? Verbum 25: 241-262.
Bogaert, Julie Van. 2009. The grammar of complement-taking mental predicate constructions in present-day spoken British English. PhD dissertation,
University of Gent.
Brinton, Laurel J. 1996. Pragmatic markers in English. Berlin: Mouton.
Brinton, Laurel J. 2008. The comment clause in English. Cambridge: Cambridge University Press.
Fischer, Olga. 2007. Morpho-syntactic change. Oxford: Oxford University Press.
Jespersen, Otto. 1927. A modern English grammar on historical principles, Part III: Syntax. Second volume. Edition 1965. London: Allen & Unwin.
Krkkinen, Elise. 2003. Epistemic stance in English conversation. Amsterdam: Benjamins.
Palander-Collin, Minna. 1997. A medieval case of grammaticalization, methinks. In M. Rissanen, M. Kyt, K. Heikkonen (eds.), Grammaticalization at
work, 371-403. New York: De Gruyter.
Tagliamonte, Sali A. and Jenifer Smith. 2005. No momentary fancy! The zero complementizer in English dialects. English Language and Linguistics 9/2:
1-12.
Thompson, Sandra and Anthony Mulac. 1991. A quantitative perspective on the grammaticization of epistemic parentheticals in English. In E.
Traugott, B. Heine (eds.), Approaches to grammaticalization, vol. 2, 313-329. Amsterdam: Benjamins.

Grammatical variation as seen in ellipsis

Joanna Nykiel (University of Silesia)

English elliptical constructions allow two realizations of remnants which have
PPs as their correlates. For example, in the construction sluicing shown in (1),
the correlate is the PP on some meds, and the remnant may be either a NP (what
meds) or a PP (on what meds). However, speakers reasons for choosing one
realization over the other remain unexplored. In this talk, I argue that what
guides speakers choices is an interaction between processing pressures and the
development of multi-word verbs in English.

In support of this argument, I offer corpus data from Early Modern
English (EModE) through present-day English (PDE). The ratio of PP remnants to
NP remnants is 68.7% to 31.3% in EModE, but changes to 54.2% to 45.8% in
Late Modern English (LModE), and finally reverses by PDE (32.8% of remnants
are PPs and 67.2% are NPs). As long as PPs remnants are the majority, NPs
usually appear in the environment illustrated in (2), as opposed to (3). The
correlate in (2) hosts an NP (a package) and the correlate in (3) a pronoun
(something), and this difference translates into the correlate in (2) having a more
accessible mental representation, because it encodes more semantic and
syntactic information (see Ariel 1990, 2001, Hofmeister 2009). Since an ellipsis
remnant, as other anaphors, signals to the hearer how accessible the correlate is,
its realization may be manipulated by choosing a more explicit remnant (PP)
over a less explicit one (NP). Research on anaphora demonstrates that speakers
indeed lengthen or shorten anaphors as a way of coding the accessibility of the
antecedent (Brennan 1995, Fowler et al. 1997). This is a processing constraint
on the use of ellipsis remnants, but it interacts with the development of multi-
word verbs in English, triggering the later decrease in the use of PP remnants.

Claridge (2000) shows that one type of multi-word verbs (prepositional
verbs) first appears in Middle English, remains stable in EModE and increases in
frequency as of LModE. This pattern matches well the increase in the frequency
of NP remnants after EModE. The growing numbers of prepositional verbs affect
the choice of ellipsis remnants in that such verbs are semantically
noncompositional to varying degrees (Brinton and Traugott 2005), and hence
processed as wholes ([V+P]) rather than as separate constituents ([V +[PP]])
(see Hawkins 2004). The preference for NP remnants is explained if we assume
that the preposition which belongs semantically with the verb is not expressed in
the remnant, because the verb isnt, either. A statistical analysis of PDE corpus
data confirms that higher degrees of noncompositionality between prepositions
and verbs are correlated with the use of NP remnants.
I conclude that the once-predominant use of ellipsis remnants as a means
to code the accessibility of the correlates has been compromised by the gradual
development of prepositional verbs. This change has implications for the
typology of languages with respect to how ellipsis remnants with PP correlates
may be realized.
(1) Pats on some meds, but Im not sure what meds/on what meds.
(2) Pats waiting for a package, but Im not sure what package/for what package.
(3) Pats waiting for something, but Im not sure what/for what.

Kunna and Mughu: On the Use of Two Modal Verbs in Middle Danish
Rie Obe
Osaka University
Over the past few decades a considerable number of studies have been made on
semantic change of modal verbs, which is considered to be examples of
grammaticalization. But only few attempts have so far been made to describe semantic
change of Danish modal verbs. In particular, a semantic system of modal verbs in
Middle Danish seems to be lacking.
However, Anders Bjerrum (1966) made several important statements on modal verbs in
a Early Middle Danish law text, namely Sknske Lov B 74. In his study the paradigm of
modal verbs in this law text is considered to consist of two members, mughu (may /
must) and skulu (should). This is not surprising considering the fact a law text has a
quite limited context, i.e. it is primarily concerned with permission, prohibition, or duty.
Regarding kunna (can), Bjerrum gives a brief reference to a single example of kunna,
however, he apparently does not consider kunna a member of the paradigm of modal
verbs.
In contrast to Bjerrums study I have done research on how kunna (and other modal
verbs) is used in the following three texts in Middle Danish, Lucidarius, Sjlens Trst,
and Karl Magnus Krnike. They differ significantly with respect to genre in comparison
with a law text. In this paper I will propound the argument of that kunna should be
regarded a member of the paradigm of modal verbs in Middle Danish: not only do
kunna and mughu partly overlap with respect to function, but kunna is also encroaching
on some meanings of mughu.
References
Bjerrum, Andersen. 1966. Grammatik over Sknske Lov efter B74. Kbenhavn:
Gyldendal.
Hansen, Erik & Lars Heltoft. 2011. Grammatik over det Danske Sprog I-III. Det Danske
Sprog- og Litteraturselskab. Odense: Syddansk Universitetsforlag.
Obe, Rie. 2011. Modalverbernes semantiske system i gammeldansk. Ny forskning i
Grammatik 18: 249-265.

A unique word for hell: Estonian prgu

Vilja Oja, Sven-Erik Soosaar
Institute of the Estonian Language
In early Estonian Bible translations the Greek words hades and gehenna as well as the
Hebrew sheol have been translated as prgu or prguhaud (haud arch. hole). In the 20th
century the expression surmavald, lit. realm of Death has been used instead of prgu in
some contexts. It has been supposed that prgu is a very old word, known perhaps even in
Finnic-Mordvin protolanguage. Words with the similar phonetic shape and lexical meaning
are unknown in other languages. The paper analyses the semantics of the Estonian noun
prgu, compares it with the terms used in the original Bible texts as well as with its
equivalents in closely related and contact languages. As a result, a new conception of the
origin of the word prgu is presented.
The rest of the Finnic languages and dialects express the concept hell with a
loanword from an Indo-European language: Finnish, Karelian and Votic helvetti < Old
Swedish helvete, Livonian e < Latvian elle or Middle Low German helle (<< Germanic);
Veps, Votic, Ingrian, Karelian and Finnish ad, aadu etc. < Old Russian < Greek
(ALFE 1: 193195; SSA 1: 155; Vasmer 1986 1: 61). The use of prgu in Estonian dialects,
place names and early written sources indicates that the word was introduced via spiritual
texts. In the vernacular use prgu designates a mythological underworld. Numerous folk tales
and place names are based on this interpretation.
Estonian prgu is most probably a loanword. Despite several etymological hypotheses
there is still no generally accepted solution (EES: 400). Spiritual Estonian has largely been
influenced by Latin, which was used in the Roman Catholic Church. Latin inferi inhabitants
of Hades or inferna place of punishment for demons and sinners' souls are not acceptable
borrowing sources of prgu for phonetic reasons. However it is not impossible that the
Estonian prgu might originate in the Latin stem prg (earlier prig), -re to clean.
References
ALFE

1 = Atlas Linguarum Fennicarum. Itmerensuomalainen kielikartasto.

Lnemeresoome keeleatlas. Ostseefinnischer Sprachatlas.
- . ALFE 1. Eds. Tuomo Tuomi, Seppo Suhonen.
Suomalaisen Kirjallisuuden Seuran Toimituksia 800. Kotimaisten kielten
tutkimuskeskuksen julkaisuja 118. Helsinki: Suomalaisen Kirjallisuuden Seura.
Kotimaisten kielten tutkimuskeskus 2004.
EES = Metsmgi, Iris; Meeli Sedrik; Sven-Erik Soosaar 2012. Eesti etmoloogiasnaraamat.
Eesti Keele Instituut. Tallinn: Eesti Keele Sihtasutus.
SSA = Suomen sanojen alkuper. Etymologinen sanakirja. Suomalaisen Kirjallisuuden
Seuran toimituksia 556. Kotimaisten kielten tutkimuskeskuksen julkaisuja 62.
Helsinki 19922000: Suomalaisen Kirjallisuuden Seura, Kotimaisten kielten
tutkimuskeskus.
Vasmer = , 1986. .
. . . . .
. , . . : .

The Problem of Language Transmission in Genetic Linguistics and Language

Classification
Natalie Operstein
Until fairly recently, the accepted view in genetic linguistics and language classification has
been that the transmission of a language is normal only when it is acquired by children as
their first language from their elders. The necessary corollary to this view has been that
languages that fail to pass the criterion of normal transmission in other words, languages
whose past histories contain significant amounts of nonnative acquisition also fail to fit
within the genetic classification scheme. Although these assumptions have underlain most of
historical linguistic writing and discussions of language classification, they have been rarely
articulated explicitly. A notable exception is Thomason and Kaufmans (1988: 9ff) extended
discussion of the issue, in which the above assumption and its corollary occupy the third and
the fourth place in the list of fundamental theoretical assumptions that underlie the concept
of genetic relationship:
Third, a language is passed on from parent generation to child generation and/or via
peer group from immediately older to immediately younger, with relatively small
degrees of change over the short run, given a reasonably stable sociolinguistic context.
() Our fourth assumption is that the label genetic relationship does not properly
apply when transmission is imperfect.
The theoretical importance of the concept of normal transmission for that of genetic
relationship is further elaborated in the following quote from page 12:
Our approach to the study of genetic relationship is thus based theoretically on the
social fact of normal transmission rather than merely on the linguistic facts themselves.
Alongside this approach within what Noonan (2010: 48) refers to as orthodox linguistic
circles, there has been growing awareness of the existence of a large number of linguistic
varieties and language transmission situations that do not fit the strict definition of normal
transmission and genetic relationship outlined in the above quotes (cf. Mufwene 1997,
DeGraff 2003, Owens 2006, McWhorter 2007, Bossong 2009). The question of the genetic
affiliation of language varieties that come into being in situations of less-than-perfect intergenerational transmission naturally arises, but is rarely explicitly discussed.
The purpose of this paper is to discuss several types of such situations and to outline a
possible approach to genetic linguistics that naturally integrates L1 and L2 acquisition at the
level of language transmission. The proposed approach builds significantly on Dixon (1997)
and incorporates recent advances in the areas of contact and sociohistorical linguistics (e.g.
Siegel 1997, Mufwene 2001, Holm 2004, Labov 2007), historical dialectology (e.g. Trudgill
2004, Kerswill and Trudgill 2005), sociolinguistic typology (e.g. Trudgill 2011), and societal
effects on language complexity (e.g. Kusters 2003).

Constructional change and schematization: grammaticalization of an adverbial locution in

Spanish
International Conference on Historical Linguistics 21, Oslo
Rosa Mara Ortiz Ciscomani, Universidad de Sonora, Mxico
ortizrm@capomo.uson.mx; ortizciscomanirm@gmail.com
It is a well-known statement that language change originates in language use and that it involves
categorization, that is, the speaker-hearer generalizes over thousands of instances (Bybee 2010) to
create schemas (Goldberg 1995), which are embodied or represented in language system. In Spanish,
there is a prepositional phrase with a lexical item, noun or adjective, with an ending as, whose
original referential meaning became semantically opaque, and whose occurrences increased in types
and tokens diachronically (1)-(3).
(1) Et fuese el fijo del labrador, et fizo lea, et trxola a cuestas (Calila e Dimna, 1251, CORDE)
and brought-it (the fire-wood) on (his) shoulder or back
(2) Dixo Moysen al Nuestro Sennor: Non puedo a solas levar esta carga (Almerich, La fazienda de
Ultramar, c. 1200, CORDE).
Not (I) can unaided/by myself to carry this load
(3) Amigos, cada uno de vosotros es puesto en tan grand peligro como es ver la muerte
Each one of you
is left
in so big problem like is to see the death
a ojos vistas, e non la poder escusar (Crnica del rey don Rodrigo, 1430, CORDE)
visibly/face to face,
In contrast with (1) with a cuestas with referential meaning, a solas, with the same ending as in (2),
has abstract-modal meaning and complexity concerning syntactical behavior, that is, the presence of
the mentioned ending as in an adjective without a noun agreeing to; the modal expression a ojos
vistas, in (3), enhances complexity as combining the adjective vistas with feminine ending and the
masculine noun ojos. Synchronically, ending as, in (2) and (3) specially, is not clear. In this paper, I
will show how an adverbial modal productive scheme with an increasing amount of adjectives and
nouns arose from the original prepositional phrase, by increasing semantic and syntactic
complexity, as of the 15th century.
From a large diachronic corpus (12th to 20th centuries), I claim that Spanish displays evidence of a
change involving a process of schematization that is a sample of constructional grammaticalization.

References:
Bybee Joan. 2010. Language, Usage and Cognition, Cambridge: Cambridge University Press.
Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to Argument Structure,
Chicago: The University of Chicago Press.

The grammaticalization of an inflectional ending: towards a formal theory of grammaticalization

Name: Fuyo Osawa
Affiliation: Hosei University
I claim that there are two kinds of elements which contributed to the grammaticalization. One is a primary
contributor which triggered the change, while the others are grammaticalized thanks to the primary
contributor. This concerns the syntactic nature of grammaticalization, that is, grammaticalization is driven
structurally as discussed by Gelderen (2004). The Present-day English determiner the is supposed to have
developed from Old English demonstrative se or seo. This is assumed to be one instantiation of
grammaticalization. However, in Present-day English, according to the DP analysis (Abney 1987), in addition
to article the, a, D head includes possessive determiner my, your, quantifiers every and the clitic -s'. The
question arises: how these D heads or more precisely their precursors in Old English contributed to the
grammaticalization. Did they contribute to the process equally?
I propose that the precursor of -s', i.e. the Old English genitive case ending es played a primary role in
the rise of a D system (cf. Rosenbach 2002, Hopper and Traugott 2003, Osawa 2003, 2007). The genitive
case ending es went first and other precursors were followers. I define the term grammaticalization in a
way different from the most widely accepted one, which is defined in Hopper and Traugott (2003) as a
process by which lexical words change into function words, and sometimes further into bound morphemes.
This is part of grammaticalization process; however, grammaticalization could be viewed as the emergence
of functional categories heading their own projection in the clause structure. This means that a space/position
is created in the clause structure. This concept of grammaticalization can solve the puzzle which has been
discussed by many researchers (Janda 1980, Allen 1997, 2008): if -s' as a D head derived from the Old
English genitive case ending es, this goes against the usual grammaticalization path from syntax to
morphology (Givn 1971) since this is a shift from morphology to syntax. In my view of grammaticalization,
this shift from morphology to syntax is indeed possible. The es ending was first grammaticalized to be a D
head in an adnominal phrase, since es ending is more suitable for a functional category whose properties are
described by Abney (1987) than the free morpheme se or seo. Once a space, i.e. an independent projection DP,
is established in the clause structure, other elements can take place in the space, or can move into the place
out of somewhere else and they undergo grammaticalization. The free morpheme such as se was
grammaticalized afterwards. I propose that grammaticalization is the creation of a new space/position rather
than the creation of new grammatical materials. Semantic bleaching, which has been considered as a
characteristic of grammaticalization, should be called into question.
Finally I claim that the analysis presented in this paper is also applied to the grammaticalization of main
verbs as auxiliaries in the history of English. A TP emergence is involved in this case.
References
Abney, Steven P. (1987) The English Noun Phrase in its Sentential Aspect. Doctoral dissertation, MIT.
Allen, Cynthia L. (1997). The Origins of Group Genitive in English, Transaction of the Philological Society 95: 1.111
-131.
Allen, Cynthia L. (2008) Genitives in Early English: Typology and Evidence. Oxford: OUP.
Gelderen, Elly van (2004) Grammaticalization as Economy. Amsterdam: John Benjamins.
Givn, Talmy. (1971) Historical Syntax and Diachronic Morphology: An Archaeologists Field Trip. CLS 7.394-415.
Hopper, Paul J. and Elizabeth C. Traugott. (2003) Grammaticalization 2nd, Cambridge: CUP.
Janda, Richard D. (1981) A Case of Liberation from Morphology into Syntax: The Fate of the English Genitive Marker
(e)s, Syntactic Change, ed. by Brenda B. Johns and David R. Strong, 60-114. Ann Arbor, MI: Department of
Linguistics, University of Michigan.
Osawa, Fuyo. (2003) Syntactic Parallels between Ontogeny and Phylogeny. Lingua 113:1.3-47.
Osawa, Fuyo. (2007) The Emergence of DP from a Perspective of Ontogeny and Phylogeny: Correlation between DP, TP
and Aspect in Old English and First Language Acquisition. Nominal Determination: Typology, Context Constraints,
and Historical Emergence. ed. by Elisabeth Stark, et al., 311-337, Amsterdam: John Benjamins.
Rosenbach, Anette. (2002) Genitive Variation in English: Conceptual Factors in Synchronic and Diachronic Studies,
Berlin: Mouton de Gruyter.

Veturlii G. skarsson, Uppsala University

Loanwords with the prefix be- in Icelandic:

An investigation of 19th century private letters
The Icelandic language got its share of loanwords of Low German, and later High German, origin
through Norwegian until the middle of the 15th century and after that through Danish. However, the
number of such loanwords, as well as of loanwords in general, is much smaller in Modern Icelandic
than in the mainland Scandinavian languages. The most important historical explanation to this is
probably the distance from the outside world, with a limited contact with foreigners through the centuries. The opposition in the 19th and 20th centuries against foreign influence on the language played
also a major role for the situation of today. It was a part of Icelands struggle for independence, where
an attempt was made to free the language from younger loanwords of Danish-German origin in order
to approach the language of the golden age of the 13th century.
The study presented in this paper focuses on a subgroup of such loanwords in Icelandic, i.e. words
with the prefix (preformative) be- (e.g. betala pay). It is based on a text corpus with more than 1,600
unpublished private letters from lay people in 19th century Iceland (ca. 18201900), and partly on
examination of various other Icelandic texts and material from the University of Iceland Dictionary, in
order to view the subject from a broader time perspective (ca. 14001900). The letter corpus is a part
of the research project Language Change and Linguistic Variation in 19th-Century Icelandic and the
Emergence of a National Standard (see: http://arnastofnun.is/page/LCLV19_project). The material is
approached from the point of view of lexicography, standardization, and sociolinguistics, and, when
applicable, applied folk linguistics. The words, their age in the language, their number, frequency, use,
and their fate in the aforementioned struggle against Danish-German loanwords, is examined. Contextual factors will also be documented, such as possible cases of code-switching, expressions indicating conscious (stylistic, humiliating etc.) use of certain words, and meta-linguistic comments about
individual words or loanwords on the whole, or on the use of or directions on the use of such words.
Information about similar factors regarding Icelandic loanwords with the German prefix an- and a
number of other Danish-German loanwords, e.g. ske happen and blfa be, remain, will be used to
shed light on the subject. It is well knownthough only partially investigated statisticallythat the
number of such loanwords decreased in Icelandic during the 19th and 20th century. Thus, there are
hardly any words with an- and be- in modern Icelandic, blfa is only usable in a very marked context,
and ske is highly stigmatized. (See e.g. skarsson 199798; 2000; 2007; 2009.)
In the study, an attempt will be made to interpret the results in the light of 19th century cultural
politics, including discussion of some of the following questions: Are there many or few examples of
the examined words to be found in texts, and are texts of lay people different from other texts in this
respect; do such words seem to be neutral or marked in their usage; do they reduce in number during
the course of the 19th century, and if so, when does that process seem to begin? Or is their number
perhaps fairly even throughout the 19th centurywhich would suggest that the opposition against
them was rather a 20th century phenomenon than one belonging the century before. And what can the
results reveal about the vocabulary and language of lay people in the 19th century? It is not certain
how the opposition to loanwords was passed on to the public, by whom and what the results were: Did
these words disappear more slowly in the language of ordinary peopleor was it the opposite: Were
these words maybe fewer from the beginning in the language of lay people than in that of intellectuals
and officials?
Cited works:
skarsson, V. 199798. Ske. slenskt ml 1920: 181207.
__ 2000. Verbet islndskt sk. Scripta Islandica 50, 3149.
__ 2007. Um ska forskeyti an- og stutta vidvl ess slensku. Or og tunga 9, 125149.
__ 2009. Um sgnina blfa, vxt hennar og vigang slensku. slenskt ml 31, 189224.

Veturlii G. skarsson
professor of Nordic languages
Uppsala University
veturlidi.oskarsson@nordiska.uu.se
tel. +46 18 471 1280

Intransitive Subject Genitive in Old Basque: a Trace of Syntactic Ergativity?

Manuel Padilla-Moyano (University of the Basque Country - IKER UMR5478)
The genitivization of O (TOGEN henceforth) in nominalised clauses has been described in
northern varieties of Basque (Heath 1972). TOGEN has traditionally been one of the features used to
demonstrate that Basque syntax is accusative (Oyharabal 1992; Trask 2002), because of the varieties
exhibiting this type of genitivization mark O, and treat both transitive (A) and intransitive (S) subjects
similarly.
In the context of the discussion about the typological situation of Basque syntax, Bossong
(1984) attempted to prove that also the genitive marking of S would be possible however, he started
from a misunderstanding of Basque data. As a result of a far-reaching search in old northern and
eastern texts, I am now in a position to provide solid evidence of genitive marking of S in nominalised
clauses in the past. In fact, intransitive subject genitives (ISGEN) are present in a range of eastern
Basque texts as late as in the19th century. In most of the cases, the genitivization rule systematically
assimilates S (1) and O (2), whilst A (2) always appears in the ergative:
(1)
(2)

Hanitz dz, sira, zu-re

huna
ji-tia.
a lot be.3sg Sir
you-GEN
here.ADL
come-NMLZ.DET
It is great that you come here.
Ingrat
lkez ni-k
zu-re
refsa-tzia.
ungrateful
be.3sg I-ERG you-GEN
reject-NMLZ.DET
It would be ungrateful that I reject you.

Although genitive marking of subjects in subordinate clauses has been described in Altaic
languages (Kornfilt 1997; Miyagawa 2008; Asarina & Hartman 2010), for a parallel of Basque it is
more pertinent to look to the history of English gerund. Indeed, in both languages a former de-verbal
name has been gaining verbal properties, becoming a gerund at some point. This process would not
have been completed yet, with genitive objects (in northern Basque) and genitive subjects (in English)
still being the norm (Trask 1995). In my opinion, Trasks hypothesis can also explain the ISGEN now
attested in old eastern Basque, even if its author did not know about this phenomenon.
Thus, the aim of this paper is to demonstrate that in subordinate clauses Basque once treated
both O and S in the same manner, i.e. in ergative alignment. At this point, ISGEN has two possible
explanations: a) it would be an archaism preserved in eastern Basque longer than elsewhere, as a
vestige of an ancient general rule; b) it would represent a non-successful innovation of the eastern
dialects, and in this case it would not have existed in pre-dialectal common Basque. Even though there
is no decisive evidence for any of these, I will argue in favour of the former explanation.
References
Asarina, A. & Hartman, J., 2011, Genitive Subject Licensing in Uyghur Subordinate Clauses, in
Proceedings of the 7th Workshop on Altaic Formal Linguistics (WAFL 7), A. Simpson (ed.),
MITWPL, Massachusetts.
Bossong, G., 1984, Ergativity in Basque, Linguistics 22: 3, 341392.
Heath, J., 1972, Genitivization in Northern Basque Complement Clauses, International Journal of
Basque Linguistics and Philology 6, 4666.
Kornfilt, J., 1997, Turkish, Routledge, London.
Miyagawa, S., 2008, Genitive subjects in Altaic, in Proceedings of the 4th Workshop on Altaic
Formal Linguistics (WAFL 4), C. Boeckx & S. Ulutas (eds.), 181198. MITWPL,
Massachusetts.
Oyharabal, B., 1992, Structural case and inherent case marking: Ergaccusativity in Basque, in
Syntactic theory and Basque syntax, J. Lakarra & J. Ortiz de Urbina (eds.), 309342.
University of the Basque Country, San Sebastian.
Trask, R., 1995, On the History of the Non-Finite Verb Forms in Basque, in Towards a History of
Basque Language, J. I. Hualde, J. Lakarra & L. Trask (eds.), 207234. John Benjamins,
Amsterdam.
Trask, R., 2002. Ergativity and accusativity in Basque, in The Nominative & Accusative and their
counterparts, K. Davidse & B. Lamiroy (eds.), 265284. John Benjamins, Amsterdam.

The Importance of Being a Process:

the Aramaic definite article as a test case
Naama Pat-El, The University of Texas, Austin
The development of the definite article from an attributive demonstrative is a classic
Grammaticalization scenario. Typically, following Greenberg (1978), the process is described
as a one-step path: Stage 0: demonstrative > stage 1: article (Diessel 1999; de Mulder and
Carlier 2010). The pathway DEM>DEF has been amply attested in world languages, and a
similar analysis was made regarding some Semitic developments, one of which is the NeoAramaic definite article, for example uroyo aq-qae the priests (aC < *hannen cp). Many
studies offer lengthy explanations as to how a demonstrative becomes a definite article, in
terms of its semantic/pragmatic properties. Such explanations suggest that there is a context in
which Grammaticalization processes operate (Heine 2002). However, this context is localized
in the particular change from demonstratives to definite articles, as if this change is detached
from a larger linguistic system, where motivation for the rise of definite marking may be. In
this paper I argue that a pathway DEM>DEF is only true if we look at the origin and end of
the process, and ignore both the process itself and various other changes in the history of the
language, Aramaic in this case, which contributed to the development of the article. I further
argue that this is an essential problem for Grammaticalization: by focusing on two arbitrary
endpoints, it is implied that there is nothing very interesting or relevant between them.

Diessel, Holger (1999). Demonstratives: Form, Function and Grammaticalization. Amsterdam: Benjamins.
Greenberg, Joseph H. (1978). How Does a Language Acquire Gender Markers? In: Universals of Human
Language. Ed. by J. H. Greenberg. Vol. 3. Stanford: Stanford University Press, 4882.
Heine, Bernd (2002). On the role of context in grammaticalization. In: Typological Studies in Language. Ed.
by G. Diewald and I. Wischer. John Banjamins, 83102.
Mulder, Walter de and Anne Carlier (2010). The Grammaticalization of Definite Articles. In: The Oxford
Handbook of Grammaticalziation. Ed. by H. Narrog and B. Heine. Oxford University Press. 522534.

