Você está na página 1de 14

ISSN 1392-0588



Lexical Grammar
INTRODUCTION distinction between grammar, on the one
hand, and lexis, on the other hand, is not
This paper concerns the relation between as fundamental as it is usually held to be
the two types of pattern that are mainly and since it is a distinction that is made
recognised as the means whereby lan- at the outset of the formal study of lan-
guage creates meaning. The terms gram- guage, then it colours and distorts the
mar and lexis will mainly be used for whole enterprise. It is worth consider-
these, but instead of grammar you will ing how far, using modern techniques,
sometimes find syntax or structure, and we can get in describing a language with-
instead of lexis you may find semantics out resorting to such a distinction.
or vocabulary. But there is always this
basic distinctions, of a component which
produces patterns of organization and a GRAMMAR AND LEXIS
component which produces items that
fill places in the patterns; the items tend The distinction between grammar and
to be chosen individually, and with little lexis is a very basic model of language;
reference to the surrounding text. there would be no motivation to recon-
The title of the paper is Lexical Gram- sider it unless new evidence gave rise to
mar and not lexico-grammar. Lexicogram- concern about its accuracy. One reason
mar is now very fashionable, but it does for such a model becoming so well es-
not integrate the two types of pattern as tablished could be simply that before the
its name might suggest - it is fundamen- computer age linguists were unable to
tally grammar with a certain amount of describe all the complexity of language
attention to lexical patterns within the at once; since it could be represented as
grammatical frameworks; it is not in any a framework and a set of choices to fit
sense an attempt to build together a gram- the frames, one of those elements could
mar and lexis on an equal basis. be held steady and the other varied against
When a dichotomy is firmly established it. So we could forget, temporarily, about
in a culture, it is difficult to find a name the patterns of semantic choice while we
for it or to talk about it as a unified look at the organization of the structures;
whole and not two different things; that and then the process could be reversed,
is the problem here. Recent research into and when we come to look at the words
the features of language corpora give us and their meanings, then we do not con-
reason to believe that the fundamental sider at this point whether they are sub-
John Sinclair, 2000

jects or objects of clauses or objects of relationship with the categories that they
prepositions, if they are noun phrases, realise (Sinclair 1999a). So grammar is
because that part of the overall organi- superficially easy to observe but much
sation is suspended. more abstract than appears at first sight.
In other words, we can put forward In contrast, the lexical patterns are very
for consideration the suggestion that this difficult to observe, because they are
initial division of language patterning realised by a large vocabulary of infre-
may not be fundamental to the nature of quent words, and so it is not easy to
language, but more a consequence of the work out the recurrent patterns that lie
inadequacy of the means of studying lan- beneath the massive variation. The pat-
guage in the pre-computer age. When terns are patterns of combination, and
the linguist had nothing but his or her this compounds the problem; whereas in
five senses, memory and internal aware- grammar the recurrence of frequent words
ness, it was difficult to analyse such a makes it fairly easy to notice patterns of
complex matter as language; consider combination, in lexis the combinations
phonetics, for example, before the sound had only been seen in a few hundred
wave could be slowed down and divid- idiomatic expressions which were so re-
ed into its components. Without the markable that they had to be accounted
ability to manipulate language external- for separately. With large corpora and
ly, the observer/analyst has to leave some powerful computers we are at the fron-
things steady, or hope they stay steady tiers of a new view of language, where
while other aspects of the whole are ex- we can appreciate its full complexity
amined. And the problem is, in lan- without getting hampered by the detail.
guage, that they dont stay steady. So It is thus no accident that linguists up
we should at least question the wisdom till now have developed grammars much
of dividing the meaningful patterns of more than dictionaries and lexicons; we
language into two at the outset. tend to have very elaborate grammars,
which contain intricate apparatus with
ranks and hierarchies and structures and
ABSTRACTIONS all sorts of categories, with many differ-
ent kinds of organization and in contrast
There is a related point to be made sep- we have very, very simple models of lex-
arately, but also a consequence of the ical structure, which are mainly one-di-
position of the human observer. It is mensional, based on the word. There is
generally recognised that the meaning- an ad hoc set of terms for multi-word
ful patterns of language are of an ab- units like idiom and clich and saying
stract nature, which is one reason why and proverb, but all these are ill-defined
they are so difficult to explain, and to terms, and there is no other network of
use in teaching; from the perspective of interconnections between one word and
grammar they are more abstract than they another.
seem to be at first sight. It is possible Again this disparity in our descriptions
that the reason for their unexpected lev- does not necessarily reflect the nature of
el of abstraction is that grammar typical- language, but rather it reflects our col-
ly is realised through the common words lective inability to process language with
and morphemes - that is, they seem to be sufficient power and understanding to
familiar, but in fact many of them are see that the complexity of the language
multiply ambiguous and in a complex as seen from a lexical point of view is
John Sinclair LEXICAL GRAMMAR 193

just as great as the complexity of the date the meaning of text. The distinc-
language as seen from the grammatical tions could have any labels at all, and at
point of view. So we may expect that best they have a mnemonic function (this
simple artefacts like dictionaries will give argument is well-supported by the re-
way to more complex lexical architec- tention in current grammars of terms like
tures - indeed the development of dic- finite or voice, which bears little
tionaries with an influence from corpus relation to meaning in current English).
research has begun to move in this di- The only meaning that grammar pro-
rection. vides is differentiation. From the valeur
of Saussure to the systems of systemic
linguistics and the choices of transfor-
MEANING AND STRUCTURE mational grammars, then, the only way
in which grammar creates meaning is by
There is one consequence of the initial setting up mutually exclusive choices,
separation of language patterning into and it exists purely as a record of the
two contrasted types that could be very choice itself; the significance of the choice
important. To bring it out clearly we - whether a past tense verb relates to
will use the terms meaning and structure. past time or present or future time or
In brief, the point is that if we ignore the modality - is determined elsewhere.
meaning while we are describing the struc- If we now view the structure/meaning
ture, then of course we have removed divide from the other perspective, and
the meaning and will not be able to get it look at semantics without structure, then
back while we are focused on the struc- the typical way of presenting the mean-
ture. That is one way of expressing the ing is the dictionary. A dictionary sim-
problem of grammar, and it has been ply lists in an arbitrary order, which we
obscured from careful examination by a call alphabetical1 the items that it re-
kind of meaning substitute. This is the gards as being meaningful, which are
curious terminology that we use, things usually the words of the language, and
like positive/negative, singular/plural, it tries to assign one or more meanings
active/passive, and so on. If we look at to each of the words. That is the charac-
them carefully, these terms are of course teristic model of a dictionary. The mean-
quite substantially inaccurate. Singular ings are denied access to the structural
does not always mean one, and plu- organization that can put them together
ral does not always mean more than and show how they work. For example
one. Present does not always refer to one meaning given in a recently-pub-
the time of the utterance, and past cer- lished dictionary for the word white is
tainly does not always mean some pre- counterrevolutionary, very conservative,
vious time. We have learned as part of or royalist; if this meaning is still cur-
our culture to suspend disbelief when rent it would take some ingenuity to
we encounter these terms, and apply a specify the structural circumstances un-
rough criterion of mutatas mutandis to der which it could occur.
their interpretation; singular means So therefore substitutes, again, are of-
not more than one, if whatever it is fered, this time standing in for the lin-
countable, otherwise general reference. guistic organization that has been dis-
The point is that because these terms are carded. There are in semantics two ma-
not sensitive to the meaning, then they jor types of organization that have been
cannot actually be used directly to eluci- imported; one of these is referential se-