Assimilation vs. Boundary Marking in German un-Derivatives

Anna-Marleen Pessara (Johannes Gutenberg-Universitt Mainz)
An unusual phenomenon is manifesting itself in present-day German morphophonology: while
assimilation normally stops short of spreading across morpheme boundaries (cf. e.g. Giegerich 1985),
a subset of un-derivatives does precisely that. For example, a Google query for unbedingt absolutely
(mis)spelled as umbedingt returns almost a million hits for .de top-level domains alone.1 A closer look
at the relevant websites reveals that the vast majority of these hits come from blogs, message boards,
and online forums, i.e. from computer-mediated communication (CMC), which constitutes a mode of
discourse in between spoken and written language (cf. e.g. Drscheid 2004). Hence the spelling of
umbedingt is phonetically motivated in that it reflects the phonetic process of turning [n] into [m] by
regressive assimilation of the bilabial feature, thus blurring the morpheme boundary in un+bedingt.
This blurring is particularly puzzling considering that German has changed from a syllable
language (OHG) to a word language (MHG>ENHG>NHG), thereby developing a strong tendency to
mark boundaries, which becomes evident in many different respects. As concerns the lexical level, for
instance, the Early New High German development of genitive constructions into genitive compounds
involves the reanalysis of genitive inflections as linking elements that come to mark the first
constituents right edge (e.g. [[desDet TeufelsN]NP Sohn]NP > [desDet [Teufel-s Sohn]N]NP >
[derDet [Teufelssohn] N]NP the devilsson). Among present-day Germans variety of linking
elements counting at least even the linking -s- stands out for it spread from its original paradigm to
other declension classes, such as the feminine declension (e.g. Anfahrtsskizze directions sketch),
thereby becoming the most productive linking element and present-day Germans finest example for
lexical boundary marking (cf. Nbling & Szczepaniak 2008). Nominal framing (e.g.
das.NOM.NEUT.SG
von
vielen
Schwierigkeiten
habenden
Studenten
boykottierte
Seminar.NOM.NEUT.SG the seminar which was boycotted by many students who are having
difficulties) presents a syntactic instance of boundary marking, illustrating that German boundary
marking is not only a lexical phenomenon (cf. Ronneberger-Sibold 2010). Even among un-derivatives
and thus on the morphological level, boundary marking seems to occur. In these cases, assimilation
across morpheme boundaries is blocked (e.g. u[n]bunt colorless), as is resyllabification (e.g.
u[n]interessant uninteresting), keeping the boundary between prefix and base intact. Why is it,
though, that two opposing mechanisms assimilation and boundary marking can and do occur in the
same context, i.e. prefixation with un- ?
This paper contributes to identifying which determinants trigger assimilation and which
preserve boundary marking in un-derivatives, assuming this instance of synchronic variation to
indicate language change. More precisely, I predict assimilation to reflect loss of semantic
transparency through lexicalization/demorphologization: Highly frequent and highly lexicalized unwords such as unbedingt ( nicht bedingt not conditional), will allow for assimilation while less
frequent and more semantically transparent ones, such as unbunt (= nicht bunt not colorful) or
uninteressant (= nicht interessant not interesting), will not but preserve boundary marking.
Explorative analyses of historical and contemporary spoken corpora (e.g. GerManC2, DWDS3,
DGD24), as well as a self-compiled CMC-webcorpus will establish to what extent Germans
pronounced predilection for boundary marking as evident on most linguistic levels may be
counteracted at the level of morphemes.
References:
Drscheid, Christa (2004): Einfhrung in die Schriftlinguistik. 2., berarbeitete Auflage. Wiesbaden: VS Verlag
(= Studienbcher zur Linguistik 8).
Giegerich, Heinz (1985): Metrical phonology and phonological structure: German and English. Cambdrige:
Cambridge University Press (= Cambridge Studies in Linguistics; 43).
Nbling, Damaris/Szczepaniak, Renata (2008): On the way from morphology to phonology: German linking
elements and the role of the phonological word. In: Morphology 18, 1-25.
Ronneberger-Sibold, Elke (2010): Der Numerus das Genus die Klammer. Die Entstehung der deutschen
Nominalklammer im innergermanischen Vergleich. In: Dammel, A., Krschner, S., Nbling, D. (eds.):
Kontrastive Germanistische Linguistik. Teilband 2. Hildesheim u.a.: Olms (=Germanistische Linguistik
206-209), 719-748.
1

http://www.google.de/search?hl=de&as_q=umbedingt&as_epq=&as_oq=&as_eq=&as_nlo=&as_nhi=&lr=lang
_de&cr=&as_qdr=all&as_sitesearch=.de&as_occt=&s
afe=images&as_filetype=&as_rights=nicht+nach+Lizenz+gefiltert, accessed on 1/28/2013
2
http://www.llc.manchester.ac.uk/research/projects/germanc/
3
http://www.dwds.de/ressourcen/korpora/
4
http://dgd.ids-mannheim.de:8080/dgd/pragdb.dgd_extern.welcome

Chance, fortune, or was it something else that happened?

Syntactic change and functional redistribution in happen-constructions
Peter Petr University of Leuven (KU Leuven)
In this talk, I relate longitudinal variation and change in complementation structures within the socalled happen-constructions (as in (i)) to a broader, typological shift in English from foreground- to
more background-marking. I do so for the period 1051-1640 in LEON 0.3 (Petr 2013).
(i) with that-complementizer: It happened (befell/fortuned/chanced/...) that he came by the
place... & the women ... came out with staues ... & made an ende of hym (1528)
(i) shows the typical narrative function of happen-constructions of marking an episode boundary and
grounding the ensuing episode (the women killing the man) by introducing in the embedded clause its
instigating event (the mans passing by) (Brinton 1996: 115-80).
Next to (i), which was the most frequent one originally, constructions (ii)-(iv) occurred. Note
that in (i)-(iii), but not in (iv), the happen-verb is impersonal (formally lacking a referential subject).
(ii) with -complementizer: And it happened in those dayes, Iesus came to Nazareth. (1538)
(iii) with oblique experiencer + to-infinitive: And at the laste by fortune hym happynd ayenste
nyght to come to a fayre courtelage ... (a1470)
(iv) with raised subject-experiencer + to-infinitive: And whyles he rode about the fielde ...
he hapned to come nere a company of Romaynes ... (1544)
From late Middle English onwards, (i)-(iv) are subject to a gradual functional redistribution. It is
argued that their respective grammatical structures are to blame.
Firstly, impersonal constructions came under pressure due to the obligatorification of subjecttopic alignment. Conversely, to-infinitival complements with raised subjects increased, as they
effectively aligned topics to main clause subjects (Los 2007). Yet constructions (iii)-(iv) gradually
shifted their preference to non-narrative uses, becoming semantically equivalent to accidentally. This
shift can be related to word order iconicity. In (i), (ii), and some variants of (iii), the happen-verb
precedes the entire embedded event, signalling a contingency relation with what preceded. The subject
may still be in control of the instigating event. In (iii) (mostly) and in (iv), however, the happen-verb
comes in between the subject and the predicate of the event, and the verbs contingency semantics now
applies to the relation between the subject and what it is doing (cf. by fortune in (iii)).
Secondly, the obligatorification of topical subjects forms part of a wider typological shift (Los
2012). Old English grammar segmented narrative as a series of foregrounded events via topic time
adverbs (typically a then) plus subject-verb-inversion in main clauses. Episode introductions,
situated in between background and foreground, received a mixed encoding of semantically empty
main happen-clause plus embedded clause. Part of the typological shift starting in Middle English
involved the development of syntactic background marking via increased subordination, which
became also used for the framing of new episodes (e.g. when Jesus was still a child, he came to
Nazareth), and which made main happen-clauses superfluous.
In sum, the reorganization of happen-constructions suggests that that/to/ZERO -variation is
probably equally determined by global typological shifts or tensions in English as well as more
finegrained factors limited to coherent constructional subsets.
References
Brinton, L. J. 1996. Pragmatic Markers in English. Grammaticalization and Discourse Functions.
Berlin-New York: Mouton de Gruyter.
Los, Bettelou. 2007. The rise of the to-infinitive. Oxford: Oxford University Press.
Los, Bettelou. 2012. The loss of verb-second and the switch from bounded to unbounded systems. In
Anneli Meurman-Solin et al. (eds.), Information structure and syntactic change in the history of
English, 21-46. Oxford: Oxford University Press.
Petr, Peter. 2013. LEON: Leuven English Old to New, version 0.3. Leuven: Department of
Linguistics. (https://lirias.kuleuven.be/handle/123456789/396725).

Time for change

FRANS PLANK
Universitt Konstanz
frans.plank@uni-konstanz.de

In order to get an angle on deep-time historical relationships between languages

(beyond what can be fathomed by the comparative method) and in order to model the
evolution of typological diversity, attention is increasingly being paid to the question of
time stability, of lexical as well as of grammatical traits. Most of this work has been
inferential: without studying diachrony directly, crosslinguistic distributions are being
interpreted as revealing how stable or unstable particular traits are. The conclusions
that have been reached about time stability in this indirect way are alarmingly
contradictory, despite all statistical sophistication. Taking seriously the early criticism
of Greenberg (1993) that "the basic fallacy [...] is the notion that we can use statistics
concerning the relative frequencies of typological features in different areas to
reconstruct remote prehistory", and that "it is rather the distribution of such typological
features [...] that itself requires historical explanation", I suggest that this research
programme be reoriented and that time stability be studied directly, i.e. diachronically.
The particular issue I would like to address here is the tempo of change: a trait will
appear relatively time-stable if the tempo of changes affecting it is slow. To the extent
that this matter is addressed at all, the literature again is remarkably contradictory, with
different authors asserting that the tempo of changes of the same kind is uniform (e.g.,
Kroch 2001) or can randomly be rapid or slow (e.g., Traugott & Trousdale 2010).
In this paper I will be looking in detail at a particular grammaticalisation development,
the reanalysis of a noun 'house, home' as a local or possessive adposition 'at' or 'of', with
the aims (i) of determining the amount of time this kind of change (or rather complex of
micro-changes) takes and (ii) of comparing its tempo across a few languages where it
has occurred. The best documented instances are French chez 'at' from Late Latin
casa/chis, Swedish/Danish/Norwegian hos 'at' from Eastern Old Norse hus, and late
Pli g 'of' from Prakritic Indo-Aryan geha, LOC.SG gehi. (The time profile for an
analogous development in Akkadian is difficult to ascertain.) All three occurrences
have indeed taken about the same amount of time to come to completion (in Sinhala and
Maldivian the adposition g would further, over quite a long time, change to a
suffix): approximately 400 years, or some 16 generations, 16 cycles of acquisition. We
conclude that grammaticalisation of this kind is very slow, but indeed seems to proceed
at a uniform tempo, and we raise the question what it is about change that can take more
or less time (answer: the complexity of micro-changes and social diffusion).

The Historical, Biological, and Developmental Dynamics of Language:

Converging Perspectives on Language Evolution and Change
Michael Pleyer, Heidelberg University; Stefan Hartmann, University of Mainz
Inquiries into the 'curious parallels' (Darwin 1871) of language change and biological evolution have
a long tradition (Atkinson & Gray 2005) and have gained increasing popularity in the past few decades
(e.g. Croft 2000; Ritt 2004). In a similar vein, recent studies on the evolution of language describe
language as a complex adaptive system (CAS, e.g. Beckner et al. 2009), emphasizing the dynamic
interaction of processes taking place at the biological, developmental, and historical/cultural timescales, respectively (Kirby 2012).
While language evolution researchers have thoroughly embraced insights from Historical Linguistics
most notably, from grammaticalization studies (e.g. Heine & Kuteva 2012) , many Historical Linguists regard insights from Evolutionary Linguistics somewhat sceptical or even explicitly exclude this
area of research from the scope of Historical-Linguistic inquiry (e.g. Campbell 2013). However, recent
cognitively oriented approaches in Historical Linguistics explicitly acknowledge that important insights can be gained from studies adopting a CAS view of language and its evolution (e.g. Frank &
Gontier 2010).
This talk addresses the question how the approaches of Historical Linguistics and language evolution
research can benefit each other in explaining dynamic aspects of language. Most importantly, converging evidence from both areas can shed light on the cognitive dimension of language and linguistic
change. The notion of construal, which lies at the heart of much Cognitive-Linguistic research (e.g.
Croft & Cruse 2004), suggests itself as a key subject matter in both areas that can only be properly
understood if all three of the aforementioned timescales are taken into account.
The relevance of construal as an integrative factor at the interface of ontogeny (Tomasello 2003), evolution (Pleyer 2012), and historical language change (Hartmann in press) is illustrated with corpus
analyses of both historical language data and language acquisition data. As will be shown, the CAS
approach provides a valuable framework allowing for an integration of multiple perspectives and lines
of evidence in acknowledging that cognitive, cultural, and biological factors are inextricably intertwined both in the origins of language and in its evolution over historical time. We conclude that the
dialogue between Historical and Evolutionary Linguistics can provide fruitful new perspectives on
language change as well as open up new avenues of research in a highly interdisciplinary paradigm.
References
Atkinson, Q. D.; Gray, R. D. (2005): Curious Parallels and Curious Connections. Phylogenetic Thinking in Biology and
Historical Linguistics. In: Systematic Biology 54, 513526.
Beckner, C. et al. (2009): Language is a Complex Adaptive System. Position Paper. In: Language Learning 59 Suppl. 1, 1-26.
Campbell, L. (2013): Historical Linguistics. An Introduction. 3rd ed. Cambridge: MIT Press.
Croft, W. (2000): Explaining Language Change. An Evolutionary Approach. Harlow: Pearson Longman.
Croft, W.; Cruse, A. (2004): Cognitive Linguistics. Cambridge: Cambridge University Press.
Darwin, C. (1871): The Descent of Man. London: Murray.
Frank, R. M.; Gontier, N. (2010): On Constructing a Research Model for Historical Cognitive Linguistics (HCL). Some
Theoretical Considerations. In: Winters, M. et al. (eds.): Historical Cognitive Linguistics. Berlin, New York: De Gruyter,
3169.
Hartmann, S. (in press): The Diachronic Change of German Nominalization Patterns: An Increase in Prototypicality. To
appear in: Selected Papers from the 4th UK Cognitive Linguistics Conference.
Heine, B.; Kuteva, T. (2012): Grammaticalization Theory as a Tool for Reconstructing Language Evolution. In: Tallerman,
M.; Gibson, K. R. (eds.): The Oxford Handbook of Language Evolution. Oxford: Oxford University Press, 511527.
Kirby, S. (2012): Language is an Adaptive System. The Role of Cultural Evolution in the Origins of Structure. In: Tallerman,
M.; Gibson, K. R. (eds.): The Oxford Handbook of Language Evolution. Oxford: Oxford University Press, 589604.
Pleyer, M. (2012): Cognitive Construal, Mental Spaces and the Evolution of Language and Cognition. In: Scott-Phillips,
Thomas C.et al. (eds.): The Evolution of Language. Proceedings of the 9th International Conference. Singapore: World
Scientific, 288295.
Ritt, N. (2004): Selfish Sounds and Linguistic Evolution. A Darwinian Approach to Language Change. Cambridge: Cambridge University Press.
Tomasello, M. (2003): Constructing a Language. A Usage-Based Theory of Language Acquisition. Cambridge; London:
Harvard University Press.

Ichl 21 abstract
Robert Ratcliffe
Tokyo University of Foreign Studies
Glottometrics: Quantifying linguistic diversity and correlating it with diachrony: Arabic as a test case
The Arabic language complex, with a written documentation reaching back some 1500 years
and a rich diversity of modern spoken forms, some of which have been influenced by intensive
contact with other languages, provide an excellent laboratory for developing and testing models of
language change and language relationship. The present communication reports on a project to
establish a quantitative measure of the diversity among several Arabic dialects and to correlate the
results with the known history of these dialects. Diversity is measured in four linguistic domains:
lexicon, phonology, inflectional morphology, and morpho-syntax. Lexical difference is measured
using a standard 100 word Swadesh list. Phonological differences are measured in terms of
difference in the phonetic realization of cognate phonemes, using the 27 consonantal phonemes
reconstructable for proto-dialectal Arabic and the 28 consonantal phonemes of Classical Arabic.
Diversity in inflectional morphology is measured in terms of the cognacy or not of the inflectional
affixes used in the verbal conjugation. Morpho-syntactic diversity is measured according to a
number of features which have played a prominent role in the literature of syntactic typology and
the principles and parameters framework. Among these are basic word orders, the positioning of
question particles, and the existence or not of auxiliary verbs and definite articles. Some preliminary
conclusions are the following: diversity in basic lexicon correlates most closely with time of
separation, and hence is most probably due to endemic, internal factors. Diversity in morpho-syntax
and to some extent phonology, by contrast, correlates more closely with geography and contact
history. Finally the inflectional morphological features are the only features of those used which
allow for a tree-like sub-classification, apparently indicating a close correlation between diversity in
this area and migration history, although in fact the actual migration history is more complex than
the tree describes. These conclusions should not be surprising, but the point is to develop an explicit,
empirically supported framework which can be applied to languages whose history is less well
documented than that of Arabic. A further goal is to move beyond the traditional, genetic, family
tree-model of language relationship to a more holistic classification that incorporates the history of
both inter-dialectal and external contact.

Johannes Reinhart (Dept. of Slavonic Studies, University of Vienna, Austria)

The History of Predicatives in the Slavic Languages.
All contemporary Slavic languages have a class of words called predicatives, so called
because of their syntactic function (A. Timberlake, A Reference Grammar of Russian, 2004:
non-verbal predicates). The Russian linguists Viktor Vl. Vinogradov (1894/5-1969) and
Alexander Isaenko (1910-1978) defined them as a separate part of speech. However, this
part of speech has not yet found general recognition, as many grammar dictionaries of Slavic
languages count them among adverbs. As examples of predicatives one could examine the
following ones: Ru. nelzja one must/need not, neobchodimo it is necessary, Ukr. koda
it is a pity, Pol. trzeba it is necessary, Cz. nanic feel/get sick, Slk. zima feel cold, Sln.
dolgas be boring, Bosn./Cro./Serb. ao be sorry, Mac. sram be ashamed, Bulg. zle
feel/get sick.
From a semantic point of view there are at least three subgenres of predicatives: modal
predicatives that describe natural phenomena and predicatives that describe physical or mental
states. From a syntactical point of view the logical subject of predicatives is either in the
dative or in the accusative. They often have an infinitive complement and possess neutral
agreement (default agreement). Morphologically, many of them are homonymous with the
adverbial form of adjectives, but some of them come from nouns or from a nominal group,
e.g. koda literally: damage, zima literally: winter, dolgas literally: long time, sram
shame.
In the earliest attested Slavic language, Old Church Slavonic, predicatives were not very
numerous, cf. e.g. ldz it is possible, permitted, trb it is necessary, useful.
Scholars have traditionally held the view that the category of predicatives already existed in
Common Slavic. A second point of view seems more probable, viz., that there were parallel
developments in the individual languages, based on common starting points.
In this lecture arguments for the second variant will be presented. Mechanisms which favour
the increase of predicatives during the history of various languages will also be analyzed.

Philological accuracy and IT resources

Rita Marquilhas, Ana Lusa Costa, Clara Pinto, Fernanda Pratas, Gael Vaamonde,
(CLUL, Universidade de Lisboa - rita.marquilhas@gmail.com)

0. Introduction
This paper addresses the problem of how to create compatibility between digital scholarly editions
intended to support historical studies (language history included) and robust annotated corpora
offering evidence for the study of language change within diachronic linguistics.
1. Project overview
We will present a project that aims to collect and publish Portuguese and Spanish personal letters
from the 16th to the 19th century, written by people from all social backgrounds: Project P.S Post
Scriptum. The documents, largely unpublished, have been kept as material proof within civil and
religious court proceedings. The paper will discuss, firstly, the methodological options that led to the
digital edition available online before focusing on issues related to the automatic modernization of
Early Modern texts and POS and syntactic annotation.
Given the increased relevance of diachronic data from non-literary texts, there has been growing
interest in corpora constituted by epistolary texts (Daybell, J., 2012; Dossena, M. & Gabriella
Camiciotti, 2012; a.o.). However, it seems to be easier to find letters written by highly educated
authors1 than by individuals belonging to lower social classes. The P.S. project publishes an online
edition of a complete range of letters written in very different social contexts, within welldifferentiated pragmatic situations. Within the field of diachronic linguistics, the dialogical nature of
these personal letters compensates to some extent for the lack of oral sources because the kind of
spontaneous interactions generated by letter exchanges can be seen as a kind of quasi-spontaneous
speech. Letters written in less formal contexts by less educated individuals, in a style that is close to
everyday speech, constitute an extraordinary resource that provides access to phonological,
morphological and syntactic data from a given period. Furthermore, scholars of history and cultural
studies will also be interested in the information it provides about anonymous individuals, their
common lives and social relations.
Given the linguistic and historical richness of this collection, the project proposes a multi-factorial
approach supported by a mixed team of both linguists and historians. The core of the working
process is the philological edition of the historical epistolary texts.
2. Paleographic work in a digital edition
Criteria that support the digital scholarly edition will be made explicit, and arguments in favor of
using technological strategies to improve philological accuracy will be presented. The digital
encoding chosen for the letters' edition follows the Guidelines prepared by the Flemish project
Digital Archive of Letters in Flanders (DALF), which is a TEI-XML compatible project. The XML
final document may be seen as a cartography of the manuscript: alongside the transcription and
facsimile, metadata, key-words, historical and situational information, English translation and a
normalized version of each letter text are made available, in order to facilitate eResearch in different
fields using different electronic tools.
3. Linguistic eResearch
Our project intends to improve diachronic linguistic research using corpus and computational
linguistics. This approach aims to consolidate previous experiments in corpus annotation and to
establish network ties with partner projects, in order to contribute to the accessibility and
comparability of diachronic data.
Recognizing the advantages of standardized spelling for the POS annotation process and lexical
research, the Portuguese letters are being semi-automatically normalized thanks to the adaptation of

See, for instance, two Portuguese language corpora containing letters from previous centuries: the Corpus Histrico do
Portugus Tycho Brahe (IEL-UNICAMP) and the Corpora Diacrnicos edited by the Laboratrio de Histria do Portugus
Brasileiro (UFRJ).

VARD 2 tool for Portuguese (Baron, A. & Rayson, P., 2008; Giusti, R. et al., 2007 and Hendrickx, I.
& Marquilhas, R., 2011).
As for the POS corpus, the Portuguese set of letters is being encoded with the Edictor tool developed
by the Tycho Brahe project, which applied the U Penn annotation system to the Portuguese
language, together with CORDIAL-SIN, (Galves, C. & Britto, H., 2002); as for the text of the
Spanish letters, it is being encoded with FreeLing for Spanish language, and will later be converted
to the U Penn annotation system (Padr, L. & Stalinovsky, E., 2012). This strategy allows for a
further step, i.e., a parsing campaign based on the syntactic annotation system implemented in the
Penn Corpora of Historical English (Kroch, T., Santorini, B. & Diertani, A., 2010).
4. Final remarks
Rather than a bottom-up process (from the paleographic transcription to the syntactic annotation),
our digital edition and corpora annotations involve a dynamic interaction between different
intervention levels. Sometimes, the late phase of syntactic annotation allows some aspects of the
previous work (i.e. palaeographic transcription and text encoding) to be improved (through
the disambiguation of paleographic abbreviations or interpretation of null syntactic categories, for
example). Examples from linguistic data that appeal to multidisciplinary work will be presented.
A digital edition of 2000 private letters (over 600,000 words) from the 16th to the 19th century is
already available.
5. References
Baron, A., & Rayson, P. (2008). VARD 2: A tool for dealing with spelling variation in historical
corpora. In Proceedings of the Postgraduate Conference in Corpus Linguistics. Aston University,
Birmingham.
DALF, Guidelines for the description and encoding of Modern correspondence material
(http://ctb.kantl.be/project/dalf/).
Daybell, J. (2012). The material letter in Early Modern England. Manuscript letters and the cultura
and practices of letter writing, 1512-1635. Hampshire: Palgrave Macmillan.
Dossena, M. & Camiciotti, G. (2012). Letter writing in late modern Europe. Amsterdam: John
Benjamins.
FreeLing (http://nlp.lsi.upc.edu/freeling/).
Galves, C. & Britto, H. (2002). The Tycho Brahe Corpus of Historical Portuguese. Department of
Linguistics,
University
of
Campinas.
Online
publication,
first
edition
(http://www.tycho.iel.unicamp.br/~tycho/).
Giusti, R., Cndido Jr., A., Muniz, M., Cucatto, L., & Alusio, S. (2007). Automatic detection of
spelling variation in historical corpus: An application to build a Brazilian Portuguese spelling
variants dictionary. In Proceedings of the Corpus Linguistics Conference CL2007. Presented at the
CL2007, Birmingham.
Hendrickx, I.& Marquilhas, R. (2011) From old texts to modern spellings: an experiment in
automatic normalisation, Journal for Language Technology and Computational Linguistics 26, n. 2,
65-76.
Kroch, A. Santorini, B. & Diertani, A. (2010). The Penn-Helsinki Parsed Corpus of Modern British
English (PPCMBE). Department of Linguistics, University of Pennsylvania. CD-ROM, first edition
(http://www.ling.upenn.edu/hist-corpora/).
Lopes,
C.
et
al.
Corpus
Compartilhado
Diacrnico:
cartas
pessoais
brasileiras (http://www.letras.ufrj.br/laborhistorico/).
Martins, A. M. (coord.). (1998-2006). CORDIAL-SIN The Syntax-oriented Corpus of Portuguese
Dialects. Retrieved from http://www.clul.ul.pt/pt/recursos/212-cordial-sin-syntax-oriented-corpusof-portuguese-dialects
Padr, L. & Stanilovsky, E. (2012). FreeLing 3.0: Towards Wider Multilinguality, Proceedings of
the Language Resources and Evaluation Conference (LREC 2012) ELRA. Istanbul, Turkey. May,
2012.
TEI, Text Encoding Initiative (http://www.tei-c.org/index.xml)

Using corpus data for comparing actual history to potential alternatives:

pinning down some effects of unstressed syllable reduction in Middle English
Nikolaus Ritt, Vienna