mantics, and the other is logical seman- Here is a brief example to show the prob-
tics; let us consider them in turn. The lems of relying on logical analogies too
assumption of referential semantics is that closely. Many commentators have noted
meanings are organized with reference that the conditional if does not always
to the world outside; words have mean- have its logical force, for example in the
ings which can be understood by indi- following instances culled from a large
cating objects, events and attributes in number of candidates in just one category
the world to which they refer; for ab- of The BNC Sampler (spoken business)
stract entities there is the figurative which is obtainable from Christian Aid if
mode which works analagously. This is people want.
simple and seems to be broadly usable Im just thinking for the meeting if we could
photocopy some Yes
for a very large range of usable phenom-
Ill be actually chairing the meeting for him.
ena, and is widely used in education, So if youd like to kick off
but from a theoretical point of view it is Mm. Yes. Mm. Erm Mm I could if I could just
absurd. Consider the proposal for a pick up one other point about you know
moment - on the one hand there is lan- And another one noticed casually in
guage, which we know is a highly orga- reading:
nized phenomenon that operates under If you believe me, I swung along that road
major constraints such as linearity, and whistling. (The Thirty-Nine Steps, John
on the other hand there is the world, Buchan)
which after thousands of years of research And one which has already occurred
we still see as pretty chaotic, exception- in this paper (one of two)
ally complex and totally unable to be If we look at them carefully, these terms are
encompassed in a simple description. We of course quite substantially inaccurate.
are asked to accept that reference to the
world can elucidate the structure of lan-
guage? We have some reason to believe THE AXES OF LANGUAGE
that language can elucidate some aspects PATTERNING
of the world, but hardly the other way
round. At best the referential links can We now move our perspective to a close-
help in for example supporting the ac- ly related dichotomy that has long been
quisition of language by a child, before recognised in language description - the
the child can cope with semantic abstrac- two fundamental axes of language pat-
tions. terning, the paradigmatic and the syntag-
The other type of imported semantic matic. They are usually depicted as hor-
structure that is popular is logical se- izontal and vertical, with the syntagmatic
mantics. It seems to have some advan- axis on the horizontal, because the lan-
tages, being rigorous and much of it be- guages of modern Europe are written in
ing quite close to the patterns of natural horizontal lines, and the paradigmatic on
language (as well it might be, being de- thevertical2 . The paradigmatic axis spec-
rived from them). But it is crucial to the ifies the possible choices at a particular
understanding of natural language that position on the syntagmatic axis, and the
the organisation is not exact, and is not syntagmatic axis controls the structure
reliable as an indication of logical rela- which is being elaborated. So what we
tionships. As with the definition of terms observe in language text is the syntag-
in grammar, there is again the problem matic; the paradigms are the total of what
of the partial fit, the inexact fit. might have been chosen instead.
John Sinclair LEXICAL GRAMMAR 195

Now, one of the interesting things about Let us consider the grammatical posi-
these two axes is that they cannot be tions a little more. In phrase-structure
simultaneously observed; you must hold rules like
one of them steady in order to look at S->NP VP
the other. We shall return to this point, the only relationship between NP and
but no doubt this is the reason why we VP is that they are both derived from S
have had the division into grammar and in the same operation; their sequence is
lexis from an early stage. It is important also determined in this single step. In
to notice that the theoretical develop- the early days of generative grammar
ment of grammars in recent years has there was a plus sign in between NP and
moved across this divide. If we were to VP,
map the grammar composite (includ- S->NP+VP
ing syntax, structure) and the lexis one but this signalled a quite spurious rela-
onto the two axes, then the obvious pair- tionship pertaining on the syntagmatic
ing would be grammar on the horizontal axis, and became unfashionable.
axis and lexis on the vertical - a model of Where syntagmatic patterns come into
language often called the slot-and-fill- being by declaration, there is no expla-
er model, the one presented at the out- nation of where they come from or how
set of this paper. The syntactic struc- they are to be deployed. The structure
tures form a series of slots, and these are of an English clause is said to involve
filled with choices from the dictionary. Subject-Predicator-Object-Adjunct, for
The well-known models of transforma- example, but these categories are mutu-
tional grammar are partly structured in ally defining, and do not have meaning
this way, for example at the interface until they are mapped into sets of choic-
between the phrase structure and the es, for example that a transitive clause is
lexicon, where the phrase structure spec- one without an object. So, neither in the
ifes the features that any word must have study of the lexis of the language nor in
in order to make a well-formed sentence, the study of the grammar of the language
and the lexicon associates each word with are the syntagmatic patterns given mean-
a bundle of features. However, other ing. This is to a great extent because
influential models insist that they are there is no framework within which they
primarily, if not exclusively, paradigmatic can be shown to have meaning, because
- notably Systemic-Functional Grammar meaning is largely held to reside either
(see Halliday 1995, 15). in the grammatical choice - on the para-
The syntagmatic patterns of language digmatic axis - or in the lexical choice of
are not given meaning in a paradigm a word to deliver a meaning.
grammar, nor, of course, are they given
meaning in a dictionary type of lexis.
The syntagmatic patterns in a grammar SYNTAGMATIC MEANING
are either offered as related through a
common node, or they are simply de- There is no effort, let us say in summary,
clared. The syntagmatic patterns of lexis to discover or create meaning on the
only appear in the byway of idiomatic syntagmatic axis; it is the responsibility
phrase, where they are offered as joint of a paradigm grammar to build in all
realisations of a single meaningful unit, possible syntagmatic meaning as con-
indicating that they have no meaning in straints on the paradigmatic choices. But
themselves. such a venture would be remarkably

complex, so in practice those grammars Now it is an axiom of the present ap-

fail to describe carefully enough the com- proach to corpus linguistics that mean-
binations of choices that are just as cen- ing and cotext are inter-related in such a
tral and meaningful and rule-governed way that involves at least partial cose-
as the single paradigmatic choices. They lection; so the knock-on effect of a para-
give tacit approval to the well-formed- digmatic choice will be felt on the syn-
ness of millions of sentences which range tagmatic axis. If we start from the other
from the odd to the bizarre, and - by axis, then any existing or proposed pat-
claiming as a series of choices phrases tern of choice on the syntagmatic axis
which we know to be a single choice - provides a framework for the interpreta-
they claim large amounts of meaning tion of any choice to be made on the
which we know those choices do not cre- paradigmatic axis.
In corpus linguistics, by contrast, we
have to work on the assumption that PRACTICAL CONSEQUENCES
meaning is created on both axes; for
want of more accurate information we The remainder of this paper gives some
may assume that they contain equal indications of the direction in which this
meaning potential. There is no reason argument is heading and the kind of
why one should have a priority in mean- consequences it is likely to lead to. First
ing potential over the other. We as- we will re-examine the nature of choice
sume a rough balance between what I and meaning, then look further into the
have called (Sinclair 1996) the phraseo- meaningful terminology of grammar,
logical tendency, the tendency of a speak- and finally pose a question about an
er/writer to choose several words at a important type of meaning that is large-
time, and the terminological tendency, the ly ignored by both the grammatical and
tendency of language users to protect lexical traditions.
the meaning of a word or phrase so that
every time it is used it guarantees de-
livery of a known meaning. As we get MEANINGS FROM NOWHERE
to know more, these assumptions may
well be revised. Let us begin by revisiting the informa-
Above we have presented a model of tion-theoretic model of paradigm gram-
language as a balance between opposing mar, which says that choice equals mean-
forces related to the two axes of lan- ing, that the number of choices deter-
guage patterning, and above that is an mines the amount of meaning available
assertion that the two axes cannot be si- in each case, and the precise positioning
multaneously observed; these sound like of the choice in the structural framework
good reasons for keeping them apart, determines much of the type of meaning
and describing them separately. How- that will be created by the choice. A
ever, the argument of this paper is that if description within this model must take
pattern and meaning are to be aligned, great care that each set of choices is ac-
then the two axes have to be inter-relat- tually relevant and applicable at each
ed for as long as possible in the descrip- point. Because if it is not - if another
tion. Consider, for example, the classic factor in the environment is affecting the
model where a choice is made on the range of choices on offer, then unless the
paradigmatic axis, which will lead to a grammar is revised it is creating more
particular word appearing in the text. meaning than is in fact available.
John Sinclair LEXICAL GRAMMAR 197