This paper discusses a new method by which diachronic corpora can be used to identify
potentially causal interdependencies among changes in different linguistic domains.
Specifically, the method we propose involves the creation of hypothetical usage data and
their comparison to actually attested ones. To illustrate it, we discuss Middle English because
it represents a language (stage) for which high quality corpus data are readily available, and
in terms of phenomena, we have selected Middle English schwa loss in unstressed final
syllables, which is acknowledged to have had wide ranging effects on the phonotactics of
both morphologically simple and morphologically complex word forms (cf. e.g. Minkova
1991) and played a crucial role in the typological shift of English from a stem based to a word
based language. (cf. e.g. Kastovsky 1992)
Basically speaking, our method is intended to help clarify to what extent changes that
succeed one another might actually be causally related in the sense that the latter changes(s)
were triggered by the earlier ones. In principle, it is very simple: first, we select a specific
change of our choice, such as, for the purpose of illustration, Middle English loss of schwa in
unstressed final syllables. Then, we take a corpus of pre-change texts and identify potential
change inputs, in our case word forms with structures such as X.C (e.g. make make or
blinde blind or X.CC (e.g. moned moaned, earmes arms), and replace relevant vowel
graphemes by a symbol for detelable schwas (We have chosen the symbol @, so that we get
forms such as mak@, blind@, mon@d or earm@s). Next, we transform their shapes so that
they reflect hypothetical outputs of the change (i.e. mak, blind, mond, or earms) and run a
search in order to establish their type and token frequencies. Finally, we repeat the same
procedure for actual post-change data, compare the results, identify differences between the
hypothetical post-change data and the actual ones, and try to interpret them.
As indicated, our own experiment, which used data from the Penn-Helsinki Parsed Corpus of
Middle English (Kroch and Taylor 2000), was intended to identify potential responses to the
effects of schwa loss in Middle English final syllables. Our focus was on the word final
consonant clusters which schwa loss created in both simple and complex word forms (end <
ende vs. moan+ed /m:nd/ < mon+ed /m:nd/). On the assumption that final consonant
clusters are phonologically dispreferred (cf. e.g. Maddieson nd, Prince & Smolensky 2004,
Hoole at al. 2012), we hypothesized that they may be dealt with more readily when they
have a morphological signaling function, i.e. when their occurrence at the end of a word
indicates its morphological complexity (as in seemed, because morph final md is unattested).
Clearly, this is not the case when morphologically produced clusters are also frequent within
morphs. Therefore, speakers might implement strategies resulting either in a reduced usage
of morphonotactically ambiguous forms or in changes that enhanced their morphological
signaling function. (cf. Dressler and Dziubalska 2006).
In order to test our hypotheses, we applied our method to Middle English final /nd/ clusters,
which were produced in the past tenses or participles of verbs ending in /n/ but had always
occurred stem-internally as well (as in hond hand, friend etc.). Our prediction was that the
number of structurally ambiguous word forms ending in /nd/ would be significantly smaller
in the actual post change data than in the hypothetical ones, which simulated a type of Late
Middle English which distinguished itself from Early Middle English only through schwa loss

but through nothing else. We found that this was indeed the case: compared to our
hypothetical post change data, the final /nd/ clusters in the actual data were indeed less
ambiguous in the sense that morphotactic and phonotactic /nd/ clusters were distributed
almost complementarily: in disyllabic word forms final /nd/ indicated past (participle) forms
in almost all occurrences, because intra-morphemic /nd/ cluster in present participles had
been replaced by /ng/ (living < livende). Among monosyllables, most forms with short Vnd
rhymes were past (participle) forms , while word forms with long VVnd rhymes were mostly
monomorphemic if their vowels were high, and mostly complex otherwise. This almost
complementary distribution seems to have been caused by the failure of EME pre-/nd/ vowel
lengthening to be implemented among non-high vowels.
Refraining from premature conclusions about causality, we believe to have demonstrated
that at least two of the changes that occurred in the wake of schwa loss seem to have had a
therapeutic effect on a semiotically suboptimal situation which schwa loss would otherwise
have created. At the same time, we hope to have shown how corpora can be used to
simulate historical development that might have occurred but did not, and how comparison
to such hypothetical histories might help one to get a better idea of what is special about
actual historical developments.
References
Dressler, Wolfgang U, and Katarzyna Dziubalska-Koaczyk. 2006. Proposing
Morphonotactics. Wiener Linguistische Gazette 73.
Hoole, Philip, Bombien, Lasse, Pouplier, Marianne, Mooshammer, Christine and Barbara
Khnert (eds.). 2012. Consonant Clusters and Structural Complexity. Berlin: de Gruyter.
Kastovsky, Dieter. 1992. Typological reorientation as a result of level interaction: the case of
English morphology. In: Kellermann, Gnter - Michael D. Morrissey (eds.). Diachrony
within synchrony: language history and cognition, Frankfurt/M.: 411-428;
Kroch, Anthony, and Ann Taylor. 2000. Penn-Helsinki Parsed Corpus of Middle English, second
edition. http://www.ling.upenn.edu/hist-corpora/PPCME2-RELEASE-3/index.html
Maddieson, Ian. nd. Syllable structure. WALS. http://wals.info/feature/12.
Minkova, Donka. 1991. The history of final vowels in English: the sound of muting. Berlin:
Mouton.
Prince Alan and Paul Smolensky. 2004. Optimality Theory: Constraint Interaction in
Generative Grammar. Oxford : Blackwell.

The Emergence of Nasal Vowels in Haitian Creole and Saramaccan

Yolanda Rivera Castillo
University of Puerto Rico-Ro Piedras
Regarding the genesis of Creole features, some propose one of two possible scenarios:
that Creole features either originate directly from lexifiers or substrata (a genetic approach),
or from some set or subset of universal features (a blank slate approach). Both types of
analysis propose a single path scenario for Creole genesis. We present an alternate analysis
and a relative chronology in the development of nasal vowels and vowel nasalization in two
Creoles: Saramaccan and Haitian Creole. Both languages have phonologically distinctive
nasal vowels (Good and McWhorter, 2012; Frre, 1983):
(1) Saramaccan:
(2) Haitian Creole:

/lagi/, inferior
/ta/, late

vs.
vs.

/lgi/, depend
/t/, time

Following findings from Diachronic Typology, historical evidence, and a phonetic and
phonological analysis of modern Haitian Creole and Saramaccan data, we demonstrate that
these Creoles belong to either one of the following typological classes:
(3) languages with both phonemic nasal and nasalized vowels (followed by coda consonants),
versus languages with nasal vowels only;
(4) languages with [+high] nasal vowels, versus those with only [-high] nasal vowels (those
with [-high] nasals have phonemic distinctions between long and short vowels).
Our analysis also shows that Haitian Creole vowels followed the first two of three possible
paths of language change, as listed below, while Saramaccan vowels followed the last one:
(5) Coda consonants constitute the main source of nasal vowels and vowel nasalization:
(Hombert, 1987): VN > N >
(6) Spreading of nasalization applies from low to mid to high vowels [Vowel Height
Parameter (Chen 1972)]: VHP: low>>mid>>high
(7) Phonologization of vowel nasalization and secondary /n/ deletion applies first to long
vowels: [Vowel Length Parameter (Hajek 1997)]: VLP: V:N >> VN
Although some studies oppose the VHP and VLP parameters, data on Creole languages
suggest that either vowel length or height can play a role in the emergence of nasal vowels.
According to previous research on diachronic typology, vowels might be at different stages in
the phonological shift from oral to nasal phonemes, where length, height (and related ATR),
place, and/or prominence determine which nasalized vowels become phonemic and which just
stay as phonetic variants. Additionally, previous studies have shown that the original
linguistic data, either from the lexifier (Nikiema and Bhatt, 2003) or substrata (Hall, 1950:
476), might be relevant regarding the set of words available in the phonologization of vowel
nasalization. Our study contributes to the general understanding of Creole genesis and the
genesis of nasal vowels.

The early development of Hawai'i Creole English: A reassessment

Sarah Roberts, Stanford University
According to Roberts (2000, 2005), Hawai'i Creole English (HCE) emerged in the first decade of the
twentieth century in the community of locally-born offspring of immigrants and native Hawaiians. The
development of a distinctive vernacular accompanied important sociolinguistic changes, particularly
language shift from ancestral languages to a form of pidginized English in social interactions among
school children. The linguistic situation prior to 1900 was described as revolving more around the
multilingual use of Hawaiian and other ALs, with greater language maintenance than in the later
period. Lexical and syntactic features suggested that the speech of (young) vernacularizing locallyborn was not much differentiated from the form of pidgin English (Hawai'i Pidgin English) spoken by
immigrants contracted for labor. Then between 1900 and 1915, certain grammatical features
uncommon in the pidgin (such as past tense been and infinitive for) became regularized in the nascent
creole, and then between 1915 and 1930 new innovative basilectal features (such as
imperfective/progressive stay [V-in] and past imperfective/progressive been stay [V-in]) arose.
New evidence suggests that this picture of linguistic development is partially flawed. The development
of the mixed speech community of locally-born offspring of immigrants was preceded by several
generations of hapa haoles ("half whites", i.e. offspring of native Hawaiians and whites), most of
whom had Anglophone parents. Recently digitized texts at the Library of Congress include
representations of the speech of hapa haoles in the 1880s, which closely resemble the early creole
attested in the 1900s and 1910s (with prolific use of been and for), suggesting that this variety
underwent development earlier than suspected by Roberts (2000, 2005). The revised picture is that the
locus for the earliest changes probably lay with the Part-Hawaiians who according to Roberts were
among the earliest shifters from ALs to a variety of English (2005:230), and who had access to both
pidginized and non-pidginized English. The acrolectal and mesolectal dimensions of the creole may
thus not only derive from exposure to the kind of English taught in the schools but reflect at least in
part the English spoken by hapa haole bilinguals in the late nineteenth century. The proto-creole of the
hapa haoles then served as input to the English spoken by the mixed population of locally born
offspring of immigrants after circa 1900.
This proposal will also necessitate a reconsideration of the term Hapa-Haole as a language label, first
popularized by Reinecke ([1935] 1969) as an antecedent to HCE. In a little-noticed paper, Powell
(1969) also supplied evidence concerning the derivation of HCE from Hapa-Haole. The term itself is
problematic because it could be interpreted as a description of language mixture (half Hawaiian, half
English) rather than a reference to the social group with the same name. But early attestations of this
language label and related expressions like "half-white dialect" suggest that the speech of hapa haoles
may have been perceived as a distinct variety prior to 1900.

Powell, J. V. 1969. Hapa Haole and the Derivation of Hawaiian English: Being a consideration of the
English Usage of 19th century Native Hawaiian-speakers, on the Basis of Extant Written
Sources. Hamilton Library, University of Hawaii, unpublished typescript.
Reinecke, J. E. 1969. Language and Dialect in Hawaii: A Sociolinguistic History to 1935. Honolulu:
University of Hawaii Press.
Roberts, S. J. 2000. Nativization and the genesis of Hawaiian Creole. In J. McWhorter, ed.
Language Change and Language Contact in Pidgins and Creoles, 257-300. Amsterdam:
Benjamins.
---------. 2005. The emergence of Hawai 'i Creole English in the early 20th century: The
sociohistorical context of creole genesis. Ph D dissertation, Stanford University.

Il mest avis que: the evolution of a French evidential marker

Amalia Rodrguez Somolinos
Universidad Complutense de Madrid
We intend to study here the evolution of the French evidential marker il mest avis que/ ce mest
avis. In Medieval French, the complex verb estre avis, which is generally classified with opinion
verbs, presents a double construction. As il/ce mest avis que p, it governs a complement clause
introduced by que. Besides, estre avis may function as a parenthetical clause: ce m'est avis/ si me
fu avis/ si come lors mert avis.
Il mest avis que p / ce mest avis are epistemic modalities expressing the belief of the speaker in
the truth of an assertion. Both structures have an evidential as well as a modal value. Their modal
value varies on the scale of the reliability of knowledge, going from probability (speculation) to
(almost)-certainty. They usually correspond to a high degree of certainty.
Il mest avis que p and the parenthetical clause ce mest avis have a very similar semantic
function in Medieval French. Due to their etymological value, they are linked to visual
perception and are frequently combined with perception verbs (veoir, esgarder): Ce mest avis
que il sont troi; le nain et la pucele voi (Erec et Enide, v. 1097) (There are three of them, as I can
see, I see the dwarf and the young lady). The speaker is a direct witness and states an
observation.
Il mest avis que/ ce mest avis can have a counterfactual use, they introduce a vision, a dream, a
direct perception which exists in the memory or the imagination of the speaker: Touz pasmez
une avison/ Vi, qui ert bele a deviser,/ [..]/ Vis me fu, que devant moi vint/ [..]Venus [a] grant
processon (Huon de Mry, Li Tornoiemenz Antecrit, v. 48) (I saw Venus, who appeared in
front of me walking in procession). Vis me fu displays here a fictitious dimension which is seen
by the speaker as if it were real.
Both structures Il mest avis que p/ ce mest avis have inferential uses as well. The inference can
be based on an immediate perception, very often visual. They manifest an operation of creation
of information (Dendale 1994). The speaker can also base his personal judgment on evidence,
from which he infers a conclusion. Very often il mest avis que expresses an opinion, a
judgment, an evaluation made by the speaker, which results from thought.
The medieval semantic values still exist in Preclassical French (1500-1650): il mest avis que
expresses a personal judgment and is still linked to visual perception, as well as with the unreal,
the dream, the imagination. The frequency of the parenthetical clause ce mest avis/ mest avis
decreases. It will disappear during the first half of the 19th century.
In Classical French (1650-1789) il mest avis que begins a process of fixation and is only used in
the present tense. It is mostly used in popular or regional language. Il mest avis que is no longer
linked to visual perception, its meaning is I think, I believe. In the 19th century, the structure is
fossilized as mest avis que, and belongs to colloquial French. Used in spoken language until the
1950s, mest avis que is not in common use nowadays.
References
AIJMER, K. (1997) I think, an English modal particle, in T. Swan O. J. Westvik eds Modality
in Germanic languages. Historical and comparative perspectives. Berlin/ New York, Mouton de
Gruyter, 1-47.
DENDALE, P. (1994) Devoir pistmique, marqueur modal ou videntiel?, Langue Franaise
102, 24-40.
DENDALE, Patrick & VAN BOGAERT, Julie (2007) A semantic description of French lexical
evidential markers and the classification of evidentials, Rivista di Linguistica 19, 1, 65-89.

The ergative proclitics of Austronesian verbs: a reconstructional error?

Malcolm Ross
Australian National University
Starosta, Pawley & Reid's 1981 paper (SPR), published only in 2009, reconstructs early forms of
Austronesian with ergative and absolutive pronominals that were Wackernagel enclitics. These
pronominals were enclitic to the verb (typically the first element of a verbal clause) as they still are in
certain Formosan and Philippine languages. For example, from Teruku Seediq (Formosan):
(1) a. me-taqi=ku
paru sapah. [Intransitive]
AV-sleep=ABS:1SG big room
I sleep in a big room.
b. byeq-un=ku=na [Transitive]
give-UP=ABS:1SG=ERG:3SG
'He gave (it) to me'
When an auxiliary or negator was first element, then they were/are enclitic to that element. Again from
Teruku Seediq:
(2) a. gisu=su
m-ahu
lukus
[Intransive; lukus is an uncasemarked generic noun]
PRG=ABS:2SG AV-wash clothes
You are washing clothes.
b. wada=ku=na
qta-un
[Transitive]
PST=ABS:1SG=ERG:3SG see-UVP
He saw me.
In two Formosan languages (Puyuma and Paiwan) and in numerous languages outside Taiwan, ergative
pronominals are always proclitic or prefixed to the verb, while absolutive pronominals remain enclitic.
From Paiwan:
(3) ku=k<in>eLem=esun
ERG:1S=<PF>beat.UVP=ABS:2SG
I beat you yesterday.
SPR propose that these clitic pronominals were once enclitics to the auxiliary which, after the deletion of
the auxiliary, were stranded and became proclitic to the verb. A similar account is offered for Philippine
languages by Wolff (1996). There is a problem with this explanation, however. In intransitive clauses the
absolutive enclitic was also attached to the auxiliary and should with auxiliary deletion have ended up as
a proclitic to the verb. But this has not happened, as (3) and (4), from Paiwan, show.
(4) uri vaik=emun
will go.AV=ABS:2P
You will go.
This suggests that SPR's analysis at least needs tweaking, and perhaps reformulating.
REFERENCES
Starosta, Stanley, Andrew Pawley & Lawrence A. Reid, 2009. The evolution of focus in Austronesian.
In Elizabeth Zeitoun, ed., Formosan linguistics: Stanley Starosta's contributions, 2:329480.
Taipei: Institute of Linguistics, Academia Sinica.
Wolff, John U., 1996. The development of the passive verb with pronominal prefix in western
Austronesian languages. In Bernd Nothofer, ed., Reconstruction, classification, description:
Festschrift in honor of Isidore Dyen, 1540. Hamburg: Abera.

HOW CAN VARIATION LEAD TO PHONOLOGICAL CHANGE. THE CASE OF YESMO IN SPANISH
Assumpci Rost Bagudanch
Universitat de les Illes Balears
Recent studies on linguistic change and specifically, on sound change, have shown that the evolution of a
language seems to start with the existence of variation (cf. Bybee 1998, 2001; Lindblom 1990a, 1990b; Ohala 1974,
2005). This is somewhat surprising if we consider that there are some forces leading to the minimum effort and to
linguistic economy. Therefore, one should expect speakers have no more than one possible solution for a given
context. This seems to break that principle. Interestingly, several investigations have set up that change is not
possible without this variation. Phonologically, it allows the explanation of the evolution sources. Following
Blevins (2004), phonetic changes occur in situations of sound misperception -which can lead to the reanalysis of the
segment/s-, problems in the segmentation of the phonetic chain -that can be followed by the reinterpretation of the
sequence and, thus, it can give place to phonological recategorization- and, finally, problems of multiple choice in
the stimulus one is exposed to when acquiring L1.
When focusing in Spanish yesmo, recent research shows clearly the nature of change in progress (see
Moreno Fernndez 2005). From the beginning of the 20th century, // has been gradually substituted by a palatal
phoneme, traditionally considered as a fricative consonant in spite of its acoustic formantic structure. Nowadays,
the most recent descriptions of Spanish phonetic and phonological system agree with its approximant nature, at
least in some regions. As a matter of fact, Navarro Toms pointed out this phenomenon in early time, and even
RAE did it in its Esbozo de una nueva gramtica de la lengua espaola (1973). In the last 50 years it has been a
recurring topic (see Alcina y Blecua 1975, Quilis 1993, Hualde 2005, RAE 2011).
What is really interesting in this process is that there is not a real dichotomy between a lateral and an
approximant consonant. Instead, there are plenty of allophones that concur as variants of //. Hence, the aim of this
paper is to focus on the huge variation in the phonetic representation of // and how it progress in a different
organization of the Spanish sound system which, obviously, has relevant phonological consequences. The analysis
of experimental data reveals that a deep reorganization of the system is in progress: research demonstrates that
several acoustic solutions can no longer be set down as // allophones because they have become variants of /l/, at
least in certain contexts, or have been assigned to //, which has taken the place of the lateral palatal.
REFERENCES
Alcina, Juan y Jos Manuel Blecua (1975), Gramtica del espaol, Ariel, Barcelona.
Blevins, Juliette (2004), Evolutionary Phonology, Cambridge University Press, Cambridge.
Bybee, Joan (1998), Usage-based Phonology, Darnell, M., E. Moravcsik, F. Newmeyer, M. Noonan and K.
Wheatley (eds.), Functionalism and Formalism in Linguistics, volume I: General Papers, pp. 211-242.
Bybee, Joan (2001), Phonology and Language Use, Cambridge University Press, Cambridge.
Hualde, Jos Ignacio (2005), The Sounds of Spanish, Cambridge University Press, Cambridge.
Lindblom, Bjorn (1990a), Explaining phonetic variation: a sketch of the H&H Theory, Hardcastle, W. and A.
Marchal (eds.), Speech Production and Speech Modelling, Dordrecht, Kluwer Academic Publishers, pp.
403-439.
Lindblom, Bjrn (1990b), Models of Phonetic Variation and Selection, PERILUS, XI, pp. 65-100.
Moreno Fernndez, Francisco (2005), Cambios vivos en el plano fnico del espaol. Variacin dialectal y
sociolingstica, Cano Aguilar, Rafael (coord.), Historia de la lengua espaola, Ariel, Barcelona, pp.
973-1010. 2nd edition.
Navarro Toms, Toms (1982), Manual de pronunciacin espaola, CSIC, Madrid. 21st edition.
Ohala, John (1974), Experimental historical phonology, Anderson, J.M. and C. Jones (eds.), Historical
Linguistics II. Theory and Description in Phonology, North Holland, Amsterdam, pp. 353-389.
Ohala, John (2005), Phonetic explanations for sound patterns: implications for grammars of competence,
Hardcastle, W. and J.M. Beck (eds.), A Figure of Speech. A festschrift for John Laver, Erlbaum, London,
pp. 23-38.
Quilis, Antonio (1993), Tratado de fonologa y fontica espaolas, Gredos, Madrid. 2nd edition.
RAE (1973), Esbozo de una nueva gramtica de la lengua espaola, Espasa, Madrid.
RAE (2011), Nueva gramtica de la lengua espaola. Fontica y Fonologa, Espasa, Barcelona.

The syntax of mood constructions in Old Japanese: A corpus based study

Kerri L. Russell and Peter Sells
This paper investigates several mood-related constructions in Old Japanese (OJ), the language of 8th
century Japan. We focus on imperative, prohibitive and optative constructions, expressing the desire
of the speaker for either the speaker or another entity to perform (or not) an event (or situation) (cf.
Bybee et al. 1994). These forms have not been discussed in any detail for OJ. Previous literature (e.g.,
Frellesvig 2010 and Vovin 2009) briefly describes them and presents several examples, but does not
investigate the grammatical properties. OJ has several inflectional forms expressing these categories:
(1)

Imperative:
Prohibitive:

Optative:

yukye
yuku na
na-yuku
na-yuki-so
na-yuki-sone
yuki-koso
yukana
yukane
yukanamu/yukanamo

Go!
Dont go!
Dont go!
Dont go!
I dont want you to go.
I want (someone) to go.
I want to go./Lets go.
I want you to go.
I want him/her/it to go.

Most of these fell out of use during Early Middle Japanese (800-1200), and only the imperative and
the yuku na prohibitive constructions survive in the modern language. The language developed a
new system which included a desiderative, but the optatives were lost.
Predicate forms dedicated to a pure optative meaning are found sporadically across the worlds
languages, and occasionally more concentratedly in certain language groups. The present study uses
the Oxford Corpus of Old Japanese, a syntactically annotated corpus, to investigate these structures.
The optative verb may take its full range of arguments, including an overt subject, as in (2):
(2)

amawotomyedomo
na
ga
na
nora-sane
diver.women
you
genitive
name say-respect.optative
I want (you) the diving women to tell me your names. (Manysh 9:1726)

We have found that the logical subject, i.e., the entity the speaker desires to perform (or not) the event
of the verb, is never marked for case in any of the mood constructions in (1) (as exemplified in (2)),
even though case-marking is found on subjects in all other true finite clause-types. It is unlikely that
this can be traced to an analysis of the subject as a vocative (cf. Zanuttini 2008), as OJ has a vocative
marker, which does not co-occur in the corpus with optative-marked predicates.
We believe that the correct analysis is that the mood constructions involve partially-finite
clauses which license subjects but not case marking on them. Non-subject arguments may be present
and bear the expected case marking as often as in regular finite clauses, but our corpus study reveals
fewer variations in constituent orders compared to finite clauses. All these facts suggest that clauses
headed by predicates inflected as in (1) are not fully finite and/or otherwise lack certain canonical
clausal properties. As only the core imperative and prohibitive uses survive beyond the 13th century,
we hypothesise that the syntax of Japanese was developing into having a more robust finite/non-finite
split in clauses, leading to the loss of these intermediate categories. In the modern language, all overt
subjects have the possibility for case marking.
References
Bybee, Joan L. and Revere Perkins and William Pagliuca. 1994. The evolution of grammar: Tense,
aspect and modality in the languages of the world. University of Chicago Press.
Frellesvig, Bjarke. 2010. A History of the Japanese Language. Cambridge University Press.
Frellesvig, Bjarke, Stephen Wright Horn, Kerri L. Russell, and Peter Sells. n.d. The Oxford
Corpus of Old Japanese. http://vsarpj.orinst.ox.ac.uk/corpus/corpus.html
Vovin, Alexander. 2009. A Descriptive and Comparative Grammar of Western Old Japanese, Volume
2: Adjectives and Verbs. Folkestone, UK: Global Oriental Press.
Zanuttini, Raffaella. 2008. Encoding the addressee in the syntax: evidence from English imperative
subjects. Natural Language and Linguistic Theory 26, 185-218.

Empty referential subjects in Old English prose and poetry

Kristian A. Rusten, University of Bergen
In recent years, renewed scholarly attention has been directed toward the occurrence of empty
referential subjects and the resulting subjectless finite clauses in the Old Germanic
languages. In this connection, Old English (OE) has also come under empirical scrutiny.
Quantitative investigations based on the YCOE (Taylor et al.) and the YCOEP (Pintzuk &
Plug 2001) have been carried out by Rusten (2010) and Walkden (2012). To a large extent,
these works independently reach the same conclusions, namely that empty referential subjects
are extremely rare in classical OE, that the richness of the verbal inflections cannot fully
account for the occurrence of the phenomenon and that the empty subjects most frequently
have third person reference. With the exception of the final result, these findings contradict
van Gelderen (2000: 121), where it is claimed that pro-drop is quite common in OE.
Walkden (2012) diverges from Rusten (2010) in using evidence from e.g. Beowulf to
suggest that a dialect split is evident in the distribution of empty referential subjects in OE. It
is argued that such subjects are ungrammatical in West Saxon but available under certain
restrictions in Anglian or Anglian-influenced texts (Walkden 2012: 191). This is an
interesting finding, yet it is also worthwhile noting that scant attention has thus far been
awarded to the distinction between the genres of prose and poetry as concerns subject
omission in OE. Importantly, it is conceivable that the high frequencies in Beowulf are due not
to its suggested Anglian features, but rather to its genre. Furthermore, it is also possible that
factors other than (or in addition to) dialect or genre, such as the date of composition of the
investigated works, may be the cause of the observed differences.
This paper presents the results of a contrastive, quantitative investigation of empty
referential subjects as they occur in 11 texts of OE prose and in 33 texts and text extracts of
OE poetry. These texts and text extracts have been analysed in their entirety, and the results
have been tested statistically. The results demonstrate that empty referential subjects are
significantly more frequent in the poetry than in the prose. This shows that more research is
needed in order to corroborate Walkdens (2012) hypothesis. Moreover, the paper shows that
the occurrence of empty referential subjects in both prose and poetry correlates with time of
composition, with such subjects being significantly more frequent in early OE prose and
poetry as opposed to later periods. Thus, this result may contribute to falsifying Amos (1980:
134) claim there is no evidence that a subject-less sentence was unsyntactical in later Old
English.
References
Beowulf = Beowulf and Judith. The Anglo-Saxon poetic records, vol. 4. Ed. by Elliot V. K. Dobbie. 1953. New
York: Columbia University Press.
Amos, Ashley. 1980. Linguistic means of determining the dates of Old English literary texts. Cambridge, MA:
The Medieval Academy of America.
Gelderen, Elly van. 2000. A history of English reflexive pronouns. Person, self and interpretability. Amsterdam:
John Benjamins.
Pintzuk, Susan & Leendert Plug. 2001. The York-Helsinki parsed corpus of Old English poetry.
http://www-users.york.ac.uk/~lang18/pcorpus.html (2 April 2012).
Rusten, Kristian A. 2010. A study of empty referential pronominal subjects in Old English prose. MA thesis,
University of Bergen.
Taylor, Ann, Anthony Warner, Susan Pintzuk & Frank Beths. 2003. The York-Toronto-Helsinki parsed corpus of
Old English prose. http://www-users.york.ac.uk/~lang22/YCOE/YcoeHome.htm. (2 April 2012).
Walkden, George L. 2012. Syntactic reconstruction and Proto-Germanic. PhD dissertation, University of
Cambridge.