If this manufacture of illusory mean- extent of the variation associated with

ing is institutionalised throughout a com- this phrase; the remaining words are fixed,
plex grammar, there are two obvious con- and do not realise any choice beyond the
sequences. One is that the grammar (and first, overall choice of meaning. So nei-
the grammarians) are misled into think- ther of the occurrences of of above are
ing that their apparatus is more power- the normal occurrence of the preposi-
ful than it actually is; the other is that tion, because of is fixed in this phrase3 ,
there is little meaning left over to be and so are the, corner and eye.
assigned by the lexical structure of the The word my can be alternated with
language. Now that it can be demon- other possessive adjectives. So here we
strated by corpus evidence that a large have a seven-word phrase which realis-
proportion of the word occurrence is the es one overall choice and at most two
result of co-selection - that is to say, more subsidiary choices. The choice between
than one word is selected in a single out of and from here is a stylistic
choice - every time that this can be dem- choice rather than a choice that delivers
onstrated there is one less item of mean- a totally different type of meaning - there
ing to be allocated to the grammar. If are not two different places, and out
you have two words that are selected in of and from are just different ways of
the same choice, then they cannot be in- expressing the same basic position.
dependently selected. Early estimates These single choices can consist of sev-
were that up to 80% of the occurrence of en words with ease; the phraseology of
words could be through co-selections, English quite frequently produces co-
which would leave, of course, only 20% selections of five, six and seven words,
for the sort of independent paradigmat- and there are even some of up to twelve.
ic choices of the grammar. A recent pa- In this connection Miller (1956) comes to
per by May Fan (1999) gave hard evi- mind. Miller showed that for most peo-
dence for this in regard to one of the ple the short-term memory handles sev-
common verbs in English. en items with ease.
Let us work through a characteristic
example. There is a phrase in English, a
common recurrent phrase, out of the CROSS-BORDER CATEGORIES
corner of my eye, as in I saw some-
thing out of the corner of my eye. There Corpus evidence consistently shows that
are seven words in the phrase, and they the ways in which a meaning can be re-
all simultaneously choose one unit of alised extend well beyond the definitions
meaning, to do with peripheral vision. of grammatical categories. In pointing
Within this primary meaning, there are out above that grammatical terminology
one or two variants of individual words, did not correspond to semantically co-
and this is where the corpus is essential, herent categories, we did not tell the whole
because the intuition cannot be relied story. Consider a term like negative,
on. out of can sometimes be replaced which will contrast with positive in a
by from, and my is a possessive two-term system of polarity. There
adjective that can have other, but prob- are a number of realisations of gram-
ably only singular forms; people do not matical negatives in English, no and
collectively see things out of the corners not and so on. There are also semi-
of their eyes, so I think their eyes is negatives like hardly and scarcely,
going to be very unusual. This is the full which share a number of features with

true negatives, but not all; these are not of, for example, the grammatical choice
normally considered as grammatical neg- between positive and negative to the study
atives. of meaning when negative meaning can
But there are also morphological neg- be created in so many alternative ways;
atives like the prefixes un- and in-, and, more fundamentally, how valuable
which are not recognised in a clause gram- is it to be able to point out that there are
mar, so that I am unhappy is positive many clauses which are grammatically
and I am not happy is negative. We negative but in relation to meaning, pos-
also find that negation as a concept can itive, and vice versa?
be lexicalised, so that the verb refuse
for example has a negative force; he
refused to go is the same as he would SEMANTIC PROSODY
not go and yet it is a positive clause in
the grammatical sense. Another important point to be made in
It is easy to understand the grammar- the study of lexical grammar is the emer-
ians wish to keep negation pure and gence of many latent categories of mean-
simple; to accept lexicalised negation is ing, which have not been recognised in
a slippery slope, and no-one knows what published grammars, and only occasion-
lies at the bottom of it. But if we are ally in the very latest dictionaries. The
intent on elucidating the meaning of run- first to be noticed were of the type some-
ning text by analysis, then all these dif- thing nasty or something worrying
ferent ways of indicating negation are or disturbing; later others like some-
perfectly acceptable realisations, and if thing magnificent, socially appropri-
supported by corpus evidence we can ate, positively constructive etc. These
take them all together, straddling the are showing up as repetitive categories
borders between grammar and seman- that are neither completely grammatical
tics. This straddling is an important fea- nor completely lexical but are neverthe-
ture of lexical structure; lexis is not the less very important from a structural point
residue of a grammatical description, but of view. So once again we have to allow
a different way of describing the same for the meaningful categories not to be
events; it is not bound by the conven- confined within the grammar as it is
tions of grammar, and it can recognise a normally presented, and if we divide
wide variety of realisations of meaning- language into these two major catego-
ful choices. ries, then we will never be able to get
The grammarian is left in a dilemma; them satisfactorily together again; also
the more sensitive grammars recognise we have to add that the grammar cannot
that categories of meaning like nega- be trusted to set up such essential cate-
tive, modal, possessive can readily gories of meaning because it is not sen-
be lexicalised - or to be more neutral, sitive to them4 .
can occur in grammatical or lexical or Here it has to be said that the percep-
morphological realisations - so a com- tions of native speakers are not to be
plex realisation route is devised for them. trusted either; the referential element in
The particular way in which they are meaning is frequently assigned a priori-
realised is then of secondary importance ty over the attitudinal, for reasons that
compared with the primary creation of are not justifiable; clearly an awareness
meaning, which is the operative process. of both aspects of meaning is necessary
The question must arise of the relevance for accurate deployment of the lexical
John Sinclair LEXICAL GRAMMAR 199