Diagnosing embedded V2 in Old French and Old English

Christine Meklenborg Salvesen and George Walkden
Over the years there has been an extensive debate as to the target of verb-movement in asymmetric V2 languages. For modern V2 languages
such as German and Dutch, the main point of contention has been whether subject-initial V2 main clauses project a CP, with den Besten (1977),
Evers (1981) and Schwartz & Vikner (1996) arguing that they do and Travis (1984) and Zwart (1993) arguing that they do not. For older V2
languages such as Old English (OE) and Old French (OFr), whereas early generative analyses proposed that the verb always moves to a head in the
left periphery (e.g. van Kemenade 1987 for OE; Adams 1987 for OFr), in recent years strong voices have made the claim that many or all cases of
apparent V2 in these languages take place within IP (e.g. Pintzuk 1993 for OE; Rinke & Meisel 2009 for OFr). This claim gains prima facie support
from the fact that more than one constituent can occur preverbally in matrix clauses in both OE and OFr (see Salvesen 2013 for OFr and Walkden
2012: 77104 for OE). We will refer to the V-to-C analysis as the C-V2 hypothesis and the V-to-I analysis as the I-V2 hypothesis.
In this paper we evaluate the two hypotheses according to their predictions for embedded clauses. If the I-V2 hypothesis is correct in its strongest
form, SpecIP has to be analysed as an A'-position not restricted to subjects (Pintzuk 1993: 25; Rinke & Meisel 2009: 111). All else being equal, then,
we would not expect there to be any asymmetries between main and subordinate clauses, given that nothing should prevent material from moving
into SpecIP in embedded clauses just as it does in main clauses. However, both Old English and Old French are usually regarded as asymmetric V2
languages.
In order to examine verb movement in Old English and Old French, we have gathered a pilot corpus of embedded clauses from both languages.
If the I-V2 is correct, we will expect to find instances of V2 embedded under all kinds of verbs. If the C-V2 hypothesis is correct, we will only
expect to find V2 under so-called bridge verbs.
For our pilot study, in Old English we have so far examined a total of 442 V2 complement clauses (an exhaustive sample from the YCOE
corpus; Taylor et al. 2003), whereas the corresponding number in Old French is 114 (taken from La queste du Graal, in the BFM corpus; Heiden et
al. 2010). In order to apply the bridge vs. non-bridge distinction in an objective manner, we have used Hooper & Thompsons (1973) distinction
between verb classes in analysing these clauses. This approach distinguishes between five different classes of verbs (AE), and it is expected that
types C and D do not admit embedded main clause phenomena.
What we find is that OFr and OE behave in a remarkably similar manner when it comes to embedded V2. In both languages, non-SV V2
complement clauses are comparatively rare: 20.2% of V2 complement clauses in OFr (114/563), and 7.0% (422/6327) in OE. 62.3% of all embedded
V2 complement clauses in Old French (71/114) are embedded under a type A verb (essentially verbs of saying). In Old English the corresponding
figure is 44.6% (197/442). The second most common verb type is type E, with 20.1% (23/114) of the embedded V2 clauses in Old French and 26.0%
(115/442) in Old English.
For OFr we have not found any non-SV V2 complement clauses under type C verbs, and only four under type D verbs. For OE we have found
four examples under type C and twelve under type D. Crucially, though exhibiting surface V2, all these examples are structurally ambiguous: most
feature heavy subjects that can be assumed to have undergone a process of rightward extraposition, as in (1) from OE, therefore yielding no evidence
that the verb has moved to the left periphery.
(1) Nis nan twyn t eow ne beo forgolden lc ra stapa e ge to Godes huse stppa
NEG-is no doubt that you.DAT NEG be requited all those steps that you to Gods house step
There is no doubt that you will be repaid for all those steps you climb to Gods house
(cocathom2,+ACHom_II,_34:259.113.5787)
The conclusion we draw from the data is that there are indeed root-embedded asymmetries with respect to V2 in Old English
and Old French (van Kemenade 1997 comes to the same conclusion for OE, though without having conducted an exhaustive search). We propose
that the difference between the classes of verbs in Hooper & Thompson (1973) is essentially a difference in the complements they may select.
Building on insight from Rizzi (1997) and Haegeman (2003), we propose that verbs of type A, B and E may select a ForceP complement containing
a high complementizer, situated under Force0. Type C and D verbs, however, necessarily select a FinP complement containing a lower
complementizer, under Fin0. Certain examples from OFr and OE strongly suggest that this is indeed the case. With type A verbs, there are several
instances of doubly spelled out complementizers, as in (2) from OFr and (3) from OE.
(2) Or dit li contes que quant Melianz se fu partiz de Galaad que il chevaucha tant que il vint a la forest gaste now
says the story that when Meliant REFL was left from Galaad that he rode so-much that he came to the forest wast
The story then tells that after Meliant had left Galaad, he rode so far that he reached the Forest Gaste
(3) Hu ne wost u nu t lc ara manna e oerne swie lufa t
how NEG know you now that each the.GEN men.GEN that other-ACC dearly loves that
hine lyst bet accian and cyssan onne oerne on br lic ?
him.ACC.SUBJ better touch.INF and kiss.INF the.ACC other-ACC on bare body
Dont you know that every man who loves another would prefer to touch and kiss the others naked body?
(cosolilo,Solil_1:42.14.539)
The data therefore support the C-V2 hypothesis for both OFr and OE.
References: Adams, Marianne. 1987. From Old French to the theory of pro-drop. Natural Language and Linguistic Theory 5, 132. Besten, Hans den. 1977. On the Interaction of Root
Transformations and Lexical Deletive Verbs. Ms. University of Amsterdam. Evers, Arnold. 1981. Verb-second movement rules. Wiener Linguistische Gazette 26, 1534. Haegeman, Liliane.
2003. Conditional clauses: external and internal syntax. Mind and Language 18, 317339. Heiden, Serge, Cline Guillot, Alexei Lavrentiev, & Lauranne Bertrand. 2010. Base de Franais
Mdival. Lyon: ENS. Hooper, Joan, & Sandra Thompson. 1973. On the applicability of root transformations. Linguistic Inquiry 4, 465497. Kemenade, Ans van. 1987. Syntactic case and
morphological case in the history of English. Dordrecht: Foris. Kemenade, Ans van. 1997. V2 and embedded topicalization in Old and Middle English. In Ans van Kemenade & Nigel Vincent
(eds.), Parameters of morphosyntactic change, 326352. Cambridge: Cambridge University Press. Pintzuk, Susan. 1993. Verb seconding in Old English: verb movement to Infl. The
Linguistic Review 10, 535. Rinke, Esther, & Jrgen M. Meisel. 2009. Subject inversion in Old French: Syntax and information structure. In Georg Kaiser & Eva-Maria Remberger (eds.),
Proceedings of the Workshop Null-subjects, expletives, and locatives in Romance, 93130. Konstanz: Fachbereich Sprachwissenschaft. Salvesen, Christine M. 2013. Topics and the left
periphery: a comparison of Old French and Modern Germanic. In Terje Lohndal (ed.), In search of Universal Grammar: from Old Norse to Zoque, 131171. Amsterdam: John Benjamins.
Schwartz, Bonnie D., & Sten Vikner. 1996. The verb always leaves IP in V2 clauses. In Adriana Belletti & Luigi Rizzi (eds.), Parameters and Functional Heads, 1163. Oxford: Oxford
University Press. Taylor, Ann, Anthony Warner, Susan Pintzuk, & Frank Beths. 2003. York-TorontoHelsinki Parsed Corpus of Old English Prose. Travis, Lisa de Mena. 1984. Parameters
and effects of word order variation. PhD dissertation. MIT. Walkden, George. 2012. Syntactic reconstruction and Proto-Germanic. PhD dissertation. University of Cambridge. Zwart, C. JanWouter. 1993. Dutch syntax: a Minimalist approach. PhD dissertation. University of Groningen.

On the Rise of Subjective Obligation Usage of English Must:

A Blended Usage Hypothesis
Keisuke SANADA
Sapporo Gakuin University, Japan
This presentation investigates how the English deontic must came to express
subjective obligation (SO), where the speaker serves as a deontic source with his/her own
want (cf. Leech 2004), from a diachronic perspective. Although many previous studies
such as Traugott (1989) and Traugott and Dasher (2002) have discussed the
subjectification of English modals, no such studies seem to have focused on the rise of
the SO usage of must. I argue that the SO usage, whose examples were rarely found in
the present research, came into existence through a blend of two usages represented by
the Old English motan: the subjective prayer (SP) (expressed in a subjunctive mood) and
objective obligation (OO) (expressed in an indicative mood) usages, illustrated in (1) and
(2), respectively.
(1) ue ic swior t / u hine selfne geseon moste feond on / frtewum fyl-werigne.
[Beowulf: 960-962]
(2) lond-rihtes mot re mg-burge monna / ghwylc idel hweorfan syan elingas /
feorran ge-fricgean fleam eowerne dom- / leasan dd
[Beowulf: 2886-2891]
Observing cases of the Middle English mo(o)t as well as of motan, I would like to
propose two kinds of motivations that may have engendered the rise of the SO usage.
One is pragmatic motivation. If one prays for some event to take place, one will feel that
the event is obligatory. This shows that prayer and obligation are in a metonymic
relation, which may have driven the blend into the existence of the SO usage. The other
is morphological motivation. In the Middle English period, a morphological distinction of
grammatical moods blurred, making it more difficult to distinguish SP and OO
morphologically in mo(o)t than in motan. This morphological complication may also
have motivated the rise of SO as a blended usage; in fact, mo(o)t could express SO in
both the indicative and subjunctive moods.
To sum up, this presentation advances the hypothesis of the SO usage of must as
a blended usage of SP and OO usages of motan, and proposes pragmatic and
morphological motivations for the blend.

Selected References
Leech, Geoffrey N. 2004. Meaning and the English Verb (3rd ed.). London: Longman.
Traugott, Elizabeth C. 1989. On the Rise of Epistemic Meanings in English: An
Example of Subjectification in Semantic Change, Language 65: 31-55.
Traugott, Elizabeth C., and Richard B. Dasher. 2002. Regularity in Semantic Change.
Cambridge: Cambridge University Press.

A Lower Sorbian sound change that got derailed

Gunter Schaarschmidt, University of Victoria
The development o > in Lower Sorbian (LSo) is a well-known but little understood
phenomenon in the history of this language. The change may have started as early as the 13th
century although documentation in Germanized place-names and loanwords is sparse. Labeling
labial and velar as peripheral (P) for the purposes of this discussion we may conveniently
abbreviate this rule as follows:
(1)

[P] o [notP] > [P] [notP]

EXAMPLES (in modern orthography): bl pain; gla heath; p after, in, along;
ksa goat; but: bok side; gowno feces, excrement; chopi to begin, to start; kobya
mare.
As this change was a mere phonetic development, did not become a phoneme but
remained an allophone. This fact may very well have been the reason why in both the toponymic
material and the early written sources was represented only sporadically as u.
(2)
This situation changed when due to the labialization of the nonpalatalized > w (a
bilabial), a) there arose a contrast PKowPK : PKwPK, as the new w < merged with bilabial w
(Schaarschmidt 1998, 126, 142). In the orthography of Muka (1911-1918) this yielded the
minimal pairs wowaki other, m sg: waki brawler, pl and kow anvil: k Pfahl; b) there
also arose a contrast wC : woC in a position after , due to the elision of as in godny [godny]
hungry : gdny suitable (Marti 2007:45); and, finally, c) due to analogy many LSo dialects
permitted in the enironment PK_PK as in mga (by) could und mgu I can, cf. m you
[sg] can, mc can [inf] (Brijnen 1994, 8182).
Many LSo dialects did not participate in this development so that remained an allophone
there. We have hypothesized that in the dialects with (2) this disturbance of a long-standing
morpheme structure rule initiated a chain reaction in that in these dialects lost its labial element
and was changed to various other vowels, specifically e and y (Schaarschmidt 1998:142-3). This
change in turn affected in all other environments. The orthography reformers of 1952 eliminated
the letter entirely presumably in the assumption that it was an allophone counter to (2) above.
The orthography reform of 1995 reintroduced the letter but not in the environment
before , i.e., an environment where it is phonemically relevant (Starosta 1999).
We shall provide some crucial data from dialect descriptions in the last 50-60 years to
show that Mukas assumption of the retention of in the environment before w < was indeed
correct.

Clause Combining as Syntactic Coding of Temporal Relations in Old German

Anna Theresa Schmidt University of Potsdam

Matthiessen/Thompson (1988) hold that syntactic clause combining is the grammaticalisation

of the rhetorical organisation of the discourse. More precisely, subordinated relations are realised through subordinated clauses.
As shown by research a distinction must be made between different types of hypotaxis (Reis
1997). Besides the notion of prototypical subordination (integrated) there is also non prototypical subordination (non-integrated). Konig/van der Auwera (1988) assume that throughout
different periods of the German language, a development occurred, from non-integrated to
resumptive to integrated. Non-integrated (n-i) for this purpose means that there are two independent constituents before the finite verb. Resumptive (RE) denotes that there is one
independent constituent before the finite verb and that the prefield is occupied by a resuming
element. Integrated (i) means that there is one constituent located before the finite verb and
that this constituent occupies the prefield. Axel (2002) turns against this assumption of a
continuum. According to her, the resumptive structure appears in all language periods.
My paper shows that the hypothesis of Konig/van der Auwera can be confirmed. For the
purpose of the analysis, the paper will refer to temporal clauses, regardless of their position
within the sentence. I point out how temporal clauses were realised syntactically in the different
language periods. A corpus consisting of text from Middle High German (MHG) and Early
New High German (ENHG) is examined from a quantitative and qualitative point of view. The
texts were specifically selected to stem from different language periods, yet to have the same
content.
(1)

a.

b.

c.

[ n-i do man den helt reinen u

z der stad in daz h
us tr
ug]. [ l
ute ] volgeten im [ ... ]
(T MHD 1076)
when one took the hero from the site into the house, people followed him
[ Do sie quamen u
f den se], [ RE do] wart ir also rechte we, daz sie nehmen musste
den tod (T MHD 95)
when they went to sea, she was in such pain that she died
[ i do er das sahe.] gieng er ein wenig f
urbas. (T FNHD 697)
when he saw it, he kept going

From a quantitative point of view, it is possible to say that the hypothesis of a trinominal development can be confirmed. Given the corpora sighted to date, significantly more constructions
with a resumptive element were observed in the MHG than in the ENHG. However, there is a
significant increase of constructions with an integrated constituent in the ENHG corpus.
(2)

a.
b.

[ Do sie quam], [ RE do] was he tod. (T MHD 965)

when she came, he was dead
[ i Als aber sy eylent dar kam.] was jr ohem tod. (T FNHD - 397)
when however she came there in a hurry, her uncle was dead

These examples illustrate a qualitative point of view. The same sentence is realised with a
resumptive element in MHG in contrast to the integrated structure in ENHG.
Going forward, the focus of this research will include temporally related but independent
clauses. It will be examined how temporal relations are coded and whether the hypothesis
of the grammaticalisation of the rhetorical organisation of the discourse can be confirmed.
References
Axel, K. (2002): Zur diachronen Entwicklung der syntaktischen Integration linksperipherer Adverbials
atze im Deutschen. In:
Beitr
age zur Geschichte der deutschen Sprache und Literatur 124, 1-43.
K
onig, E. & van der Auwera, J. (1988): Clause integration in German and Dutch conditionals, concessive conditionals, and concessives. In: J. Haiman & S. Thompson (eds.), Clause Combining in Grammar and Discourse. Amsterdam: Benjamins.
101-133.
Mattiessen, C.& Thompson, S. (1988): The structure of discourse and subordination. In: J. Haiman & S. Thompson (eds.),
Clause Combining in Grammar and Discourse. Amsterdam: Benjamins. 275-329.
Reis, M. (1997). Zum syntaktischen Status unselbst
andiger Verbzweit-S
atze. In: C. D
urscheid, K. H. Ramers & M. Schwarz, Hrsg.,
Sprache im Fokus. Festschrift f
ur Heinz Vater zum 65. Geburtstag. T
ubingen: Niemeyer. 121144.

The changing fate of modal periphrases in Spanish:

a multivariate approach within diachronic corpus linguistics
Kim Schulte, Universitat Jaume I, Castelln
Jos Luis Blas Arroyo, Universitat Jaume I, Castelln
The factors involved in the choice between different periphrastic modal constructions in
Spanish have been the subject of discussion in both prescriptive and descriptive work since
the 18th century. Until very recently, the focus has been almost exclusively on whether or not
the use of different modal periphrases coincides with different types of modality (epistemic or
deontic). Whilst such a distinction does appear to apply, to a certain extent, in 16th- and 17thcentury Spanish (Blas Arroyo, 2013), its absence in later periods leads many scholars to
conclude that the respective modal constructions occur in free variation (e.g. Gmez
Manzano, 1992: 162; Maeseneer, 1998: 40). However, such a unidimensional focus on
semantic differences ignores the existence of a wide range of linguistic and extralinguistic
factors that may trigger the use of, or be involved in the choice between, one of these
periphrases. Furthermore, diachronic shifts in their relative frequency cannot be explained by
free variation alone; it is highly likely that other, as yet unidentified variables are involved.
In order to establish which variables have contributed to the choice between five available
modal constructions in actual language use over the centuries, this study, which is part of a
wider research project on the factors involved on diachronic variation funded by the Spanish
Ministry of Science and Education, systematically examines all relevant tokens that occur in
an extensive diachronic corpus (of more than 3.2 mil words), consisting entirely of personal
correspondence, private memoirs and other texts with a similar degree of communicative
immediacy (cf. Koch & Oesterreicher, 1985, Oesterreicher, 2004), with the explicit aim of
representing, as closely as possible, the real-world use of Spanish since the 16th century,
rather than literary norms or trends; this is particularly important in view of the prescriptive
pressures that have artificially influenced the literary usage of these constructions over the
past centuries.
As there is, to date, no systematic account of which factors are at play in the choice between
different modal periphrases, for the purpose of this study a wide range of potentially relevant
variables are considered. For each token present in the corpus, the same set of linguistic and
sociolinguistic features are examined and receive a binary value. In the area of semantics, in
addition to the traditional modal deontic/epistemic distinction, variables such as
animate/inanimate subject, aktionsart of the main verb, and the degree of (im)personality of
the clause are taken into account; morphosyntactic features examined include person, number,
tense and mood, presence or absence of an overt subject, the internal syntax of both the main
and the subordinate verb, the degree of adjacency between main and subordinate verb, as well
as whether the clause is negative or affirmative. Phonological variables taken into
consideration are, among others, the number of syllables of the entire VP and the
phonological interface between the two verbs contained in the VP. In addition to such purely
linguistic features, sociolinguistic variables such as age, sex, geographical provenance and

social status of the author, the type of relationship between author and intended reader, as well
as the degree of privacy or formality of the correspondence are also included.
Departing from this raw data, each factor is tested for statistical significance, with particular
attention to changes in significance over time. This variationist approach provides some
intriguing insights into how the relative importance of some of these factors has changed over
the centuries. For instance, when comparing the modal preiphrases deber + inf. and deber de
+ inf., in the 16th and 17th century, the most decisive factor is the semantic distinction
between deontic and epistemic modality, but in more recent times, variables such as the tense
of the main verb or polarity (negative/affirmative) have superseded the former in terms of
significance. On the other hand, perhaps even more intriguingly, some sociolinguistic factors
such as the age of the speaker/author have retained a similar degree of relevance over the
centuries, with younger speakers using deber de more frequently than their older
contemporaries to such an extent that age has been a statistically significant variable since the
16th century.
Beyond allowing us to assess the relative importance of a wide range of individual factors, the
corpus-based approach to language variation and change presented in this paper makes it
possible to undertake a diachronic multivariate analysis by taking into account the
simultaneous effect of all relevant independent variables (Labov, 2004). The great advantage
of such multivariate analyses is that they can identify the complex interplay between different,
at first sight often unrelated factors to cast light on what it is that really drives a particular
linguistic change. A number of examples in which different variables conspire to trigger a
preference for a certain construction in particular contexts will be presented, followed by a
discussion of how such multivariate analyses could, in future, be facilitated by incorporating
additional information into annotated corpora.
References:
Blas Arroyo, Jos Luis & Margarita Porcar Miralles. 2013 (in press). Patrones de variacin y
cambio en un rea de la sintaxis del Siglo de Oro. Un estudio variacionista de dos
perfrasis modales, RILCE, Revista de Fillologa Hispnica.
Koch, Peter & Wulf Oesterreicher. 1985. Sprache der Nhe Sprache der Distanz
Mndlichkeit und Schriftlichkeit im Spannungsfeld von Sprachtheorie und
Sprachgeschichte. Romanistisches Jahrbuch 36: 15-43.
Gmez Manzano, Pilar. 1992. Perfrasis verbales con infinitivo (valores y usos en la lengua
hablada). Madrid: UNED.
Labov, William. 2004. Quantitative reasoning in linguistics, in Ulrich Ammon, Norbert
Dittmar, Klaus J. Mattheier, and Peter Trudgill (eds.). Sociolinguistics/Soziolinguistik: an
international handbook of the science of language and society. Volume I. Berlin: Mouton
de Gruyter, pp. 6-22.
Maeseneer, Rita. 1998. Sobre algunos problemas relacionados con las perfrasis obligativas,
Linguistica Antverpiensia, 32: 39-53.
Oesterreicher, Wulf. 2004. Textos entre inmediatez y distancia comunicativas. El problema
de lo hablado escrito en el Siglo de Oro, in Cano Aguilar, Rafael (ed.), Historia de la
lengua espaola. Barcelona: Ariel, pp. 729-769.

Interaction between Discourse Structure and Clause

Evidence from a Corpus Study in Early New High German
Ulyana Senyuk
University of Potsdam

While there is widespread agreement in the literature that a discourse represents a hierarchical construct
with either coordinating or subordinating relations between discourse units (see e.g., Mann & Thompson
1988, Asher & Lascarides 2003), there is disagreement on how these relations interact with the syntactic
structure of a given piece of discourse. On the one hand, it is assumed that there is a relatively strong
correlation between discourse structure and syntax: coordinating and subordinating discourse relations are
mostly expressed by coordinating and subordinating clause combining (Mathiessen & Thompson 1988). On
the other hand, however, deviations from this strong correlation are pointed out showing that (i) clause
combining with and as a coordinating connector does not necessarily convey a coordinating discourse
relation (Ramm & Farbicius-Hansen 2005) and that (ii) clause combining with a genuine subjunction can
also imply a coordinating discourse relation (Delort 2008).
The goal of this study is to make a contribution to this discussion by examining different types of clause
combining from the later period of Early New High German (1350-1650) like in (1-a-c): Here all three
highlighted clauses convey the same semantic relation (namely temporal) with respect to the preceding
clause. This relation is encoded by clause-initial pronominal expressions. Syntactically, however, the
clauses in (1) seem to have a different status: In (1-a) we obviously have a main clause with the finite verb
in second position, in (1-b) we have a dependent clause, a so-called continuative relative clause, introduced
by a wh-word and displaying verb final order. Example (1-c) illustrates a so called relative-like clauses
whose syntactical status is still controversial (Author, in preparation): it is introduced by a demonstrative
and shows verb final order typical for dependent clauses.
(1)

a.

b.

c.

Dieses alles soll in Gegenwart der Knigin und anderer grossen Damen geschehen. Hierauff
(M,128)
wird der Knig nach den Frontiren verreisen.
This all should happen in the presence of the queen and other greatest ladies. Hereupon the
king will go to the border region.
Am vergangenen Sonntage ist die Pbstliche Krhnung mit gewhnlicher Solenniteten gesche(M,230)
hen. Worauff der Pabst das Volck gesegnet [...] hat.
On last Sunday the papal coronation took place with a habitual ceremony, whereupon the
pope blessed the people.
Den 29. dieses [...] ist die Auwechselung der Ratificationen zu Breda solenniter geschehen.
Darauff die Publication eine Danck Predigt [...] gehret/
(M,140)
On the 29th of this month the replacement of the ratifications of Breda took solemnly place.
Hereupon/whereupon the audience heard a sermon of thanks.

In my talk I will address the following question: To what extent can the distribution of the illustrated clause
types be explained by their discourse status? I will follow up this question by using the diagnostics by Asher
& Lascarides (2005). One of these diagnostics is, e.g., the type of discourse continuation. Preliminary
results suggest that there is a discourse-structural difference between clauses with verb second order like
in (1-a) and clauses with verb final order like in (1-b-c): The latter can be regarded as being deeper
embedded in the discource than the former. These findings seem to support the correlation assumption
mentioned above. Finally, I will also discuss the implications of this study for the issue of the syntactic
status of relative-like clauses like in (1-c).
References
Asher, N. & Lascarides, A. (2003): Logic in Conversation. Studies in Natural Language Processing. Cambridge: Cambridge University Press.
Asher, N. & Vieu, L. (2005): Subordinating and Coordinating Discourse Relations. In: Lingua 115, 591-610.
Author (in preparation): Zum Status relativhnlicher Stze im Frhneuhochdeutschen. (PhD Thesis).
Delort, L. (2008): Exploring the Role of Clause Subordination in Discourse Structure: The Case of French avant que. In: C. Fabricius-Hansen &
W. Ramm (eds): Subordination vs. Coordination in Sentence and Text. Amsterdam, Philadelphia: Benjamins, 241-255.
Mann, W. C. & Thompson, S. A.: (1988): Rhetorical Structure Theory: Toward a Functional Theory of Text Organisation. In: Text 8 (3),
243-281.
Matthiessen, C. & Thompson S. A. (1988): The Structure of Discourse and Subordination. In: J. Haiman & S. A. Thompson (eds): Clause
Combining in Grammar and Discourse. Amsterdam, Philadelphia: Benjamins, 275-329.
Ramm, W. & Fabricius-Hansen, C. (2005): Coordination and discourse-structural salience from a cross-linguistic perspective. In: M. Stede, C.
Chiarcos, M. Grabski & L. Lagerwerf (eds): Salience in Discourse: Multidisciplinary Approaches to Discourse 2005. Mnster: Nodus Publ./
Amsterdam: Stichting, 119-128.

From parenthetical to main clause: The case of the problem is in the history of American English
Reijirou SHIBASAKI
Meiji University

Discourse markers (DMs, hereafter) have been well explored in the history of English. However,
the vast majority of preceding studies on DMs or comment clauses focus on the earlier stages of
English (e.g. Brinton 1996, 2008). This study pays attention to the recent development of DMs
especially in Late Modern through Present Day English, with special focus on the problem is,
based on The Corpus of Historical American English (COHA, hereafter) and The Corpus of
Contemporary American English (COCA, hereafter); other corpora are used as needed basis.
The main survey results are as follows. Unlike comment clauses, the problem is shows the
recent development as a DM. Furthermore, the problem is appeared as a parenthetical (i.e. DM)
like the problem is, (c. 1837), developing into the problem is, that-complement (c. 1879) and
then into the problem is that-complement (c. 1900). The changing direction from parenthetical to
main clause reflects syntactic tightening, because the problem is takes on a new dimension as a
main clause. Bybee (2001) states that main clauses are innovative and subordinate clauses are
conservative; her arguments prove to be much to the point from a cross-linguistic perspective.
However, what Bybee (2001) points out is not a syntactic rising of main clause but a functional
expansion of main clause, for example, as seen in the advent of comment clauses. On the other
hand, what this study uncovered i.e. the process from parenthetical to main clause has
received but scant attention, probably because it is in direct conflict with the outcomes of
grammaticalization research. As a consequence, it has not come under close scrutiny.
In addition, this study addresses other related constructions such as the thing is, the fact is, the
point is, etc., in order to specify the directions of change in this particular construction i.e. (the)
NOUN is (that). For example, the thing is developed: the thing is, (c. 1796-1801) > the thing is,
that-complement (c. 1821-22) > the thing is that-complement (c. 1920), while the fact is followed
the opposite direction: the fact is that-complement (c. 1770) > the fact is, that-complement (c.
1775) > the fact is, (c. 1791). The vicissitudes of complement types especially at the dawn of
constructional expansion is neither regular nor unidirectional, which does not necessarily give
full support for a theory of grammaticalization.
Brinton, L. J. 1996. Pragmatic markers in English: Grammaticalization and discourse functions.
Berlin and New York: De Gruyter Mouton.
Brinton, L. J. 2008. The comment clause in English: Syntactic origins and pragmatic development.
Cambridge: Cambridge University Press.
Bybee, Joan. 2001. Main clauses are innovative, subordinate clauses are conservative. Complex
Sentences in Grammar and Discourse, eds. by Joan Bybee & Michael Noonan, 1-17.
Amsterdam: John Benjamins.

The Three Masculine Musketeers and the case morphology of Middle Danish
Eva Skafte Jensen esj@dsn.dk
The Danish Language Council
In 1956 the Danish grammarian Aage Hansen declared that the three nouns thiufr 'thief', nithr
'kinsman' and kostr 'object' ought to be called "The Three Masculine Musketeers". His reason for
doing so was the common pratice in grammars of Middle Danish of stating as a fact that nouns of
the strong declensions displayed a four way case distinction: the nominative, the accusative, the
dative and the genitive. However, in these grammars, the nominative was always represented by
one of the three mentioned words, as the majority of the nouns of the strong declensions did not
have a separate nominative form distinguishable from the accusative hence the label "The Three
Masculine Musketeers".
Hansen critisized the grammarians for conducting squinting grammar (cf. Jespersen), and to some
extent the critique was justified, but the four way paradigm wasn't a complete fabrication on the part
of the accused grammarians the forms were actually there in the manuscripts for everyone to see.
This calls for a methodological discussion of what to count as a part of a morphological system at a
given time. In order to uncover productive vs. relict forms it is necessary to conduct a thorough
investigation of the forms belonging to the morphological paradigms in a given text material.
This I have done. Based on the case forms found in the Scanic Law of approximately 1350, I've
mapped out the paradigms for nouns, numerals, adjectives and pronouns (cf. Jensen 2011), and in
this paper I will present my findings. I will show how within the nominal paradigms there are
subparadigms of four way, of three way and of two way distinctions. For some words, even, one
form seems to suffice.
In addition, I will show how my investigations reveal that the different nominal declensions follow
different paths. In the strong declensions of the nouns, for instance, there is a tendency towards a
complete syncretism between the nominative and the accusative, the old accusative being the
extensive case form ousting the nominative. In most of the pronouns a similar tendency towards
syncretism between the nominative and the accusative can be observed, but contrary to the situation
pertaining to the nouns, in the pronoun paradigms, the old accusative gives way to the nominative
form.
Hansen, Aage (1956). Kasusudviklingen i dansk. I: Festskrift til Peter Skautrup. Aarhus:
Universitetsforlaget i Aarhus. pp. 183-193.
Jensen, Eva Skafte (2011). Nominativ i gammelsknsk - afvikling og udviklinger. Kbenhavn:
Universitets-Jubilets danske Samfund & Syddansk Universitetsforlag.

A Phylogenetic Analysis of the Ancient Greek Dialects

Christina Skelton
Program in Indo-European Studies, University of California, Los Angeles
This paper aims to show how phylogenetic systematics can be used to help improve
our understanding of both the treelike and non-treelike development of a language family or
dialect group, using the dialects of ancient Greek as an example.
The development of the Greek dialects is generally well-understood, with the
exception of well-known questions such as the existence of an Aeolic subgroup (Colvin 2007,
40-41), how Proto-Greek developed into the major Greek dialect groups (e.g. Risch 1955),
and the subgrouping of some dialects within the major dialect subgroups, especially West
Greek (Colvin 2007, 44-45). It is generally agreed that the early development of the Greek
dialects into the major dialect groups was, at some level, fundamentally treelike. However,
dialects of different dialect groups came to be adjacent, and they exchanged linguistic features
through borrowing and diffusion of linguistic change.
A phylogenetic analysis of the Greek dialects using Maximum Parsimony successfully
recovered the four major dialect groups, as well as much of their internal structure and their
higher-order subgrouping. All phylogenetic characters with a consistency index lower than 1
were subjected to a statistical and a geographical analysis to identify patterns of borrowing
and diffusion between adjacent dialects. One analysis used these phylogenetic characters to
create a GIS dialect map of Greece, which was qualitatively examined for patterns. A second
analysis, designed to examine the directionality of borrowing, calculated the conditional
probability of each dialect showing a borrowed dialect feature given that an adjacent dialect
also showed that feature. For two dialects, one of which tends to create innovations and the
other which tends to borrow them, the borrowing dialect would be expected to share a very
high proportion of its non-treelike linguistic features with the innovating dialect, since the
innovating dialect would be its major source of innovations. However, the innovating dialect
would share a lower proportion of its linguistic features with the borrowing dialect, since the
borrowing dialect would show more innovations, some of which had not spread to the
borrowing dialect. Thus, the conditional probability of the borrowing dialect given the
innovating dialect should be higher than the conditional probability of the innovating dialect
given the borrowing dialect.
These analyses correctly identified the dialects and regions that scholars believe were
involved in significant borrowing and diffusion of linguistic features. Thus, this example
shows that phylogenetic analysis can be used to successfully reconstruct the evolution of at
least some examples of languages and dialect groups that show extensive amounts of nontreelike evolution, and that further analysis of the homoplastic characters can reveal the
patterns of borrowing and diffusion between taxa.

Colvin, S. C. 2007. A Historical Greek Reader: Mycenaean to the Koine. Oxford: Oxford
University Press.
Risch, E. 1955. Die Gliederung der griechischen Dialekte in neuer Sicht. Museum Helveticum
12: 6175.

Thomas Smitherman
University of Bergen
On the Methodology of Argument Structure Reconstruction
The ability of historical linguists to reconstruct syntactic forms with any degree of accuracy is a matter for
current debate. There has been recent work on the application of the Comparative Method to syntax within the
IECASTP project (see Bardal and Smitherman, forthcoming), suggesting the use of Construction Grammar and
constructicons. This represents an attempt at scientific progress but leaves some unanswered questions on the
proper order of procedure and the approximate weight to give different units considered in such a complex
reconstruction.
This paper will discuss issues of procedure and methodology in the attempt of the reconstruction of
argument structure in proto-languages, with a focus on the experience of the IECASTP project in the comparison of
oblique non-canonical syntax in Indo-European languages. We can see that at least for certain branches of IndoEuropean like Germanic (Bardal and Eythrsson 2003) and Slavic (Grillborzer 2010, Smitherman 2010) that we
should be able to reconstruct oblique subjects at least to certain proto-branches. The next, more difficult step is the
comparison of these branch-level data to create a Proto-Indo-European reconstruction. For example, in previous
papers, it has been argued that there was a PIE construction: NPDAT + *nh3-to/no- + [NPNOM] with the
meanings, I know of X; I am familiar with X; I understand X, as opposed to the active-verb form in a canonical
(NOM-ACC) construction with the meanings, I know X; I am finding out X. The basis for this reconstruction is
attestations in Greek, Latin, Old Slavonic, and Old English. This paper will address the objections to such
reconstructions with reference to the posited history of the word-form(s) of the predicate from its PIE origin to the
daughter
branches
and
the
particular
syntax
of
the
attested
argument
structures.
In addition, the paper will address methodological difficulties in the reconstruction of argument structure.
Such problems include:
1) What is an acceptable syntactic reconstruction given that one is dealing with all possible reconstructed units of a
language (phonemes, morphemes, lexemes, semantics, and syntactic units)?
2) Cognate roots rather than full stems are usually sufficient for the reconstruction of PIE verbs at a lexical level.
To what degree might affixation on a predicate verb (or, sometimes, a noun or adjective) skew the reconstruction of
a syntactic unit that is to say, what morphological derivations on a predicate may create reasonable doubt that an
oblique-subject construction, in which that predicate is involved in a given language or languages, be inherited?
3) To what degree might one reconstruct a category without (many) specific examples? In the IE case, could the
relative stability across several IE branches of the semantic fields in which dative- and accusative-subject
constructions occur allow, by itself, for the reconstruction of these categories with the implication that many
predicates that may have been involved in these constructions in Late Proto-Indo-European were renewed in the
given branches?
4) What is the proper method for proceeding must one first investigate subjecthood properties of ancient IE
languages and propose such properties for PIE, or are the traditional syntactic subjecthood tests too restrictive, i.e.,
would they bias the reconstruction of PIE towards a certain morphosyntactic alignment type (namely, accusative
alignment)?
Suggestions in response to such methodological problems will be proposed in the context of IECASTP experience.
References:
-Bardal, Jhanna & Thomas Smitherman. forthcoming. The Quest for Cognates: A Reconstruction of Oblique
Subject Constructions in Proto-Indo-European. To appear in Language Dynamics and Change 3(1).
-Bardal, Jhanna & Thrhallur Eythrsson. 2003. The Change that Never Happened: The Story of Oblique
Subjects. Journal of Linguistics 39: 439472.
-Grillborzer, Christine. 2010. Dative Subjects in Russian Modal Constructions Synchronic and Diachronic
Account. A talk delivered at the conference Subjects in Diachrony: Grammatical Change and the Expression of
Subjects in Regensburg, 34 December.
-Smitherman, Thomas. 2010. An Analysis of Dative Subjecthood in Old Russian An Inherited Construction?.
Paper delivered at workshop on Variation and Change in Argument Realization. Naples-Capri, 27-30 May 2010.

Plural Formation between Latin and Gallo-Romance:

a case study in typological divergence
John Charles Smith
University of Oxford
(St Catherines College & Research Centre for Romance Linguistics)
A general consensus concerning the evolution of Latin into Romance is that it involves a consistent shift in
morphosyntactic typology. This is sometimes expressed in the over-simplistic formula: Latin is synthetic,
Romance is analytic; more subtly, and more accurately, Coseriu (1988) argues that, whilst Latin is essentially
synthetic, Romance languages retain synthetic exponence for non-relational categories, whilst innovating by
adopting analytic exponence for relational categories.
In this paper, I examine a single phenomenon (plural formation in nouns) in a single branch of the Romance
family (Gallo-Romance: i.e., French, Occitan, and Franco-Provenal) and show that, contrary to claims made at
the macrolinguistic level, at this microlinguistic level there is no coherent typological evolution between
Latin and the modern languages. Dryer (2005a; 2005b) examines a sample of 957 languages and distinguishes
eight mechanisms of plural formation in descending order of frequency: plural suffix; plural word; plural
prefix; no distinct plural; plural clitic; reduplication of stem; change of stem; plural tone. In Latin, most nouns
form their plural suffixally, although some nouns, such as those of the fifth declension, give the impression of
having no distinct plural (compare nom. sg. RES thing, nom. pl. RES things); but this is true only of the
nominative if all case-forms are taken into account, then singular and plural have distinct suffixal paradigms.
Gallo-Romance, in contrast, arguably manifests every one of the types of plural formation enumerated by Dryer.
Suffixal plurals of the Latin type continue to be frequent; they may involve a zero/suffix alternation, as in many
Occitan varieties and some Norman varieties, or an alternation between singular and plural suffixes, as in other
varieties of Occitan and Franco-Provenal (Ronjat 1937). On the other hand, as a result of regular sound change
(chiefly the loss of final [-s]), the plural of most nouns no longer has a distinct form in standard French, many
French dialects, and some varieties of Occitan. However, as is well known, the mark of plurality may often be
carried by another element in the phrase, especially a determiner (Epstein 1994) compare French [la] the
cat vs. [lea] the cats corresponding to Dryers mechanism of a plural word (which need not have the sole
function of marking the plural). The minority of audible plurals in these varieties are best analysed as involving
a change of stem (compare French [j] eye, [j] eyes). Most recently, Sauzet (2011) has shown that singular
and plural in the Occitan variety of Sant Jlia de Cremsa are distinguished by different phonological tones. In
addition, I shall demonstrate that the sandhi phenomenon of liaison in French (Encrev 1988) can be analysed as
creating a prefixal plural form in certain contexts and a plural clitic in others, and that reduplication is involved
in the formation of emphatic plurals in a number of Gallo-Romance varieties.
The detailed survey of the data shows that a variety of phonological and morphological changes have
resulted in great typological divergence in the exponence of plural number amongst the varieties traditionally
referred to collectively as Gallo-Romance. It follows that, whilst this term may be a convenient label for the
Romance languages which emerged from the Latin spoken in Gaul, we should be wary of interpreting it as
implying that the varieties involved have necessarily undergone a consistent typological development from
Latin and now possess typological unity or even typological coherence.
References
Coseriu, Eugenio, 1988. Der romanische Sprachtypus: Versuch einer neuen Typologisierung der romanischen Sprachen. In
Energeia und Ergon: Sprachliche Variation, Sprachgeschichte, Sprachtypologie. Band I, Schriften von Eugenio
Coseriu (19651987), ed. J. Albrecht, 207-224. Tbingen: Narr.
Dryer, Matthew S. 2005a. Coding of nominal plurality. In The World Atlas of Language Structures, ed. Martin
Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie, 138-141. Oxford: OUP.
Dryer, Matthew S. 2005b. Occurrence of nominal plurality. In The World Atlas of Language Structures, ed. Martin
Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie, 142-145. Oxford: OUP.
Encrev, Pierre. 1988. La Liaison avec et sans enchanement. Paris: Seuil.
Epstein, Richard. 1994. The development of the definite article in French. In Perspectives on Grammaticalization, ed.
William Pagliuca, 63-80. Amsterdam & Philadelphia: John Benjamins.
Ronjat, Jules. 1937. Grammaire istorique [sic] des parlers provenaux modernes. Tome III: Deuxime Partie,
Morphologie et formation des mots; Troisime Partie, Notes de syntaxe. Montpellier: SLR.
Sauzet, Patric. 2011. Los morfmas de plural nominal a Sant Jlia de Cremsa : [-w] e lo ton bas. In Actes du 9e Congrs
de lAssociation Internationale dtudes Occitanes., ed. A. Rieger, 827-842, Aachen: Shaker.

Katerina Somers
Queen Mary, University of London
Analogy and innovative agreement in the history of German

This paper fits into the larger body of scholarship on the grammaticalization of verbal inflection in West Germanic
and looks to more concretely establish the diachronic link between the introduction and extension of innovative
agreement, on the one hand, and the development of complementizer agreement, on the other.
Somers (2009) offered a data-driven account of the distribution of the innovative second person singular ending
(-st) in Old High German (OHG) which made reference both to the syntactic environment in which the innovative
ending tended to occur, as well as the particular phonological shape of those endings that often exhibited the new
-st over the original -s. Specifically, this work showed that both syntactic position and ending shape were crucial in
predicting where the innovative ending might surface in those historical texts in which a mixture of the old and
new agreement suffixes were attested. The current paper will take a diachronic perspective and argue that the OHG
data can be interpreted in a way that is consistent with analyses presented in works such as Kathol (2001) and De
Vogelaer (2010). Namely, the verb that surfaces in a position in which pronominal encliticization is possible is
more likely to show an innovative agreement suffix, which subsequently can be extended along analogical lines
beyond its original syntactic environment.
Where the OHG data diverge from Kathol (2001) and De Vogelaer (2010) is that the data do not show evidence of
the intermediary stage that is part of these works analogical pathway to full grammaticalization of the new suffix:
(1)

An analogical pathway for innovations in the position enclitic to the verb (from De Vogelaer (2010:7), who
draws from Kathol (2001:65-71))
enclitic to:
Vinfl in VS-clauses > Complementizer > Vinfl in SV-clauses > Vinfl in subclauses

Though in the transitional texts of OHG we see evidence of the new suffix in VS-clauses, as well as is SV-clauses and
Vfinal clauses, there is no evidence of extension of the enclitic cluster to the complementizer. That this step of the
pathway is not attested, despite the fact that the new suffix becomes fully grammaticalized well before the end of
the OHG period, in addition to the fact that these transitional OHG texts do show evidence of clitic groups
comprising complementizer hosts, is puzzling and will also be a topic of investigation.

Works cited:
De Vogelaer, G. 2010. Morphological change in continental West Germanic: Towards an analogical map.
Diachronica 27:1.1-31.
Kathol, A. 2001. Syntactic categories and positional shape alternations. Journal of Comparative Germanic
Linguistics 3.59-96.
Somers, K. 2010. The introduction and extension of the -st ending in Old High German. Journal of Germanic
Linguistics 23.2.41-81.

An extra-demonstrative source for demonstrative system expansion

Lameen Souag, CNRS
The existence of person-oriented demonstrative systems has been recognised for over a
century (Brugmann 1904). In such systems, the proximal and medial terms respectively refer
to near speaker and near hearer; as Greenberg (1985:277) put it, person becomes the
model for place as the icon.
Diachronically, this suggests that a person-oriented demonstrative system could develop from
a non-person-oriented one by incorporating elements of the person marking system. However,
such a phenomenon has not been widely reported. The synchronic incorporation of person
markers in the system would entail demonstrative agreement with the addressee, an iconic but
typologically almost unknown phenomenon. The only well understood way that new
adnominal demonstratives emerge is by the addition of locative adverbs here, there, often
reinforcing the original demonstrative, occasionally replacing it (Heine & Kuteva 2002:172,
295). Such a change does not bridge the gap between a person-oriented system and a distanceoriented one, since locative adverbs are themselves demonstrative as Diessel (fc) notes in
defense of his claim that demonstratives cannot be traced back to lexical expressions.
This implies that there must be other ways of re-orienting demonstrative systems, and closer
examination bears out the prediction above: at least three languages display addressee
agreement on the demonstrative, and in all three cases this appears to be the outcome of the
formation of a new distance (or replacement of an old one) by the incorporation of a 2nd
person marker, a change also attested beyond these languages. The clearest data on this can be
drawn from Siwi and other eastern Berber languages, which display these phenomena in a
particularly convenient form; Semitic and Italian, however, also provide important data, while
extra-Semitic parallels in turn cast light on the century-old debate about the origins of Semitic
distal -ka (Barth 1913). Attested means of incorporating 2nd person markers into the
demonstrative system include, at least, the extension of an existing demonstrative with a
locative adverbial near you, and the extension of a presentative with a dative to you.
Given the existence of the grammaticalisation pathway in question, the cross-linguistic rarity
of addressee agreement on demonstratives appears unexpected. However, placing this within
a broader typology of allocutivity across two syntactic dimensions host category, and host
function allows us to explain its rarity in terms of a cross-categorial generalisation: since
allocutivity is a property of the whole utterance, it is preferentially marked at the level of the
main clause, rather than in embedded contexts such as arguments or modifiers.
References
Barth, Jakob. 1913. Die Pronominalbildung in den semitischen Sprachen. Leipzig: Hinrichs.
Brugmann, Karl. 1904. Die Demonstrativpronomina der indogermanischen Sprachen.
Leipzig: Teubner.
Diessel, Holger. fc. Where do grammatical morphemes come from? On the development of
grammatical markers from lexical expressions, demonstratives, and question words. In Kristin
Davidse, Tine Breban, Liselotte Brems, & Tanja Mortelmans (eds.), New Reflections on
Grammaticalisation. Amsterdam: John Benjamins.
Greenberg, Joseph H. 1985. Some iconic relationships among place, time, and discourse
deixis. In John Haiman (ed.), Iconicity in Syntax. (Typological Studies of Language 6).
Amsterdam: John Benjamins.
Heine, Bernd & Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge:
Cambridge University Press.

Sverre Stausland Johnsen

University of Oslo

Vowel reduction in Old English

Introduction. An original unstressed long * generally develops to a in Old English (OE). In some morphological categories, however, it appears both as u and a. Previous literature suggests that * was raised to
u if the following syllable historically had *u, otherwise it developed to a, then followed by generalization in
either direction. I hypothesize that * reduced to u in synchronic medial syllables due to the short duration of
such syllables, irrespective of what vowels might have followed historically, and that * regularly developed
to a in nal syllables. A statistical study of verbal forms in the largest early OE text strongly supports this
hypothesis.
van Heltens rule. An original unstressed * generally develops to a in OE, cf. *dags > dagas days and
*wundi > wunda woundeth. In the past tense forms of -verbs, however, both u and a are found: pret.
andswarude ~ andswarade answered, perf. wundud ~ wundad wounded. van Helten suggested that *
had been raised to u when the following syllable originally contained a *u, e.g. perf.f.nom.sg. *wunddu >
wundud, but otherwise developed to a, e.g. perf.m.nom.sg. *wunddaz > wundad. All main grammars of OE
accept this theory, adding that both u and a were later generalized beyond their original distribution.
Vowel reduction. Unstressed vowels are shorter than stressed vowels and high vowels are shorter than
low vowels. e articulators need to travel a longer distance for low vowels than for high vowels, so low
vowels require more time to be realized. e extra time needed to articulate low vowels is antagonistic to
the reduced time alloted to vowels in unstressed syllables. It is therefore oen found that mid/low vowels
reduce to high vowels in unstressed syllables. Importantly, unstressed vowels are shorter in medial syllables
than in nal syllables, so the likelihood of vowel raising should be even greater in medial syllables.
New hypothesis. Based on these properties of vowels in unstressed syllables, I suggest that an original
unstressed * reduced to a short high u in synchronic medial syllables, and developed to a in synchronic nal
syllables. Despite later generalizations, we should be able to detect this paern in the data.
Corpus.
text.

All past tense forms of -verbs (n = ) were gathered from Ms Haon , the largest early OE
u in mid and final syllables
90%
Forms with u
75% 80% 85%
Before *u

Elsewhere

Fig. : van Heltens rule

70%

70%

Forms with u
75% 80% 85%

90%

u before original *u

mid

final

Fig. : New hypothesis

Results. ere is no support for van Heltens rule (Fig. ). e u is actually less common before an original
*u than elsewhere, but this dierence is not signicant (mixed eects logistic regression, p = .). Fig. shows
that there is good support for the new hypothesis u is signicantly more common in medial syllables
(p < .).
Conclusion. OE follows a cross-linguistic tendency of reducing shortened unstressed vowels to high vowels. Based on the verbal forms gathered from the largest text of early OE, a paern can be detected in which
an original unstressed long * reduced to a high u in shortened medial syllables, but developed to a in the
longer nal syllables.