item, and if this is not available it is ging - in other words it was found nec-
arguable that more difficulty may arise essary to use some syntactic information
from a mistake on the pragmatic side in order to complete what was originally
than on the referential. To give a real- a morphological analysis.
life example, in the preparation of a dic- This movement of focus is well recog-
tionary for native speakers of English by nised in corpus linguistics - the need to
the Cobuild team, there was a strong examine the context of an item in order
feeling among editors and publisher that to determine its function or meaning. But
whereas for learners of English it may be nothing seems able to shake belief in the
necessary to state the attitudinal mean- underlying assumption that all the words
ing, this is already available to native of a language naturally fall into a small
speakers. So the Cobuild definition (1987) number of classes. The information from
for scrawny is unpleasantly thin and a computer examination of a corpus sug-
bony, while in Todays English Dictio- gests quite otherwise, as I have argued
nary (1995) it is thin and bony; the two on several occasions6 . Since few inflec-
dictionaries define prattle identically but tions survive into Modern English, and
Cobuild adds an informal word, often since one of the most productive areas of
used showing disapproval5 . development in the modern language is
the ability of words to move across word
classes, it may be preferable to accept
WORD CLASS what the corpus seems to be signalling,
which is the need for a major overhaul
Professional linguists should not be sur- of the notion of word class.
prised to experience a rather disturbing In general, we must move toward a
effect from the massive surge in the avail- theory that reconciles the paradigmatic
ability of evidence and the growing so- and the syntagmatic dimensions and al-
phistication of the tools for examining it lows the description of the language to
and testing hypotheses against it that remain sensitive to both dimensions for
corpus linguistics has brought. Some of as long as the correlation is productive;
the vague but useful categories of tradi- no doubt there will be some residue of
tional language analysis, which have specifically grammatical and specifical-
served humans well for centuries, are ly lexical information after that stage,
not easily replicated in computational but we must wait to see what it is, and
routines; for example parts of speech what categories and processes are best
or word class labelling. Human be- used to describe it.
ings have little difficulty assigning words
to a dozen or so word classes, but ma-
chines have exposed just how untidy a LEXICAL STRUCTURE
categorisation this is. For English, which
has had a lot of attention over many years, At present, the lexical structure is pre-
there is little or no consensus about how sented separately, insensitive to the gram-
many labels there are - the variation from mar in the same way as the grammar is,
one analysis to another is very large - or traditionally, insensitive to the lexis. It
how they are defined. The persistence is probably a valuable exercise to prior-
of researchers has resulted in a signifi- itise the lexical patterning and to push a
cant movement of focus, so that the pro- lexical description as far as it is reason-
cess is now called morphosyntactic tag- able to do so; the justification is that so