Moving out of PPs in the history of Icelandic

Brynhildur Stefnsdttir and Jhannes Gsli Jnsson, University of Iceland
In this paper, we report the results of our investigation of movement out of PPs in the history
Icelandic, using the IcePaHC Corpus. Movement out of PPs involves movement of the DP object
(P-stranding) or movement of the preposition (DP-stranding). Through the history of Icelandic,
these two phenomena seem to be in an inverse relation. Whereas DP-stranding is attested in Old
Icelandic, P-stranding was more or less excluded but the situation is reversed in Modern Icelandic,
where P-stranding is possible but DP-stranding is out.
1. P-stranding
The oldest clear-cut example of P-stranding in the corpus is from 1525, as shown in (1). This is not
surprising in view of the fact that Old Icelandic had very few if any examples of P-stranding
(Haugan 2001, Delsing 2003, Faarlund 2004, and Jnsson 2008).
(1)

gaf barni
sr ekki neitt a
essum
orum
keisarans
these-DAT words-DAT the.emperor- GEN gave the.child REFL not at.all to
The child paid no attention to these words of the emperor
(ID 1525.GEORGIUS.NAR-REL,.866))

2. DP-stranding
Examples of DP-stranding found in the IcePaHC corpus can be roughly divided into two categories.
The first involves strings with preposition + verb/participle + object as in (2):
(2)

og hversu hann er fr rekinn snu

a saklausu
rki
and how
he
is from driven REFL.-DAT state.-DAT by innocence
and how he is wrongfully driven from his state
(ID 1275.MORKIN.NAR-HIS,.1882)

The corpus has almost 70 examples of this type before 1500. Younger examples are less than 40.
Examples like (2) can be analyzed as incorporation of the preposition into the verb (see Hrarsdttir
2000:247-252). This is supported by the fact that preverbal prepositions are often written as parts of
the verb in texts from the 17th and 18th centuries.
The second type of DP-stranding involves strings where the preposition precedes the object with (i)
many words in between or (ii) one word which is not a verb/participle. There are some 20 examples
of this kind in the corpus. In some cases, these strings result from applying Heavy NP Shift to the
object of the preposition:
(3)

himintunglin benda fyrir me gn og hryggleik au

strtendi,
the.planets
point to
with threat and sadness
those-ACC big.news.-ACC
er eftir munu koma
that later will come (ID 1300.ALEXANDER.NAR-SAG,.1118))

In other cases, it looks more like the preposition has moved to the left, and many of these examples
could be analyzed as some kind of incorporation of the preposition:
(4)

og hefir Oddur af viring mlunum

and has Oddur of respect the.affairs.and Oddur gains respect from the affairs
(ID 1450.BANDAMENN.NAR-SAG,35.555))

The development of early Middle English : spelling evidence

Gjertrud F. Stenbrenden, University of Oslo
This paper investigates (a) the backing and raising of Northern Middle English (ME) to ,
usually referred to as Northern Fronting (NF), and (b) the raising of Southern Middle English
to [u:] in the so-called Great Vowel Shift (GVS). Traditionally, NF is supposed to be the earlier
of the two changes and to have reached completion before the GVS started. The first stage of NF,
/o:/ > [:], is believed to have started c. 1300, and the next phase, [:] > [:], to have begun in the
fifteenth century. The early stages of the GVS are dated to c. 1400 by most scholars (Luick 1914
40: 406, 581; Jordan 1968: 53-54; Wright and Wright 1928: 55). Jespersen, however,
assuming that the raising started in unstressed syllables, dates the incipient stage to the fourteenth
century (1909: 8.31).
The spelling evidence has been abstracted from three ME corpora which collectively cover the
period c. 1150-1500: LAEME, LALME, and SMED. These corpora make use of texts in local
language, and the language of each text has, in most cases, been more or less precisely localised.
The paper aims (1) to assess the types of evidence used, and the problems which attend the
interpretation of orthographical material in terms of phonetic correspondences; (2) to discuss
whether the spellings in fact support the assumed developments of eME ; (3) to date and
determine the chronology of the changes to eME in Southern and Northern ME, with special
reference to the early vowel-shift hypothesis (Stockwell & Minkova 1988); and (4) to identify, if
possible, the dialects which stand out as loci of change. More recent claims regarding NF by
Britton (2002), Britton & Williamson (2002), and Williamson (2006) will also be examined.
References
Britton, D. 2002. Northern fronting and the north Lincolnshire merger of the reflexes of ME /u:/ and ME /o:/.
Language Sciences 24; pp. 221-229.
Britton, D. & K. Williamson. 2002 (ms.). A review of Northern Fronting and its developments in England and
Scotland. Paper read at the 12th ICEHL, Glasgow, Scotland, 23. August 2002.
Jespersen, O. 1909. A Modern English Grammar, Part I. Heidelberg: Carl Winters Universittsbuchhandlung.
Jordan, R. 1968. Handbuch der mittelenglischen Grammatik: Lautlehre. Heidelberg: Carl Winter
Universittsverlag.
Kristensson, G. 1967. A Survey of Middle English Dialects 1290-1350. The Six Northern Counties and
Lincolnshire [SMED1]. Lund: CWK Gleerup.
Kristensson, G. 1987. A Survey of Middle English Dialects 1290-1350: the West Midland Counties [SMED2].
Lund: Lund University Press.
Kristensson, G. 1995. A Survey of Middle English Dialects 1290-1350: the East Midland Counties [SMED3].
Lund: Lund University Press.
Kristensson, G. 2001. A Survey of Middle English Dialects 1290-1350: the Southern Counties. I. Vowels (except
Diphthongs) [SMED4]. Lund: Lund University Press.
Kristensson, G. 2002. A Survey of Middle English Dialects 1290-1350: the Southern Counties. II. Diphthongs
and Consonants [SMED5]. Lund: Lund University Press.
LAEME, see Laing, M. 2008.
Laing, M. 2008. A Linguistic Atlas of Early Middle English. University of Edinburgh.
http://www.lel.ed.ac.uk/ihd/laeme1/laeme1.html
LALME, see McIntosh, Samuels, Benskin et al. 1986.
Luick, K. 1914-1940. Historische Grammatik der englischen Sprache, Vol. I, Parts 1 & 2. Oxford: Basil
Blackwell.
McIntosh, A., M.L. Samuels, M. Benskin et al. 1986. A Linguistic Atlas of Late Mediaeval English, Vols. I-IV.
Aberdeen: Aberdeen University Press.
SMED, see Kristensson 1967, 1987, 1995, 2001, 2002
Stockwell, R.P. & D. Minkova. 1988. The English Vowel Shift: problems of coherence and explanation. In:
Kastovsky, D. and G. Bauer (eds.)1988. Luick Revisited. Tbingen: Gunter Narr Verlag; pp. 355-394.
Williamson, K. 2006 (ms.). Further reflections on the outcomes of Northern Fronting in Older Scots. Paper
read at 14 ICEHL, University of Bergamo, 21-25 August, 2006.
Wright, J. & E.M. Wright. 1928. An Elementary Middle English Grammar. Oxford: Oxford University Press.

A DIACHRONIC ACCOUNT OF DATIVE BY GENITIVE REPLACEMENT

IN PAPYROLOGICAL GREEK
JOANNE STOLK - UNIVERSITY OF OSLO

Distinct morphological case marking for the dative case was lost between Classical and Modern
Greek and the functions of the former dative case were taken over by the genitive and accusative
cases and prepositional phrases. Interchange between the dative and the genitive cases is found
in the Greek of the documentary papyri (Humbert 1930), especially with pronouns that are
in a position in which they might belong to the preceding VP or the following DP (Horrocks
1997:121-2). The aim of this study is to investigate the overlap between the post-verbal dative
object pronoun and the pre-nominal genitive possessive pronoun in order to establish the semantic
and syntactic basis for the merger of the two cases.
Gianollo (2010:127) shows that the pre-nominal genitives in New Testament Greek overlap
with the distribution of the dative external possession construction. I examined all pre-nominal
first person singular genitive pronouns in the papyri (300BCE-800CE), but this did not reveal
external possession constructions. Instead, the data included examples of affected alienable
possessors (1), (malefactive) source constructions (2) and genitives functioning ambiguously as
possessor and benefactive (3) or possessor and goal (4).
(1) > | < they burned my (gen.) threshing floor (P.Petr.3.34, 5-6;
210-183BC)
(2) > he took away my (gen.) gold ornaments (SB 16.12326, 8; late
III)
(3) > | >
(l. >
) buy me/my (gen.) part of the
olive grove (BGU 2.602, 5-6; II)
(4)
| (l. ) | > give to the one who brings you
(gen.) the letter (P.Oxy.2.296, 3-4; I)
Therefore I argue that the origins of the merger of dative and genitive can be found in the semantic
extension of the genitive case from the role of an affected alienable possessor to the dative roles of
benefactive/malefactive and goal. Following the principles of Diachronic Construction Grammar
(e.g. Fried 2009), I also propose a reconstruction of the chronological stages of this change.
References
Fried, Mirjam. 2009. "Construction grammar as a tool for diachronic analysis", in: Constructions
and Frames 1:2, 262-291.
Gianollo, Chiara. 2010. "External possession in New Testament Greek", in: G. Calboli & P.
Cuzzolin, Papers on Grammar XI. Rome: Herder Editrice, 101-129.
Horrocks, Geoffrey. 1997. Greek: A History of the Language and its Speakers. London-New
York: Longman.
Humbert, Jean. 1930. La disparition du datif en grec (du Ier au Xe sicle). Paris: Champion.

The Evolution of Verbal Inflection in the Scandinavian Languages

Oscar Strik
University of Groningen
Since the Proto-Germanic period, the verb systems of the Germanic languages have been characterised by a
multitude of conjugation classes. These are generally categorised into a strong and a weak inflection type, each
with its own (sub)classes.
Compared to the other Germanic languages, the Scandinavian languages show many interesting developments
in verbal inflection. On the one hand, relatively isolated languages like Icelandic, Faroese, and Elfdalian have
retained much of the inflectional richness of the Old Germanic period. Remarkably, however, multiple weak
classes have been retained in all Scandinavian languages, while many non-Scandinavian Germanic languages
have only one weak class left (Frisian is a notable exception). Bokml and Swedish have even developed a new
weak class (Dammel 2009). On the other hand, we find simplifications in some other respects. For example,
there is loss of person distinction in the plural in Faroese verbs, and a wholesale loss of person/number
distinction in Danish, Bokml, and Swedish. Surprisingly, Elfdalian shows loss of person distinction in the
singular, but not in the plural.
We can contrast system-level developments such as the above with developments at the individual word level.
Throughout the history of the Scandinavian languages, we can find many verbs that have shifted to a different
inflection class entirely, or have acquired forms from another class in part of its paradigm. For example:
Old Norse snta ~ snttOld Swedish sitia ~ sitin
(Old Norse dumpa ~ dumpa-

>
>
?>)

Bokml snyte ~ snt

Modern Swedish sitta ~ suttit
Elfdalian dimpa ~ damp

blow ~ blew [ones nose]

sit ~ sat (sup.)
thump ~ thumped

I will argue that these individual changes are best explained as a result of analogical extension, based on
phonological similarities with other verbs. My preliminary research on this type of change in the history of
Swedish and Frisian indicates that analogical computer models (Skousen 1989, Albright & Hayes 2003) can be
used to predict a significant part of such developments.
Whether we are dealing with system-wide or verb-specific changes, there are a few key principles that govern
morphological changes in verbs. Drawing upon examples from the various Scandinavian languages (past and
present), I will argue that phonological changes can pave the way for analogical extension of patterns, as well
as levelling in paradigms. Analogical similarities between phonological verb forms can explain the direction of
individual inflection class shifts. At the same time, type and token frequency play an similarly important role in
determining the direction of such shifts, and even in determining whether a shift will take place at all. Finally,
sociolinguistic factors determine the spread and maintenance of innovative forms beyond the idiolects in
which they originate.
References:
Albright, A. & Hayes, B. (2003). Rules vs. analogy in English past tenses: a computational/experimental study. In: Cognition 90.
119161.
Dammel, A. (2009). How - and Why - Do Inflectional Classes Arise? A Case Study on Swedish and Norwegian Conjugation. In:
Montermini, Fabio (et al.) (ed.). Selected Proceedings of the 6th Dcembrettes. 1221.
Skousen, R. (1989). Analogical Modeling of Language. Dordrecht / Boston / London: Kluwer Academic Publishing.

Ulla Stroh-Wollin, Department of Scandinavian Languages, Uppsala University, Sweden

Strong and weak adjectives in definite noun phrases in Old Swedish

The weak declension of adjectives is a Proto-Germanic novation after which attributive adjectives
had two different guises for each combination of gender, number and case, a strong form and a
weak form. In the modern Scandinavian languages the distribution of strong and weak forms
correlates well with the indefinitedefinite distinction. However, the weak forms are rare in runic
inscriptions and still appear with low frequency in early medieval manuscripts, which means that the
obligatory use of the weak form in semantically definite noun phrases is of a later date. Surprisingly
little attention is paid to this fact in handbooks and, with few exceptions (Delsing 1994, Simke 2012),
elsewhere.
Delsing 1994 shows that adjectives following a possessive attribute did not necessarily appear in the
weak form in Old Swedish, contrary to Modern Swedish. The primary aim in Simke 2012 is to capture
the use of strong and weak adjectival forms in noun phrases in Old Swedish texts from c. 1300 to
c. 1450 in quantitative terms, and the survey shows a gradually decreasing use of strong adjectives in
semantically definite noun phrases during the investigated period of time.
I my research I try to follow up the mentioned, mainly descriptive studies from the theoretical point
of departure that adjectives presumably interrelate formally with other constituents within the noun
phrase. Thus, I take the distribution on strong and weak forms as one clue to a better understanding
of the process whereby the Scandinavian languages develop obligatory definiteness marking in noun
phrases, above all visible by means of definite inflection of nouns and formally definite determiners.
One approach to handle the formal interrelation between the constituents of a noun phrase is in
terms of feature sharing in the sense of the operation Agree, preferably understood as in Pesetsky &
Torrego 2007 (rather than e.g. Chomsky 2001).
The investigation is founded on excerpting semantically definite noun phrases in Old Swedish texts in
order to gain a detailed knowledge of their possible and, not least, impossible structures and to date
different steps in the process towards the modern patterns. In modern Swedish for instance a definite noun phrase must be headed by a definite article if the head noun is preceded by an adjective,
even though the noun itself is inflected for definiteness. Further, the adjective in such a phrase is
always weakly inflected. However, I have in Old Swedish texts found five different combinations that
may result in the same definite interpretation, whereas three patterns are not attested, cf. (1).
(1) Possible patterns:
Non-attested patterns:
no art. + strong adj. + unmarked noun
no art. + strong adj. + def. noun
def. art. + strong adj. + unmarked noun
def. art. + strong adj. + def. noun
def. art. + weak adj. +unmarked noun
no art. + weak adj. + unmarked noun
no art. + weak adj. + def. noun
def. art. + weak adj. + def. noun (= the modern pattern)

Chomsky, Noam. 2001. Derivation by Phase. Ken Hale: A Life in Language, Michael Konstowitz (ed.), 152.
Cambridge, MA: MIT Press.
Delsing, Lars-Olof. 1994. Hans sjukt ben Om starka och svaga adjektiv i fornsvenskan. In Sprkbruk,
grammatik och sprkfrndring: En festskrift till Ulf Teleman 13.1.1994, 99108. Department of
Scandinavian Languages, Lund University.
Pesetsky, David & Esther Torrego. 2007. The Syntax of Valuation and the Interpretability of Features. In Phrasal
and Clausal Architecture: Syntactic Derivation and Interpretation, [Linguistik Aktuell/Linguistics Today
101], Simin Karimi, Vid Samiian & Wendy K. Wilkins (eds), 262294. Amsterdam: John Benjamins.
Simke, Rico. 2012. Die Entwicklung des schwachen Adjektivs. Studie zur altschwedischen Nominalphrase.
Master thesis [unpubl.], Friedrich-Alexander-Universitt Erlangen-Nrnberg.

Wh-Interrogatives in Laurentien French: Extensive Variation, Diglossic Situation

Sandrine Tailleur
Universit du Qubec Chicoutimi
The cause of the extensive variation found in the Laurentien French [LaF] (that is, French spoken in
Canada excluding the Acadian dialects) wh interrogative system has never been satisfactorily
explained. In todays LaF, examples 1 to 6 below are all attested and equivalent, the choice of variant
being based solely on the discourse situation (1 is very formal; 2, 3 and 6 are fairly neutral; 4 and 5 are
more informal see Elsig 2009 for details on usage).
1. O vas-tu?
where go-you
2. O tu vas?
where you go
3. O est-ce que tu vas?
where is-it that you go
4. O cest que tu vas?
where it-is that you go
5. O que tu vas?
where that you go
6. Tu vas o?
you go where
Where are you going?

Wh-fronting with inversion

Wh-fronting without inversion
Wh-fronting with reinforcer
Wh-fronting with reinforcer
Wh-fronting with reinforcer
Wh-in situ

By looking at different types of data, historical as well as contemporary (written and oral corpora,
metalinguistic documents, etc.), this work shows that:
i.

ii.

the high degree of variation is due to the use of wh est-ce que, the interrogative reinforcer
(and all the variants that are derived from it, see 3-5 above) (which appeared as early as
Old French (11th century) as an interrogative cleft (Rouquier 2003 and others));
between Old and Modern French, wh est-ce que has gone through a typical cycle of
grammaticalisation (assuming a generative definition of grammaticalisation (Roberts &
Roussou 2003) as well as a theory of cyclic change (van Gelderen 2004)).

These grammatical changes over time led to todays LaF wh system, which is dominated by the wh
estce que and variants (over 98 percent of use Elsig 2009).
Moreover, the data presented in this paper provides strong evidence towards a theory of diglossia
within the French diaspora (Zribi-Hertz 2011). It seems to be the case that wh interrogative variants
are divided according to two fundamental structural differences; some have wh movement (high,
formal register; see Ex. 1 and (perhaps) 2) while others do not (vernacular and neutral register, see Ex.
3-6).
This divide between two groups of wh variants explain the multitude of options that are available to
French speakers, as well as provides a logical explanation for the lack of consensus among native
speakers when it comes to the grammaticality of wh in situ (see Mathieu 2004).

Selected References
Elsig, Martin. 2009. Grammatical Variation across Space and Time: The French interrogative
system, Amsterdam/Philadelphia : John Benjamins.
van Gelderen, Elly. 2004. Grammaticalization as Economy. Amsterdam: John Benjamins.
Mathieu, Eric. 2004. The mapping of form and interpretation: the case of optional WH-movement
in French. Lingua 114 : 1090-1132.
Roberts, Ian & Anna Roussou. 2003. Syntactic Change. Cambridge: Cambridge University Press.
Rouquier, Magali. 2003. La squence est-ce dans les interrogatives en qui/que en ancien et en
moyen franais. French Language Studies 13: 339-362.
Zribi-Hertz, Anne. 2011. Pour un modle diglossique de description du franais : quelques
implications thoriques, didactiques et mthodologiques. Journal of French Language studies
21.2: 231-256.

THE PENUTIAN PHYLUM AND THE COMPARATIVE METHOD

Marie-Lucie Tarpent
Mount Saint Vincent University (retired)
The label "Penutian" was extended by Sapir from a California grouping to a larger "phylum"
comprising a number of languages spoken further North (Sapir 1929). His few criteria were mostly
typological, along with a few morphemes, and did not always apply to all the component languages.
He thought that a genetic relationship was probable, but would remain "beyond the reach of the
comparative method." Since then, the "Penutian phylum" has been broken up into several smaller
groups along with a few isolates, yet a few researchers have continued to consider the "phylum" as a
potential genetic grouping. Obstacles to progress or denial on this front concern the availability of data
and the handling of these data in a comparative perspective.
The North American West Coast is a "residual zone" with many small and varied linguistic areas. In
California, considerable work of documentation was carried on in the second half of the twentieth
century, but farther North, documentation has been less adequate, and several of the languages became
extinct early. At least as important than the scarcity of documentation was the lack of adequate
comparative/historical knowledge among linguists trained mostly for fieldwork and description:
proto-language reconstruction works for languages which are already considered related because of
morphological similarities which strongly suggest a common ancestor. Outside of obvious linguistic
groups, reconstructing vocabulary while ignoring morphology leads to failure, as happened with
"Takelman", a proposed grouping of the Oregon languages Takelma and Kalapuya (Tarpent& Kendall
1998). Another problem was the lack of familiarity with the range of variation in phonological
change. For instance, Hymes made a pan-Penutian search for potential cognates, apparently finding
many among sets of words beginning with "*C and *S" (Hymes 1964). He did not realize that in
historical terms these sounds, especially [s], can be the end result of a long chain of changes over
centuries and millennia, and that the ancestors of the supposed cognates could have had quite different
initials.
In recent years there has been renewed interest in the Penutian phylum, helped by a number of new or
much better language descriptions (DeLancey & Golla 1997). In particular, Tarpent's independent
descriptive and reconstructive work in Tsimshianic, the northernmost and most isolated of Sapir's
Penutian families, has provided a basis for searching for common morphological features and
individual morphemes among the various families. This work has uncovered many widespread
grammatical morphemes and lexical roots, showing regular phonological correspondences, linking
Tsimshianic especially with languages of Oregon (Tarpent 1997). Far from being the "most
problematic" group in the "Penutian phylum" (Silverstein 1979), Tsimshianic appears to be a key
member. These still early findings make it likely that it will ultimately be possible to work on ProtoPenutian reconstruction.

REFERENCES
DeLancey & Golla 1997 The Penutian hypothesis: Retrospect and prospect. IJAL 63:171-202
Hymes 1964 Evidence for Penutian in lexical sets with initial *C and *S. IJAL 30:213-42
Sapir 1929 Central and North American Indian languages. Encycl. Brit. 14th ed. 5: 138-141
Silverstein 1979 Penutian: An assessment. in Campbell & Mithun, eds: The languages of Native America: A
historical and comparative assessment. Austin, Tex.
Tarpent 1997 Tsimshianic and Penutian: Problems, methods, results and implications. IJAL 63
Tarpent & Kendall 1998
On the relationship between Takelma and Kalapuya: Another look at
Takelman. SSILA meeting, NYC.

Expanding the basis for the genetic classification of Panoan languages

Daniel Valle

Adam Tallman
University of Texas at Austin

The Amazon region is among the most linguistically diverse areas of the world, and at the
same time among the least studied (Dixon and Aikhenvald 1999:1). There remains much work to be
done with regard to genetic relationships and sub-grouping within Amazonian language families. One
such understudied linguistic family is the Panoan language family. Currently, only 17 Panoan
languages, of the 30 known, are still spoken today in the Amazon regions of Peru, Brazil and Bolivia
(Fleck 2011). Previous classifications of Panoan languages have been carried out using lexicostatistics (dAns 1972), phonological reconstruction (Shell 1985), and a combination of lexical,
phonological and some grammatical features (Loos 1999; Fleck 2001; Valenzuela 2003). While some
patterns seem to link these proposals, consensus as to how to sub-group Panoan languages is still
lacking.
A widening of the spectrum of data seems to be the next step toward a deeper understanding of
the historical relationships among Panoan languages. Indeed, as Girard (1971) points out, Panoan
languages have undergone very few phonological innovations but much "morphological restructuring"
making morpho-syntax a place to look for a wider range of innovations. This study explores this idea
of focusing on grammar to better understand the internal relations Panoan languages share among
themselves by studying a total of 232 grammatical features such as constituent order, alignment,
valency-changing morphology, evidentiality, etc. The choice of the languages for this survey depended
on the availability of sources and the need to have a balanced sample, that is, one language from each
sub-group as suggested by the previous studies on this family. The languages chosen were: Shipibo
Konibo, Matses, Kakataibo, Amawaca, Chcobo, Cashinawa and Yaminawa. In addition, two non
Panoan languages, Quechua and Ashaninka, were included in the sample as a way to detect possible
contact influence since these languages have apparently had a long history of contact with Panoan
languages. The grammatical features were coded using the software SplitsTree4 (Huson and Bryant
2006).
Preliminary results support the most recent classifications based mainly on lexical items and
phonology (Fleck 2001). That is, Kakataibo, Matses and Chcobo seem to independently diverge from
a core group formed by the other Panoan languages studied here. While clustering languages on the
basis of typological features may pick up areal instead of phylogenetic relations (Donohue et
al.2006), this bias decreases when considering the results in combination with proposals based on non
grammatical data.

References
dAns, Andre-Marcel. 1972. Estudio glotocronolgico sobre nueve hablas pano. Lima, UNMSM.
Dixon, R.M.W. and A. Aikhenvald. 1999. The Amazonian Languages. Cambridge, CUP.
Donohue, Mark et al.. 2008. Typology, areality, and diffusion. Oceanic Linguistics 47:223-232.
Fleck, David. 2011. Panoan languages and linguistics. Ms.
Girard, Victor. 1971. Proto-Takanan phonology. UCPL, vol. 70. Berkeley: University of California Press.
Huson, D. and D. Bryant. 2006. Application of Phylogenetic Networks in Evolutionary Studies. Mol. Biol. Evol.,
23(2):254-267.
Loos, Eugene. 1999. Pano. In R.M.W. Dixon and A. Aikhenvald. The Amazonian Languages. Cambridge, CUP,
pp.227-250.
Shell, Olive. 1985. Estudios pano III: Las lenguas pano y su reconstruccin (2nd. ed.). Yarinacocha, ILV.
Valenzuela, Pilar. 2003. Transitivity in shipibo-konibo grammar. PhD Dissertation. Oregon, University of
Oregon.

A Rendaku Database for Old Japanese

Timothy J. Vance (NINJAL) and Mark Irwin (Yamagata University)
The term rendaku refers to a well-known set of morphophonemic alternations in modern
Japanese. In many cases, a morpheme that begins with a voiceless obstruent word-initially
appears with a voiced obstruent instead as a non-initial element in a compound, as in /tama/
ball and /me+dama/ eyeball. Old Japanese (OJ) is the language reflected in 8th-century
texts, the earliest substantial written records of Japanese, and rendaku was already
widespread. The attested OJ lexicon is fairly small and is covered quite thoroughly in a
single-volume dictionary (Jdai-go Jiten Hensh Iin-kai 1967). We are nearly finished compiling a database containing all the compounds that are listed as headwords in this dictionary
and that end with a noun element that provides a potential rendaku site. In phonological
research, attention necessarily focuses almost exclusively on the examples attested in phonograms in one of the important sources for OJ listed by Frellesvig (2010:23), and we will describe some of the problems of interpretation that OJ phonogram orthography presents. The
OJ rendaku database will be finished and publicly available by the time of ICHL21.
To demonstrate the value of the OJ rendaku database, we will use it to evaluate a proposal
involving Lymans Law, which in modern Japanese is a constraint that prohibits rendaku
from affecting a morpheme that contains a medial voiced obstruent. For example, /tabi/ trip
does not alternate with */dabi/ in compounds like /funa+tabi/ boat trip. In OJ, a stronger
version of Lymans Law held (Unger 1975:9; Vance 2005). The usual statement of the strong
version is that rendaku was also prohibited if the morpheme preceding the potential rendaku
site contained a voiced obstruent, as in OJ/midu+tori/ water bird. Miller (1984) interprets an
obscure passage in an 11th-century Buddhist text as a statement of an even stronger version of
the constraint. We will not address the question of whether Millers interpretation is correct.
Instead, we will investigate whether or not the constraint that Miller states actually held in OJ.
Although Millers statement is not entirely clear, there seems to be only one reasonable way
to construe it: in addition to the strong version of Lymans Law, a sonorant consonant in the
syllable immediately preceding the potential rendaku site also prevented rendaku. Miller adds
some provisos that effectively insulate his claim from disconfirmation, but we have chosen to
treat it as an empirical statement.
By examining all the relevant compounds in our OJ rendaku database, we can show that
Millers version of the constraint did not hold in OJ. Preliminary figures (restricted to compounds in which the final element has one or two syllables) show rendaku in 58 of 159
unproblematic examples with a sonorant consonant in the last syllable of the first element. We
will report final figures that include compounds with second elements longer than two syllables, but such elements are rare and will have only a negligible effect on the overall proportion of relevant examples with rendaku.
Finally, we will compare a much larger database of modern Japanese compounds (Irwin
and Miyashita 2012) and investigate the extent to which we find a trace of the strong version
of Lymans Law as a statistical tendency.
References
Frellesvig, Bjarke. 2010. A history of the Japanese language. Cambridge: Cambridge University Press.
Irwin, Mark, and Mizuki Miyashita. 2012. Rendaku database. http://www-h.yamagata-u.ac.jp/~irwin/site/
Rendaku_Database.html.
Jdai-go Jiten Hensh Iin-kai, ed. 1967. Jidai-betsu kokugo dai-jiten: Jdai-hen. Tky: Sanseid.
Miller, Roy Andrew. 1984. The first statement of Lymans Law. Bochumer Jahrbuch zur Ostasienforschung
7: 3756.
Unger, J. Marshall. 1975. Studies in early Japanese morphophonemics. Yale University Ph.D. dissertation.
Vance, Timothy J. 2005. Sequential voicing and Lymans Law in Old Japanese. Polymorphous linguistics: Jim
McCawleys legacy, ed. by Salikoko S. Mufwene, Elaine J. Francis, and Rebecca S. Wheeler, 27 43. Cambridge: MIT Press.

Language contact effects on syntax and pragmatics: a case study from early English
Ans van Kemenade, Radboud University Nijmegen, CLS/Dept. of English.
A.v.Kemenade@let.ru.nl
This paper will trace the effects of language contact on the syntax and pragmatics interface,
with a case study from early English. It is well-known that English in its extant history was
subject to various historical layers of language contact, including contact with Old Norse in
the Northeast in the 9th and 10th centuries, and contact with Norman French following the
Norman Conquest in 1066. Thirdly I surmise a language contact situation which is perhaps
less well-known: 14th and 15th century London, an important source of texts, was a melting
pot of many dialects, including those of Northern speakers with their linguistic legacy of
contact with Old Norse. Language contact situations, especially in societies where written
communication is not general and the language not (yet) standardized, typically lead to
untutored second language acquisition (SLA), including late acquisition, which in turn is
thought to affect primarily the interfaces of grammar, according to e.g. Sorace's (2005)
Interface Hypothesis. I will focus here on the interface between syntax and pragmatics, and
present a quantitative case study on the development of Verb Second syntax in the transition
from Middle English to early Modern English. The choice of case study is inspired by two
considerations: 1) it has been established that the specific V2 patterns of Old English and
early Middle English show an interplay of syntactic and pragmatic factors: in V2 syntax, the
first constituent is most typically filled by a constituent that forms a link to the preceding
discourse (Los (2009)), and V2 variation (variation in subject-finite verb inversion) is codetermined by the information status of the subject (given vs non-given), and the type of finite
verb (van Kemenade and Westergaard 2012); 2) given this interplay of syntax and pragmatics,
the hypothesis is that the development of V2 in Middle English and its loss in the transition to
early Modern English is plausibly due to the effects of multilingualism resulting from
increasing dialect contact, affecting the discourse-linking character of the first constituent, and
neutralizing the distinction in information status of subjects, ultimately resulting in rigid SVO
word order.
The case study presented provides a systematic comparison of V2 patterns in texts
from London, from the (more remote) North and Southwest in the fourteenth fifteenth and
sixteenth centuries, focusing on the syntactic and pragmatic properties of first constituent,
finite verb and subject, in order to explore the relative syntactic and pragmatic developments
in these three varieties.
Kemenade, Ans van, and Marit Westergaard. 2012. Syntax and Information Structure: V2
Variation in Middle English. In Information Structure in the History of English, ed. MaraJos Lpez-Couso, Bettelou Los, and Anneli Meurman-Solin. Oxford: Oxford University
Press
Los, Bettelou 2009 The consequences of the loss of verb-second in English: Information
structure and syntax in interaction. English Language and Linguistics 13: 97-125.
Sorace, Antonella (2005)
Selective optionality in language development. In L.Cornips &
K.Corrigan (eds.) Syntax and Variation: Reconciling the Biological and the Social, pp. 55-80.
Amsterdam: John Benjamins.

The diachrony of complex predicates: causatives in Romance

NIGEL VINCENT
The University of Manchester
nigel.vincent@manchester.ac.uk
It is common within both descriptively and theoretically oriented work to make a distinction between
periphrases, in which a non-finite form is accompanied by an auxiliary, and complex predicates, in which the
non-finite form goes together with a so-called light verb. The former include many of the staples of the
grammaticalization literature (e.g. Romance futures or English be going to), and have therefore received extensive
diachronic scrutiny. Complex predicates by contrast have not been subject to much close historical analysis, but
when they have (e.g. Butt & Lahiri 2002 for Indo-Aryan), the items in question, unlike emergent auxiliaries,
appear to be remarkably stable over long timespans. According to Butt & Lahiri (2002), this circumstance is not
coincidental; in their words the stability of the light verb is predicted by its very status. On their account, a light
verb is intrinsically linked with a main verb partner with which it shares its morphological structure. The light
verb potential inherent in main verbs with suitably general meanings (give, take, go, put, etc) is either
activated, in which case the language displays the relevant complex predicate, or dormant, in which case it does
not. Rather than a historically defined relation of bleaching between a main verb and a light verb, there is a
synchronic relation of lexical underspecification. The diachronic options are then only three: switching on the
light verb option at a given historical moment; the switching off of an existing light verb, or the loss of the main
verb, which automatically entails the loss of its light concomitant. Crucially, on this account, light verbs do not
evolve into auxiliaries. In her wider survey, Bowern (2008) supports Butt & Lahiris view that light verbs are
diachronically stable and do not fit into the clines of grammaticalization that lead to the formation of auxiliary
verbs, although she disputes their conclusion that light verb constructions constitute diachronic dead-ends.
In the present paper, by contrast, we focus on the initial phase when a light verb comes into existence.
Our case study involves the Romance causative construction built out of the do verb (Fr. faire, Ital. fare, Span.
hacer, Cat. fer, Port. fazer < Lat. facere) plus the infinitive. This construction obeys all of Butts (2010) tests: the
light verb and the accompanying lexical verb form a monoclausal construction; both verbs contribute jointly to
that clauses argument structure; the construction has remained constant over close to a thousand years of
attested history; in each language the morphology of faire, etc is the same regardless of whether it serves as light
or full verb. The difference from Butt & Lahiris case is that whereas they seek to reconstruct the complex
predicate pattern as part of the grammar of the earliest attested stages of Indo-Aryan, it is clear from the Latin
evidence that the widespread use of reflexes of facere do plus the infinitive to express causativity is a Romance
innovation and not reconstructible to the ancestor language. While a verb meaning do is a plausible member of
Butt & Lahiris claimed set of universal predicates, it is not clear that this is sufficient to explain the genesis of
the new construction. Rather, in line with Bowern (2008) we need to invoke a mechanism of reanalysis from the
bi-clausal accusative and infinitive construction to the monoclausal complex predicate. This reanalysis plus
semantic bleaching due to extension of contexts of use is sufficient to explain the new pattern without invoking
a universal predicate list with its attendant ontological and epistemological problems (how many items figure on
the list? how do speakers come to know them?). Interestingly, the pattern of change we see in the genesis of the
facere construction is repeated in the history of Portuguese with another verb mandar (< Lat mandare order,
entrust). The development of a parallel structure from a different and semantically less general verb confirms
our conclusion that the mechanism at work proceeds via semantic loss or bleaching rather than the activation of
an inherent universal predicate. Crucially such semantic loss does not affect the argument structure potential of
the light verb and it is the latters role in contributing to the argument structure of the clause which ensures its
diachronic stability, although this may change witness for instance the passivizability of the causative in Italian
but not in French or even be lost (Bowern 2008: 175-7).
References
Bowern, Claire. 2008. The diachrony of complex predicates. Diachronica 25:161185.
Butt, Miriam. 2010. The light verb jungle: still hacking away. In Amberber, Menghistu, et al (eds). 2010. Complex
predicates: Cross-linguistic perspectives on event structure. CUP, 48-78.
Butt, Miriam & Aditi Lahiri. 2002. Historical stability vs. historical change. Unpublished Ms. http://ling.unikonstanz.de/pages/home/butt/.

Better characters for better phylogenies: Slavic test cases

Ruprecht von Waldenfels (Bern), Johanna Nichols (Berkeley)
Computational phylogenies have mostly been done using wordlist characters, and for the
most part cognacy codings are based on root comparisons rather than specific word comparisons.
This is true, surprisingly, even for well-studied language families like Indo-European daughter
branches. We test the several other kinds of characters on the well-studied and well-documented
Slavic family, an excellent test case in view of its mix of inherited and contact situations and its wellstudied differences between the modern family tree with its three branches (East, West, South) and
the reconstructable history of dispersal and spread. A good tree needs to show (a) the three
branches and their accepted subbranching; (b) evidence of the secondary consolidation of South
Slavic from at least two early sources; (c) evidence of transitional status for Slovene, which had West
Slavic affinities until the German expansion and Hungarian entry, after which it shares the South
Slavic history; (d) earlier separation of West Slavic from the rest.
Using SplitsTree we show that (1) lexicostatistics using root cognacy for 100-word and 200word Swadesh lists yields mediocre to poor trees; (2) word cognacy yields noticeably better trees,
even in small datasets; (3) adding grammatical information (such as inflectional and derivational
classes) to lexical information considerably improves trees; (4) combined lexical and grammatical data
extracted from parallel corpora yields excellent trees.
For (3), and one of the small datasets in (2), we use a standard list of 18 pairs of verbs used to
typologize the causative alternation. Root cognacy for these 18 pairs (=36 items) alone yields much
better trees than the larger Swadesh list, partly because the items are verbs (the Swadesh list is
mostly nouns) and partly because they span a larger range of stability levels. Adding information
about derivational types within the pairs (e.g. the non-causative is reflexive, the causative underived)
produces an excellent tree, even if only 9 pairs are used.
For (4), we use texts taken from a corpus of parallel translated texts that are automatically
lemmatized, POS-tagged and aligned to each other on sentence and word level. We obtain the
characters for input into SplitsTree by extracting all verb forms across the aligned languages and
classifying them in terms of their verbal prefixes (if any). Classification is at this point simply based on
pattern matching against a set of regular expressions that define 11 prefix classes. In this way, we
arrive at characters that combine grammatical (word class), lexical (prefix class), and semantic
(similarity in meaning as measured by alignment in the corpus) information; see Figure (1) for the
resulting network. This method yields the best tree of all, showing the branches and subbranches, the
transitional position of Slovene, and the diffuse nature of South Slavic.
Figure (1): Network based on prefix classes of ca. 10.000 verb forms in 13 Slavic versions of Stanisaw
Lem's Polish classic Solaris (incl. multiple translations into the same languages).

Thus surprisingly small character sets, if appropriately chosen and grammatically analyzed,
can yield very good trees with rather little effort at extraction and coding; and corpus data involving
both lexicon and grammar can yield excellent trees also with manageable levels of effort now that
the annotation and extraction are largely automated.

The pluperfect in 17th-century translations from Polish into Russian

Christine Watson, Uppsala University

Modern Russian and Polish do not have a pluperfect tense. However, it existed in
earlier stages of both these languages, as well as in Church Slavonic, which functioned as the
bookish written language in Russia until the 17th century. In Russian language history, the
pluperfect is often divided into two varieties: the Church Slavonic pluperfect, with the
auxiliary verb byti to be in the aorist or the imperfect, and the Russian pluperfect, with the
auxiliary verb in the perfect tense.
The pluperfect is traditionally said to refer to an event preceding another event,
expressed in a past tense form, in Church Slavonic and Old Russian typically the aorist. It
could also be used to signify absolute remoteness in time. In early non-bookish texts, the
Russian pluperfect could signify an interrupted, cancelled or reversed event. Remnants of this
meaning of the tense still exist in Russian, in constructions with the particle bylo, such as the
following:
On podnjal bylo stakan, no razdumal
He was about to pick up the glass, but changed his mind
Originally, the perfect-tense auxiliary of the Russian pluperfect was conjugated, but
occasionally in 16th-century texts and even more frequently in the 17th century, we also find
occurrences of a pluperfect with the auxiliary in the neuter singular, but with the main verb,
i.e. the participle form, agreeing with the subject. This is the origin of the modern construction
with bylo, which is a neuter singular form.
The 17th century, when agreeing and non-agreeing forms still competed, was also a
period of strong Polish influence on the Russian language. The Polish pluperfect form was
common in the 16th and 17th centuries, always with the auxiliary verb in agreement.
In this paper, I will study the use of the pluperfect in a number of excerpts from
translations from Polish into Russian, made in the late 17th century. It will be seen that the
Polish pluperfect was not always translated with a Russian or Church Slavonic pluperfect,
which implies that this tense was less accepted in Russian than in Polish. Syntactic and
semantic factors that may have influenced the choice of tense in the translations will be
discussed.
The results will be compared with earlier studies, based on original Russian texts. By
studying translations made in a time period when a presumably language-internal change was
underway in Russian, an additional dimension will be added to the diachrony of the pluperfect
in Russian written language.

References:
Burzywoda, Urszula, Danuta Ostaszewska, Artur Rejter & Mirosawa Siuciak, Polszczyzna XVII wieku. Stan I
przeobraenia. Katowice 2002.
Goeringer, Keith, The motivation of pluperfect auxiliary tense in the Primary Chronicle, Russian Linguistics
19, 1995, 319332.
Gorkova, K. V. & G. A. Chaburgaev, Istorieskaja grammatika russkogo jazyka. 2nd ed. Moscow 1997.
ivov, V. M., Usus scribendi. Prostye preterity u letopisca-samouki. Russian Linguistics 19, 1995, 4575.
Petruchin, P. V., Lingvistieskaja geterogennost i upotreblenie proedich vremn v drevnerusskom letopisanii.
Unpublished dissertation. Moscow 2003.
Uspenskij, B. A., Istorija russkogo literaturnogo jazyka (XIXVII .). 3rd ed. Moscow 2002.

Discontinuous constituents in Classical and Later Greek

Eirik Welo, University of Oslo
Discontinuous constituents (in Greek, hyperbaton) are a typical feature of Classical Greek. Modifiers
of various categories (adjectives, genitives) appear separated from their heads. Recent studies, in
particular Devine and Stephens (2000), have shown that discontinuity is sensitive to the discourse
function of the parts of a complex phrase (topic, focus). In Classical Greek, several kinds of
hyperbaton are found: as in continuous phrases the modifier may precede the head (premodifier
hyperbaton, shown in 1a) or follow the head (postmodifier hyperbaton, shown in 1b).
(1) a. tina
ekhei
dunamin? = 'Which power does it have?'
which-ACC.SG has-3.SG
power-ACC.SG (Plato, Republic 358b)
b. ep' andras
strateuometha agathous = 'We fight against good men.'
againstmen-ACC.PL
wage_war-1.PL
good-ACC.PL (Herodotus 7.53)
In later Greek texts, the frequency of discontinuous phrases varies. Some texts are written in close
imitation of Classical models and often contain the highest frequency of hyperbaton (Welo 2008). The
frequency for the different types of hyperbaton is, however, not the same. Examples of hyperbaton
where the modifier precedes the head are much more frequent, at least with adjectival modifiers.
Lindhamer (1908) collected data from many later Greek texts, illustrating different types of
hyperbaton. She did not, however, distinguish between premodifier and postmodifier hyperbaton.
Based on Lindhamer's data, Welo (2008:29-34) contains a preliminary discussion of the historical
development of hyperbaton within Greek, concluding that there is a sharp rise in the frequency of
premodifier hyperbaton.
The hypothesis of my paper is, first, that the frequency of pre- and postmodifier hyperbaton is
related to the word order patterns of modifier and head in continuous phrases. Secondly, I argue that
even in texts which strive to imitate classical models as closely as possible on the levels of lexicon,
morphology and (some aspects of) syntax, the order of modifier and head is determined more by
diachronic developments than by the word order patterns of Classical Greek texts (see Gianollo 2011
for analysis of the word order of genitive modifiers in postclassical Greek). Consequently, the word
order patterns of discontinuous phrases may provide supportive evidence for the diachronic
development of word order in Greek.
The hypothesis will be tested by comparing the frequency of different types of hyperbaton to
the frequencies of head-modifier and modifier-head order in complex phrases in authors who imitate
classical models in order to look for correlations between the two groups. The different word order
patterns of the continuous phrases in these authors will then be compared to the word order patterns in
texts which do not imitate classical models strongly, e.g. the Greek New Testament, in order to see to
which extent their word order is comparable.

References
Devine, A. M. and Stephens (2000). Discontinuous Syntax. Hyperbaton in Greek. New York- Oxford:
Oxford University Press.
Gianollo, Chiara (2011). "Native syntax and translation effects: adnominal arguments in the Greek and
Latin New Testament", in Welo, E. (ed.), Indo-European syntax and pragmatics: contrastive
approaches (= Oslo Studies in Language 3(3)), Oslo: University of Oslo, 75-101.
Lindhamer, L. (1908). Zur Wortstellung im Griechischen. Eine Untersuchung ber die Spaltung
syntaktisch eng zusammengehriger Glieder durch das Verbum. Ph. D. thesis, Ludwig
Maximilian-Universitt Mnchen.
Welo, Eirik (2008). The information structure of hyperbaton--a case study of Galen's commentary on
Prognosticum. Ph. D. thesis, University of Oslo.

Late Middle English word order: The effect of

information structure and sociolinguistic factors
Marit Westergaard & Tams Eitler
University of Troms & Etvs Lornd University
During the Old English period (OE), the variation in verb second (V2) word
order was generally stable, with V2 in non-subject-initial declaratives mainly
appearing with NP subjects and non-V2 with pronouns. This has been analyzed
as a so-called IP-V2 grammar (e.g. Kroch & Taylor 2007) or one in which word
order is the result of information structure (IS) factors (e.g. Bech 2001,
Westergaard 2009). It is well known that the V2 variation in the Middle English
(ME) period was more extensive: While in early ME, the relative word order of
subject and verb was very similar to what it was in OE, the word order at the
end of the ME period was relatively close to Present-day English in this respect
(i.e. a Non-V2 grammar).
The late ME period also saw a certain rise of V2 in some linguistic contexts,
e.g. with auxiliaries and unaccusative verbs (Warner 2007, van Kemenade &
Westergaard 2012). Furthermore, northern dialects as well as East Anglian
displayed a grammar where V2 was a syntactic requirement (referred to as
CP-V2), due to influence from Scandinavian languages in these areas. Thus,
there was considerable synchronic variation at this time, with different V2
grammars existing side by side. This word order variation has been analyzed
as the result of dialect differences and sociolinguistic factors, e.g. Eitler
(2006).
The present paper discusses V2 variation in late ME by providing a close
investigation of four texts written around 1450. The texts are all produced by
the same author, John Capgrave, and show considerable variation in V2. Our
investigation takes into account both IS factors as well dialectal
differences/social factors such as the intended readership of the written texts,
i.e. a local, regional or national audience. We show that both types of factors
are important to understand this intra-speaker variation at this crucial time in
the history of the English language.
References
Eitler, Tams. 2006. Some sociolectal, dialectal and communicative
aspects of word order variation in late Middle English. Doctoral
dissertation, Etvs Lornd University.
Kemenade, Ans van & Marit Westergaard. 2012. Syntax and information
structure: Verb-second variation in Middle English. In Anneli Meurman
Solin, Mara
Jos
Lpez-Couso
&
Bettelou
Los
(eds.)
Information Structure and Syntactic Change in the History of English
(Oxford Studies in the History of English 2), 87-118. New York:
Oxford University Press.
Kroch, Anthony & Ann Taylor. 1997. Verb Movement in Old and Middle
English: Dialect variation and language contact. Parameters of
Morphosyntactic Change ed. by Ans van Kemenade & Nigel Vincent,
297325. Cambridge: Cambridge University Press.
Warner, Anthony. 2007. Parameters of variation between verb-subject and
subject-verb order in late Middle English. English Language and
Linguistics 11.1, 81112.
Westergaard, Marit. 2009. The Development of Word Order in Old and
Middle English: The Role of Information Structure and First Language
Acquisition. Diachronica 26.1, 65-102.

The diachronic development of std & tonal accent in North Germanic

Allison Wetterlin & Aditi Lahiri
University of Oxford
Tonogenesis in Scandinavian has been a topic of much scholarly debate. There is a great deal
of controversy around the cause and time line of the development of the tonal distinction.
There are three competing views on tonogenesis. The traditionally prevailing hypothesis in
Scandinavia is that North Germanic tonal accent preceded std and that one is a further
development of the other (cf. Kock, 1901, Oftedal, 1952, Riad, 2000). The opposite route is
proposed by Liberman (1982), who claimed that std actually preceded tonal accent in the
history of these languages. In a more recent paper, Rischel (2001) presents a third possibility,
proposing that std and tonal accents are two separate developments that did not follow from
each other. Despite the controversies, there is agreement on two fronts; the accentual
development comes from an HL contour, which turned into Accent 2, while the L pitch accent
manifested itself into Accent 1 and std. This logic suggests that the accent and std patterns
correspond across the languages. Rarely, however, in these discussions are the synchronic
accentual and std patterns compared directly with one another. A careful examination
suggests that there are systematic differences across the accent/std manifestations as we see
below; Danish ven does not correspond with the presence of Accent 1 in Swedish and
Norwegian, and Swedish plural noun humrar fails to match the others.
SG
PL
DEF

Norwegian
venn1, lem1, hummer1
venner2, lemmer2 hummere1
vennen1 lemmet1 hummeren1

Swedish
vn1, lem1, hummer1
vnner2, lemmar2, humrar2
vnnen1, lemmen1, hummeren1

Danish
ven, lem, hummer
venner, lemmer, hummere
venn/en, lemmet, hummeren

We will argue that the complex synchronic correspondences are a reflection of a diachronic
scenario, which is best understood if we assume that following Rischel, the prosodic
constriction of std was a result of vowel lengthening. This became lexicalized due to
encliticisation of the definite article leading to recursive prosodic word formation.
Consequently, the definite clitic had a considerable influence on the lexicalisation of std,
which even now is consistently reflected in the modern system. We will also show that it is
not the bitonal, contour which led to the accent innovation, but rather the development of an L
tone Accent 1 via the glottal constriction of the std. This pattern is consistent across
languages, where despite newer innovations the definite singular overrides "normal"
accentual patterns. For instance, where a word becomes disyllabic due to suffixation, it would
normally get Accent 2, but this is blocked if the additional syllable has clitic status.
Furthermore, the vowel length and std correspondences consistently show up in Danish;
even now std can survive only if the syllable is bimoraic, but the reverse is not true.
Bimoraicity does not automatically lead to std. Our hypothesis also fits with Oeftedal's
Hypothesis A, which concludes that the cliticisation was the most important step towards
lexical accent opposition.
Kock, A. (1901). Alt- und neuschwedische Accentuierung. Strassburg, Karl J. Trbner.
Liberman, A. (1982). The Scandinavian Languages. Minneapolis, U of Minn Press.
Oftedal, M. (1952). On the Origin of the Scandinavian Tone Distinction. Norsk tidsskrift for
spogvidenskap 16: 201-225.
Rischel, J. (2001). "Om stdets opkomst." Pluridicta 38: 16-25.
Riad, T. (2000). The origin of Danish std. In: Aditi Lahiri (ed.) Analogy, Levelling and
Markedness. Principles of change in phonology and morphology. 261300.

The grammaticalization of a new negative modal and the spread of negative concord in Welsh
David Willis, University of Cambridge (dwew2@cam.ac.uk)
Some northern varieties of Welsh use cau, historically a reduced form of a verb found in
literary Welsh as nacu refuse, as a negative modal meaning roughly wont:
(1)

Mae
r drws (yn)
cau agor.
be.PRES.3SG the door (PROG) CAU open.INF
The door wont open.

In the nineteenth century, while the phonological reduction of nacu to cau is well attested, it
is not used in this negative-modal sense. This paper traces the development of this new modal
over the last century using written sources and data from fieldwork for the ongoing Syntactic
Atlas of Welsh Dialects, to argue that the following stages need to be recognized:
(i) cau ceases to impose the requirement that its subject is volitional/agentive, and therefore
goes from being a control verb to being a raising verb (loss of independent argument
structure), a change that is well attested in other cases of the grammaticalization of modals
(cf. English will) (Bybee, Perkins & Pagliuca 1994);
(ii) cau enters the negative system proper and begins to trigger negative concord; thus, in (1),
the affirmative form of be, mae, is replaced by the negative-concord form of the same verb,
namely dydy.
These stages are reflected in ongoing dialect variation, as different varieties have experienced
grammaticalization of this item to differing extents. Minor dialect patterns suggest that further
stages are emerging today:
(iii) cau becomes a concording element itself and may be the target of negative concord with
negative indefinites:
(2)

Does
neb
cau dod
allan.
NEG.be.EXIST.3SG no.one CAU come.INF out
No one will come out.

(iv) cau triggers strong negative concord i.e. both negative concord on the verb (as with
dydy above) and presence of the negative marker ddim not.
The process in (i) is common in auxiliation (cf. Kuteva 2001), although the clear historical
precedence of phonological reduction of semantic and syntactic change marks out the current
example as significant. The co-option of cau into the negative system proper is at first sight
unexpected, but makes sense if the wider verbal system of the language is considered. I argue
that cau instantiates negation and mood, thereby amounting to a composite head. Welsh has
other such heads, as with pairs such as gallu be able vs. methu be unable (negation +
mood) or wedi perfect marker vs. heb negative perfect marker (negation + aspect). The
existence of complex heads of this type is likely to have favoured the same treatment for cau,
suggesting that the tendency for acquirers to make cross-category generalizations of this kind
is a factor in language change.
References
Bybee, Joan L., Revere D. Perkins & William Pagliuca. 1994. The evolution of grammar:
Tense, aspect, and modality in the languages of the world. Chicago: Univ. of Chicago Press.
Kuteva, Tania. 2001. Auxiliation: An enquiry into the nature of grammaticalization. Oxford:
Oxford University Press.

Medieval Sardinian: New Evidence for Syntactic Change from Latin to Romance
Sam Wolfe, University of Cambridge
Modern Sardinian is a topic of considerable interest in the syntactic literature (cf. Jones 1993; Remberger
2006, 2010; Cruschina 2012), yet little diachronic work on Sardinian has been undertaken. The aim of
this paper is to present findings of a 50-page syntactic corpus study of medieval Sardinian word order and
set these findings in their diachronic context. Within the Old Sardinian corpus, V1 orders represent 71%
of the matrix sample (1). In addition, V1 orders constitute the only attested order in embedded clauses (2):
(1) Postince Bicturu Plana sa parone sua dessa terra de Collectariu
donated Bicturu Plana the portion his of-the land of Collectariu
Bicturu Plana donated his portion of the land at Collectariu
(2) Custos fiios de Gavini Formiga, ki posit iuige Gostantine
these children of Gavini Formiga that donated iudice Gostantine
These children of Gavini Formiga, that Iudice Gosantine donated
Old Sardinian also shows alternations between S-V orders (3) and V-S orders (4), which have been noted
by previous authors (Virdis 1996, Lombardi 2007) but have never been systematically analysed:
(3) Isse levedi ad Margarida
he took to Margarida
He took Margarida
(4) Postince Furatu Icalis paronce sua de binia....
donated Furatu Icalis portion his of vineyard
Furatu Icalis donated part of the vineyard...


The preverbal placement of Higher adverbs (cf. Cinque 1999) and generalised enclisis in matrix
clauses, along with a lack of Complementiser-XP-V orders in embedded contexts suggest Old Sardinian
had generalised V-to-CFin movement. The S-V/V-S alternations depend on the Information Structure of
the Subject in question, with S-V generally occurring when an old information Subject raises to a Topic
position in the left periphery and V-S when a new information Subject remains in its in situ position in
Spec-vP.


The predominance of V1 orders in the Old Sardinian corpus appears distinct from the V2 patterns
reported elsewhere for Old Romance (cf. Beninc 1983-4, Vance 1997, Ledgeway 2012). Proposing that
these V-initial orders are derived by V-to-CFin movement, however, has the desirable consequence of
showing that the grammars of these Old Romance varieties are underlying similar, as V movement into
the C-domain is commonly assumed in deriving V2 effects in other Old Romance varieties (Adams 1987,
Beninc 2004, Ledgeway 2007). I will suggest that V-to-CFin movement is the commonality of the
abstract Medieval Romance syntax described by Beninc (2004:245), with the makeup of the other
projections in the clausal left periphery the point of variation between varieties.