little research has been done in this area, These theories will be developed from
especially as compared with the immense the kinds of hypotheses that are taking
attention that the grammar has had over shape in corpus-driven linguistics.
the centuries. But such an effort should
not be misunderstood; it must be seen
simply as an interim step towards an EXAMPLE
eventual holistic description, and there
is no imperialistic dimension to lexical Let me give as a conclusion an example
description. of the kind of semantic prosody that Im
In the meantime, there are structures talking about. Consider the English word
of a particularly lexical nature that are effort; it is a countable noun and so it has
worthy of attention, and which are in- a plural, efforts. And one of the most
troduced in recent publications, particu- notable collocates of efforts is the word
larly Sinclair (1998). These begin with to, which follows efforts, and which is
collocation, the co-occurrence of words, the infinite marker. So essentially we
and go on to colligation, which in this are focussing on a structure which has a
work is defined as the co-occurrence of core of efforts plus an infinitive. In the
words with grammatical choices, then Bank of English7 in Birmingham, which
semantic preference, which is the co-oc- is the reference corpus that I normally
currence of words with semantic choic- use, the one that lies behind the Cobuild
es, and semantic prosody. The semantic publications, there are 9,617 instances of
prosodies express attitudinal and prag- efforts followed by to. For the Figure, the
matic meaning; they are the junction of computer has selected 21 by the simple
form and function. The reason why we expedient of picking the first one in text
choose to express ourselves in one way sequence, then dividing 9,616 by 20 (=
rather than another is coded in the pros- 480 in round figures) and then selecting
ody, which is an obligatory component the 481st, 961st etc instance through the
of a lexical item. corpus.
The ways in which the prosody is ex- If we examine these, it becomes fairly
pressed are extremely varied, and seem clear that we use this phrasing - we talk
to have no limits as to position or shape; of efforts to do something - when they
we can thus anticipate severe technical appear to be very unlikely to succeed, to
problems in retrieving them computa- be heading for failure, or already unsuc-
tionally. This is the central problem in cessful. In other words, the prosody that
analyzing open text and one of the prin- appears in almost every example is the
cipal reasons that the performance of speaker/writers prejudgement of the ef-
devices which depend on some kind of forts, that they are heading for failure.
language understanding is so poor. At So when we are discussing the machine
the present time the goal of the machine understanding of language, if we were
understanding of language is far more to talk of the efforts of computational
difficult than it needs to be, because we linguistics to comprehend natural lan-
are not using appropriate theories - once guage, we would imply that they are
the meaning created by lexical structures doomed to failure. There are a number
becomes available, and integrated with of adjectives, for example, like hysterical,
what we already know through gram- frantic, futile, strenuous; verbs like blun-
mar, then theories will be articulated that der, hamper, were overwhelmed people close
predict the prosodies and the computer ranks against efforts, or achieve things
will then know where to look for them. despite efforts; efforts exhaust us, and so
John Sinclair LEXICAL GRAMMAR 201

on. So if this is a representative sample the lexical structure of the language, and
of the behaviour of the word, we can some tentative hypotheses have been
expect to find in the left-hand cotext of formulated.
efforts to, some indication of the likely
failure of the efforts.
In a contrast which is almost ironic, CONCLUSION
we can expect to find in the right-hand
cotext a set of verbs which are creative, Despite the recent rush to welcome cor-
which talk about creative action, like pora into the resource collections of many
please, revive, work together, protect, sup- students of language, we must note that
port, gain, raise, activate, kindle (a debate), the vast majority of work with corpora
help, give (the city something good), save, still takes place under the assumptions
etc. So before a reader/listener discov- of pre-corpus linguistics, and is thus in-
ers that the efforts are to do something sensitive to the possibilities put forward
constructive and beneficial, they are al- here. It is clear that the first step to-
ready sabotaged. Our first draft of the wards a new view of language has now
lexical item that has as its core efforts to been taken by the linguistics profession,
will thus contain three elements of struc- in recognising that corpora are relevant
ture - the core, the semantic preference and useful; this has been effectively com-
for a verb of constructive action, and the pleted approximately thirty years from
semantic prosody of anticipated failure. the advent of electronic corpora. It is
The selection of the item is controlled by only natural that to begin with scholars
the prosody, because the whole point of will appreciate the security of familiar
expressing oneself in this way is to pre- concepts in engaging with such a total
evaluate the actions, which would oth- revolution in the availability of evidence
erwise be positively evaluated by the of usage, and only gradually will they
reader/listener. accept that some of those concepts are
It is likely that other expressions with sorely in need of being revised and up-
structural similarity to this tentative item dated.
will be found; attempts to may be simi- The initial separation of grammar and
lar8 , etc. The singular forms, effort, at- lexis in language description, and the
tempt may show some tendency in the subsequent prioritisation of grammar at
same direction. A set of forms sharing the expense of lexis, is one of the most
similar meanings could be a further step firmly-held positions among theoretical
in the mapping of an organisational frame- and descriptive linguists, and it will take
work for lexis. This work is only begin- some time before it is held up to scrutiny
ning; a few probes have been made into and approached with an open mind.