V1 structures are evident in the textual record from the earliest stages of Latin (Kroll 1918, Adams
1976, Wanner 1987), yet occur in progressively more pragmatic-driven and structural contexts in later
periods (Bauer 1995). It will be proposed that Old Sardinian provides crucial evidence for the period of
generalised V-fronting outlined by Salvi (2004) and Ledgeway (2012), which followed a reanalysis of the
late Latin Primary Linguistic Data by child language acquirers as inconsistent with an SOV grammar.
Generalised V1 in Old Sardinian is therefore a result of this reanalysis. In other Old Romance varieties a
further reanalysis took place, consistent with an innovative V2 grammar, where the preverbal position is
near-obligatorily filled.
BENINC, P. (2004). The left periphery of Medieval Romance. Studi linguistici e filologici online, 2(2), 243297
LEDGEWAY, A. (2012). From Latin to Romance: morphosyntactic typology and change. Oxford: Oxford University
Press SALVI, G. (2004). La formazione della struttura di frase romanza: ordine delle parole e clitici dal latino alle
lingue romanze antiche. Tbingen: Max Niemeyer Verlag.

Person marking in the history of Japanese: The role of linguistic signs

Toshiko Yamaguchi
University of Malaya
In a recent study on person marking, Heine and Song (2011) investigate personal pronouns from the
perspective of grammaticalization theory. The main argument of their study is that change is relatively
regular cross-linguistically and is subsumed within three aspects. First, personal pronouns belong to
functional categories owing to their schematic meaning. Second, their change to a desemanticized
category is unidirectional: the movement from noun to pronoun represents linear desemanticization.
Third, their evolution involves only a handful of conceptual sources. However, Japanese diachronic
data suggest, on the contrary, that: (i) Japanese expressions of person are lexical, open-class items; (ii)
unidirectionality is not essential to change; and (iii) there are factors other than conceptual sources.
One reason for these discrepancies is that the evolution of Japanese person markers was invariably
influenced by the rules of social verbal interactions, or taig (literally treatment). As such, they
served to reveal interpersonal relationships between speaker and addressee in the form of respect, selfdepreciation, or contempt, all of which were integral parts of a hierarchical Japanese society.
This paper provides a new analysis of person markers that is independent of chaining, directionality,
or categoriality of conceptual notions. As early as Kobayashi (1927) and Sakuma (1966), it has been
known that Japanese has several person markers derived from nouns or demonstratives. This paper
investigates a selected number of first- and second-person markers. Three significant facts should be
noted. First, some markers (koko here, I; konata this direction, I, you; temae I, you) indicate both
the first person and the second person. Second, one marker (anata over there) came to indicate the
honorable second person based on its reference to spatial remoteness. Third, some markers (konata)
ceased to be active markers. In order to explain these historical facts, I rely on the theory of the
linguistic sign.
Following Keller (1998 [1994]), the linguistic sign can be characterized in three ways, as
symptom, icon, and symbol. Symptoms are the most archaic signs whose interpretation is that of
causal inference, while icons are ad hoc and highly contextual signs. Symbols are more advanced,
rule-based signs. The fact that some proximal demonstratives are conflated with the first person can
be explained by their symptomatic nature. The fact that a number of person markers were short-lived
or replaced by other markers can be explained by their iconicity. Furthermore, the fact that some
markers survived until today can be ascribed to their establishment as symbols. Icons are the most
intriguing type of sign as they played a role in substituting the first-person marker for the secondperson marker, thereby generating socio-pragmatic meanings (e.g. respect or contempt) that were
determined according to the speech situation.
This paper will show how the evolution of person markers in Japanese can be accounted for by
the structure of the linguistic sign. This approach does not fit perfectly in the pathways mapped by
Heine and Song. One obvious reason for this is that the creation of person markers in Japanese is the
result of a gradual process of semantic innovation.
Keywords: Japanese, linguistic signs, person marking,
Keller, Rudi. 1998 [1994]. A Theory of Linguistic Signs. Oxford: Oxford University Press.
Sakuma, Kanae. 1966. Gendai Nihongo no Hygen to Goh [Expressions and Usage of Modern
Japanese]. Koseisha.
Kobayashi, Yoshiharu. 1927. Nihongo Kokubunp Ygi [The Essentials of Japanese Grammar].
Tokyo: Kyobunsha.
Heine, Bernd and Kyung-An Song. 2011. On the grammaticalization of personal pronouns. Journal of
Linguistics 47: 587-630.

Rhetorical Questions in Middle English Religious Prose: Their Illocutionary and

Perlocutionary Acts, Textual Functions, and Related Genres
Fumiko Yoshikawa
Hiroshima Shudo University
Abstract
Papers in Halmari and Virtanen (2005) have suggested several rhetorical strategies
relating to different genres such as political speech and judicial argumentation. These
rhetorical strategies are primarily used to persuade readers and listeners of the
reasonableness of their authors opinions. Halmaris contribution in the volume
examines several rhetorical strategies seen in State of the Union addresses by two U.S.
Presidents: Ronald Reagan and Bill Clinton. One of the strategies she treated in the
paper is using rhetorical questions. She showed that both presidents used rhetorical
questions to create an effect by engaging listeners and making them think (Halmari
2005: 116) and also to introduce a new topic (Halmari 2005: 117). These effects,
persuading the listener or the reader and using rhetorical questions as a text
segmentation, are not new but were even used in the Archaic Period, and the same
strategy is frequently utilized in the Middle English period.
In some Middle English religious texts such as Julian of Norwichs Revelations of
Divine Love (hereafter Revelations), we find only a few rhetorical questions. In others
such as Ancrene Wisse, The Cloud of Unknowing, and Nicholas Loves The Mirror of
the Blessed Life of Jesus Christ (hereafter The Mirror), rhetorical questions are
frequently used. In Ancrene Wisse and The Mirror, they seem to be used when
introducing a new topic. In other words, rhetorical questions are used as a textual
segmentation.
In this paper, several works of religious prose are compared in their uses of
rhetorical questions, and some correlations with the use of rhetorical questions are
examined, for example, relationships between the use of rhetorical questions and those
of other persuasive rhetorical strategies, between the use of rhetorical questions and the
reader-writer relationship, and between the use of rhetorical questions and the purpose
of writing the text. A tradition of writing style with rhetorical questions and genres
related to the their usage are discussed in the concluding sections.
Works Cited
Halmari, Helena (2005). In search of successful political persuasion: A Comparison
of the Styles of Bill Clinton and Ronald Reagan. In H. Halmari & T. Virtanen
(2005) pp. 105-34.
Halmari, Helena & Virtanen, Tuija (2005). Persuasion Across Genres: A Linguistic
Approach. Amsterdam /Philadelphia. John Benjamins.

Syntax and semantics of split intransitivity in Japanese:

A comparative study of Old Japanese and Modern Japanese
Zixi You (University of Oxford)
zixi.you@orinst.ox.ac.uk
Split intransitivity refers to a fundamental dichotomy within intransitive verbs.
Since the formulation of Perlmutters Unaccusative Hypothesis (1978) that divides
intransitive verbs into unaccusatives (whose subjects originate as direct objects) and
unergatives (whose subjects originate as subjects), split intransitivity has received a great
deal of attention among linguists. However, most studies on split intransitivity so far have
been focused on modern European languages, with few studies examining Asian languages
and ancient languages. This paper contributes a comprehensive comparative study that
explores split intransitivity in Old Japanese (700-800 AD) and Modern Japanese (1600present).
Large corpora with a big amount of textural and grammatical information merit
descriptive and analytical linguistic research not only on modern languages but also on
ancient languages. While the investigation of Modern Japanese was based on previous
literature (Kishimoto 1998, among others) and native speakers intuitions, the Old Japanese
part of the research was entirely based on the newly completed Oxford Corpus of Old
Japanese (OCOJ). As a part of the collaborative research project Verb semantics and
argument realization in pre-modern Japanese, OCOJ consists of nearly all attested Old
Japanese texts, romanized and xml tagged with a wide range of linguistic information (e.g.
orthography, part-of-speech, morphology and syntactic constituency) following the
conventions of Text Encoding Initiative (TEI).
Semantically, this paper looks at Old Japanese verbs in their contexts in OCOJ to
investigate the interaction between semantic factors that play crucial roles in the classification.
Those are then compared with the semantic parameters that were proposed for Modern
Japanese. Syntactically, through comparison, this paper identifies manifestations that are
exclusive to Modern Japanese (e.g. kake suffixation), those that are exclusive to Old
Japanese (e.g. perfective auxiliary selection), and also discusses manifestations which are
apparently shared by both language stages yet with slight differences (e.g. VV-compounding
and so on). These manifestations further reveal that both Modern Japanese and Old Japanese
have the feature of covert split intransitivity, rather than overt split intransitivity. From
another perspective, data in Old Japanese show evidence of deep unaccusativity, whereas
those in Modern Japanese reveal both deep unaccusativity and surface unaccusativity.
Aiming at introducing new corpus for ancient language and sharing interesting data, the
presentation of this paper is expected to be of the interest of historical linguists, especially to
people who are doing synchronic or diachronic work on ancient languages through corpora.

References
Perlmutter, David. 1978. Impersonal passives and the Unaccusative Hypothesis. In
Proceedings of the Fourth Annual Meeting of the Berkeley Linguistic Society,
157-189. Berkeley: Berkeley Linguistics Society, University of California.
Kishimoto, Hideki. 1998. Semantic Parameters of Unaccusativity. Kobe Papers in Linguistics
1: 15-34.

When history repeats itself: The evolution of the palatal lateral consonant in
Spanish and Portuguese
Andr Zampaulo
The Ohio State University & University of Wisconsin-La Crosse
In the evolution of the sound patterns of contemporary Romance languages, several have
been the pathways taken by their consonant and vowel systems from shared roots in Latin. The
palatal order of consonants, in particular, presents a revealing case of how such languages have
differed in their development. Within this class of segments, the palatal lateral // is of utmost
importance due to the uniqueness of its evolution and its implications for the sound inventories
and patterns of Romance languages, both in the past and in the present. Focusing on Spanish and
Portuguese, this paper takes a novel approach to the evolution of this consonant, by considering
the phonetic motivation of its development and by connecting its synchronic dialectal distribution
with its diachronic pathways of change. In the history of Spanish, more specifically, this
consonant arose from two different groups of Latin sources, which displayed distinct evolutionary
pathways in time. A first // (< Latin /-lj-, -kl-, -gl-/) incurred delateralization and evolved into
the palatal obstruent // in Old Spanish, as illustrated in (1) (e.g. Ariza 1994; Penny 2002):
(1)

MULIERE > mu[lj]ere > *mu[]ere > mu[]er woman (Old

OC(U)LU > o[kl]o > *o[]o > o[]o eye (Old Sp.)
TEG(U)LA > te[gl]a > *te[]a > te[]a tile (Old Sp.)

Sp.)

A more recent // (< Latin /pl-, kl-, fl-, l/) also delateralized but started to merge with the palatal
obstruent // since the 16th century, as shown in (2) (e.g. Lloyd 1987):
(2)

CABALLU > caba[]o (Old Sp.) > caba[]o horse (Modern Sp.)
PLANU > []ano (Old Sp.) > []ano flat (Modern Sp.)
CLAMARE > []amar (Old Sp.) > []amar to call (Modern Sp.)
FLAMMA > []ama > (Old Sp.) > []ama flame (Modern Sp.)

In the history of Portuguese, however, the palatal lateral // has only shared the same sources of
the first Spanish // (< Latin /-lj-, -kl-, -gl-/): MULIERE > mu[]er, OC(U)LU > o[]o, TEG(U)LA >
te[]a (e.g. Williams 1962). The significance of these historical developments comes to light
when one considers the recurrence of very similar processes of change in the palatal lateral of
current dialects in the aforementioned Romance languages. Recent research reveals the advanced
delateralization stage of // in current dialectal Spanish (e.g. Canfield 1981; Lipski 1994; Alvar
1996a, 1996b; Colantoni 2001, 2004) as well as in certain varieties of Portuguese (Aguilera 1989;
Cristfaro Silva 1998; Freire & Marques de Lucena 2011). Several authors have pointed out the
articulatory complexity of this consonant (e.g. Navarro Toms 1967; Ladefoged & Maddieson
1996) and how its production may vary from language to language (e.g. Bladon & Carbonaro
1978; Recasens 1990). In addition, the acoustic features of this segment also reveal its intricacies,
with its first and second formants displaying patterns similar to those of palatal vowels /e/ and /i/
(e.g. Quilis 1981:287). However, while descriptions of the historical steps and the phonetics of //
are present in the literature, very few scholarly works have centered their efforts in unfolding the
crucial part that the phonetic characteristics of this segment (and its surrounding environments)
have taken in its evolution. Thus, this study presents a unified, listener- and constraint-based view
of the role that the articulatory and acoustic patterns of the palatal lateral have played in its
development throughout the history of Spanish and Portuguese and continue to shape its current
dialectal manifestations. More specifically, the current presentation brings forth the determining
contribution of phonetic motivation for the study of recurring pathways of sound change, thus
casting light upon the fuzzy boundary between the synchrony and the diachrony of the Spanish
and Portuguese sound patterns.

OnthehistoryofthedativealternationinEnglish
EvaZehentner
UniversityofVienna
The present paper investigates the history of the dative alternation in English, a syntactic phenomenon which
enablesditransitiveverbstoappearintwodifferentconstructions,cf.thefollowingexamples:
(1)JohngaveMaryabook
(2)JohngaveabooktoMary.
The main issues to be addressed here are the changes in the distribution of these variants from Old English to
PDE, as well as changes in the factors determining the choice of one variant over the other. According to established
opinion, the periphrastic variant of the double object construction arose in late Old English/ early Middle English, most
probably due to the broad erosion of the inflectional system and the subsequent need for more explicit expressions to
avoidambiguity(cf.Allen1995Fischer1992McFadden2002Visser1963).
While in the beginning, the competition appearstohavebeen resolved rather randomly, withvariation beinghigh,
the alternation is believed to be largely functionally motivated in todays English. The choice here has been shown to
depend on various cognitive/ processingrelated factors such as syntactic weight (cf.Hawkins1994 Rohdenburg 1996),
animacy, definiteness, discourseaccessibility and others (cf. Wolk et al. forthc. Bresnan et al.2007 Rappaport Hovav
&Levin2008).
Furthermore, some idiosyncratic preferences can be observed, the motivations behind which remain to be
identified. Sowere,e.g.certain verbclassesassociated tothe doubleobjectconstructioninOldEnglish,butlatercameto
be confined to the prepositional pattern (e.g. verbs of saying cf. Levin 1993 de Cuypere 2010), while others are only
found inthesyntheticvariant. Allthis suggests thatsubstantialchangesinthelanguagesargumentstructure systemtook
place between Old English and PDE, which seem tohaveresultedin substantial functional reorganisationanda more or
lessconsistentresolutionofthecompetition.
This paper now aims to substantiate these claimsonthebasisof anextensive corpus studyof ditransitives in the
Penn Corpora of Historical English (PPCME2/ PPCMEME), seeing that no longterm, largescale diachronic
investigation of the factors conditioning the alternation has been carried out so far (cf.de Cuypere 2010). The data
acquired in the analysis is further supplemented by and compared to results obtained in research on the construction in
Late Modern English and PDE (Levin 1993 Bresnan et al.2007 Wolk et al. forthc.). Moreover, the paper takes into
account work doneonditransitiveverbsin OldEnglish, andtacklesthequestionof whetherand towhatextentthedative
alternation continues the earlier variant word orders of DOIO and IODO (cf. de Cuypere 2010, Koopman 1990
McFadden2002Polo2002).
Concerning theoretical framework, the paper attempts to combine elements of usagebased, functional models
withevolutionarylinguisticapproaches(cf.Nedergaard2006Bybee2007Bybee&Hopper2001andothers).
References
Allen,Cynthia.1995.Casemarkingandreanalysis:grammaticalrelationsfromOldtoEarlyModernEnglish.Oxford:OUP.
Bresnan,J.Cueni,A.Nikitina,T.Baayen,R.2007.Predictingthedativealternation.InBoume,G.etal.(eds.).Cognitive
foundationsofinterpretation.Amsterdam:RoyalNetherlandsAcademyofScience,6994.
Bybee,JoanL.(2007).Frequencyofuseandtheorganizationoflanguage.Oxford:OUP.
Bybee,JoanL.Hopper,Paul.(eds.)2001.Frequencyandtheemergenceoflinguisticstructure.Typologicalstudiesinlanguage45.
Amsterdam/Philadelphia:Benjamins.
Cuypere,Ludovicde.2010.TheOldEnglishdoubleobjectalternation.Sprachwissenschaft35,337368.
Fischer,Olga.1992.Syntax.InBlake,Norman(ed.).TheCambridgehistoryoftheEnglishlanguage.Vol.II,10661476.
Cambridge:CUP,207408.
Hawkins,JohnA.1994.Aperformancetheoryoforderandconstituency.Cambridge:CUP.
Koopman,Willem.1990.ThedoubleobjectconstructioninOldEnglish.InAdamson,SylviaLaw,VivienVincent,NigelWright,
Susan.(eds.).Papersfromthe5thICEHL.Amsterdam/Philadelphia:Benjamins,225244.
Levin,Beth.1993.Englishverbclassesandalternations:apreliminaryinvestigation.Chicago:UniversityofChicagoPress.
McFadden,Thomas.2002.TheriseofthetodativeinMiddleEnglish.InDavidW.Lightfoot(ed.).Syntacticeffectsof
morphologicalchange.Oxford:OxfordUniversityPress,107123.
NedergaardThomsen,Ole(ed.).2006.Competingmodelsoflinguisticchange.Evolutionandbeyond.Amsterdam/Philadelphia:
Benjamins.
Polo,Chiara.2002.Doubleobjectsandmorphologicaltriggersforsyntacticcase.InDavidW.Lightfoot(ed.).Syntacticeffectsof
morphologicalchange.Oxford:OxfordUniversityPress,124142.
RappaportHovav,MalkaLevin,Beth.2008.TheEnglishdativealternation:thecaseforverbsensitivity.JournalofLinguistics44
(1),129167.
Rohdenburg,Gnter.1996.CognitivecomplexityandincreasedgrammaticalexplicitnessinEnglish.CognitiveLinguistics7(2),
149182.
Visser,FredericusTh.1963.AnhistoricalsyntaxoftheEnglishlanguage.Leiden:Brill.
Wolk,ChristophBresnan,JoanRosenbach,AnetteSzmrecsanyi,Benedikt.forthc.DativeandgenitivevariabilityinLateModern
English:exploringcrossconstructionalvariationandchange.

The development of subject positions in Old English

Richard Zimmermann
Universit de Genve,
Richard.Zimmermann@unige.ch
In Old English, subject placement is relatively variable. While the majority of subjects
are placed in the canonical, initial subject position as in Modern English (1), a nontrivial
number of subjects are found elsewhere, post-verbally in clause-medial position (2) or
extraposed, in clause-final position (3) (subjects in bold).
(1)

a
ara treowleasra cyninga beboda wi cristenum monnum grimsedon.
when of.the faithless
kings command against Christian men
raged
When the commands of the faithless kings burst forth against the Christians,
(cobede,Bede_1:7.34.12.274)

(2)

a t a ongeaton a rran gewinnan t se Romanisca here ws onweg gewiten,

when that then understood the former enemies that the Roman army was away gone
When the former adversaries then understood that the Roman army had gone away,
(cobede,Bede_1:9.44.20.379)

(3)

a gehldan hiene a apostole Petrus & Iohannes,

when healed
him the apostles Peter and John
When the apostles Peter and John healed him,
(cobede,Bede_5:2.390.8.3885)

I will present a quantitative corpus study (YCOE, Taylor et al. 2003), similar in spirit
to Kemenade, Milicev & Baayen (2007), which investigates the development of subject
placement during the Old English period. The material will be limited, firstly, to adverbial
subordinate clauses so that cases of high subject placement in the C-domain can be avoided
and, secondly, to full, non-pronominal subjects in order to allow genuine variation between
discourse new and given subjects. Independent variables considered are period, verb position,
the information-structural status of the subject (e.g. Bech 2001, Los & Dreschler 2012), as
well as various morphological and semantic properties of the subject. The resulting model can
answer the question how and at what rate the canonical, pre-verbal subject position
strengthens during the Old English period. Subsequently, I will discuss the findings of the
empirical study with reference to the clausal architecture of Old English and the question of
what role information structural factors play in language change.
References
Bech, K. (2001) Word order patterns in Old and Middle English: A syntactic and
pragmatic study. Doctoral dissertation, University of Bergen.
Los, B. L. J. & Dreschler, G. A. (2012). The loss of local anchoring: From adverbial local anchors to
permissive subjects. In: Nevalainen, T. & Traugott, E.C. (Eds.) The Oxford Handbook of the History
of English. Oxford: Oxford University Press.
Kemenade, A. van, Milicev, T. & Baayen, R.H. (2007). The balance between discourse and syntax in
Old English In : Dossena, M. & Maurizio G. (eds.) Selected papers from the 14th International
Conference on English Historical Linguistics. Amsterdam: John Benjamins.
Taylor, A., Warner,A., Pintzuk,S & Beths, F. (2003). The York-Toronto-Helsinki Parsed Corpus of
Old English Prose. 1.5 million words of syntactically and morphologically annotated text. Available
through the Oxford Text Archive.

Schminke:MechanismsofGenderIntegrationandGenderChangeinGerman
DavidZubin(StateUniversityofNewYorkatBuffalo)
KlausMichaelKpcke(UniversityofMnster)
The classic problem with the study of grammatical gender, the one that distinguishes genderfromother
grammatical systems, is the apparent complexity and/or arbitrariness of gender class assignment. And
the study of this problem ultimately resolves back to the investigation of gender change and gender
integration. The typical approach to gender has been deep diachronic, lacking detail of either the
psychological or the cultural context in which the gender change/integration occurred. Inthispaperwe
will follow the lead of Osthoff and Brugmann in their Neogrammarian Manifesto (preface to
Morphological Investigations) by focusing on the synchronic cognitive context of gender
change/integration in process, as well as the cultural context in which the changes are occurring. To
adequately capture these two dimensions of context we will focus on a specific concept domain
Decorative Cosmetics exhibiting a number of attributes beneficial to the study. Firstly, within the
cultural context of cosmetic use, nouns are currently being integrated into the lexicon from other
languages, notably English and French, resulting in ongoing genderintegrationandnounsalreadystable
in the lexicon are being appropriated to new uses in the context, resulting in some cases in gender
change. Secondly, there is extensive interactive discourse on Decorative Cosmetics providing relevant
quantifiable data: discourse recorded in promotional websites, blogs, and chatrooms. Finally, Many of
the nouns in this domain exhibit extensive gender variation, suggesting that the domain is ahighlyactive
zone of gender integration and gender change inthelexicon,and providingdetailedquantitativedatafor
testinghypothesesconcerningtheunderlyingmechanisms.
Mechanismswewilldiscussinthepaperinclude:
a.genderborrowing:TeintborrowedasMfromFrench
b. idiosyncratic integration conventions: in particular the m>n convention for classes of French loans,
includingBlushandParfum
c.analogytothesituationalequivalent(SE):AftershaveN~GesichtswasserN
d. analogy to a translation equivalent (TE): Brush (e.g. Stippling Brush), in addition toMfromtheSE
Pinsel,showssignificantFvariationfromtheTEBrste.
e.subcategorizationeffect:FangoMMineralklasse
f. assimilation to field default (field effect): many nouns in the cosmetics domain show a slight to
moderate tendency toward N, despite the force of more local mechanisms Eyeshadow and
FoundationshowsomeNvariationdespitethestrongSELidschattenMandthesuffixionF.
g. highfrequency headdrop: CamouflageN<CamouflageMakeUpN,significantvariationwithM
from suffix age Labello M < LabelloStift M Kajal M < Kajalstift M (n.b. sig. variation with
earlierKajalN)
h. suffixbased integration: Modellage is integrated as F based on the convention for borrowing age
from French Lipmarker, Conditioner, Concealer areintegratedasMbasedonanalogybetweenthe
GermanandEnglishderivationalsuffixeser
i. gender change: Butter (Body Butter, Shea Butter) and Milk (Body Milk) show significant N
variation,despitetheFgenderof thesenounsinthecorelexicon.Puder(Kompaktpuder,losesPuder)
significant N variation despite previous integration as M.Thegoalofthestudywillbeto a)providea
theorydriven taxonomy of integration and change mechanisms, and b) analyze the outcome of
competitionamongcompetingmechanisms.

Gurn rhallsdttir (University of Iceland), sta Svavarsdttir (The rni Magnsson Institute for
Icelandic Studies), Eirkur Rgnvaldsson (University of Iceland), Haraldur Bernharsson (University
of Iceland), Jhannes B. Sigtryggsson (The rni Magnsson Institute for Icelandic Studies), Veturlii
skarsson (Uppsala University), Wim Vandenbussche (Vrije Universiteit Brussel)

Is it possible to reverse a linguistic change?

Language change and standardization in 19th-century Icelandic
Fundamental changes took place in Icelandic society in the 19th and 20th centuries, which
affected the language in various ways. The status of Icelandic changed from that of a remote
minority language in the Danish kingdom to a national language in a sovereign state. In
assuming the role of an official language, a common linguistic norm had to be defined and
developed. The research team behind the project Language Change and Linguistic Variation in
19th-Century Icelandic and the Emergence of a National Standard is currently focusing on the
beginnings of the standardization of Icelandic in the 19th century and studying its linguistic,
sociolinguistic and ideological foundations.
The dominant model for the emerging standard was the medieval language and the attitudes
tended towards purism. The questions we want to answer concern the nature and the spread of
internal and external linguistic changes; the distribution of linguistic variants and their
correlation with sociolinguistic and stylistic factors; and the influence of expressed opinions on
linguistic features and of organized intervention on the linguistic development. Earlier research
has revealed the importance of the medieval linguistic model in 19th and 20th century language
ideology, but it has never been investigated empirically if and how contemporary linguistic and
sociolinguistic factors affected the proposed standard.
A central and fundamental question, with general theoretical relevance, is the following: Is it
possible to reverse a linguistic change? A potential case is a change affecting the inflection of a
set of masculine nouns, where the nom. sg. ending -r was reinterpreted as belonging to the
stem. The older pattern was allegedly extinct by the mid-19th century, when a well-known
language reformer suggested a return to the earlier inflection. The older variants were in fact
standardized and are dominant in the modern language. This would thus be an example of the
revival of an extinct linguistic feature by deliberate effort, if it indeed holds true that the earlier
inflection had practically disappeared.
We are analyzing selected linguistic variables, relevant with respect to the standardization, by
applying a variationist approach and methodology. We apply two corpora of written texts from
the 19th century as a basis for our analysis, one of private letters and the other of newspapers
and journals, and indications of an emerging standard are both sought in actual language use
and in contemporary views on language use expressed in 19th-century newspapers and
textbooks. The frequency and distribution of different variants are studied and correlated with
stylistic and sociolinguistic factors in order to reveal their potential influence on the choice of
standard variants. The present paper will give examples of preliminary results by discussing
investigations of certain morphological and morphosyntactic changes where the 19th-century
corpora have proved immensely useful.
Reference
Language Change and Linguistic Variation in 19th-Century Icelandic and the Emergence of a
National Standard. See http://arnastofnun.is/page/LCLV19.