Alphabetical order is an order whose only house in UK. A fruit machine consists of three
virtue is that it is taught to all literate mem- revolving cylinders, each of which bears a
bers of societies which use it. The fact that it number of drawings of fruit. The player pulls
is the only means of organising the vocabu- a handle at the side which causes the cylin-
lary of a language merely emphasises the fail- ders to spin independently of each other, and
ure of linguists to find a better one. come to rest in a chance combination. A row
The most available instance of this mecha- of three fruit thus appears in a central grille,
nism in some societies is the fruit machine, and if the row corresponds to one of those in
which used to be found in almost any public a list on the side of the machine, (eg banana

banana banana) then the player wins, and I argued that most of the common words in
receives several times his or her stake. The English have individual patterns of occur-
central row corresponds to the syntagmatic rence, and do not fit into the general word-
axis, and the cylinders contain the paradig- classes.
matic choices; a well-formed structure is one (c) As a contribution to the NERC Report
of those that wins a prize. (1996), I pointed out that in English, as well
It would be a digression to argue here that of as words which function as nouns, and those
is not a preposition in such structures; for which function as verbs, there is a substantial
that see Sinclair 1991 class which function as both (I called them
Here the new Pattern Grammars (Hunston norbs). This is a kind of underspecification
and Francis 1999) take the innovative step, which postpones a very difficult set of deci-
guided by corpus evidence, of associating sions until perhaps the analytical system is
some of these meanings with structural pat- better able to deal with them.
terning. 7
At the time of retrieval the Bank contained
This was one of the few arguments that I, as almost 350 million words of broad general
Editor-in-Chief of Cobuild, lost; but I am still English text, from native speakers in many
puzzled at the conviction that native speak- parts of the world, their spoken and written
ers may need to know the referential mean- expression. The corpus is jointly owned by
ing of a word but not its attitudinal/prag- HarperCollins, publishers, and The Universi-
matic one. ty of Birmingham, and access to it can be
(a) in Tickoo (ed) 1989, reprinted as Chapter arranged via the Cobuild Home Page,
6 of Sinclair 1991, I showed that the second <www.cobuild.collins.co.uk>
commonest word in English, of, had very lit- 8
The appearance of despite, failed, unsuccessful,
tle in common with other prepositions, and desperate, repeated, several,as very significant
was mainly used in a unique syntactic func- collocates of attempts to suggests considerable
tion. similarity, but measures have not yet been de-
(b) I followed this up in Sinclair (1999), where vised to compare collocational profiles.

Fan M. 1999. An investigation into the perva- Sinclair J. 1999a. A Way with Common
siveness of delexical chunks in authentic lan- Words. Out of Corpora, by H.Hasselgrd and
guage use and the problems they present to S.Oksefjell. Amsterdam and Atlanta: Rodopi.
L2 language learners. Language Analysis, Sinclair J. 1999 b. The Lexical Item. Contras-
Description and Pedagogy, ed. by R.Berry et al. tive Lexical Semantics, ed. by E.Weigand.
Language Centre, HKUST, Hong Kong. Amsterdam/Philadelphia: John Benjamins.
Miller G. 1956. The Magical Number Seven, Volume 17 of series Current Issues in Lin-
Plus or Minus Two: Some Limits on Our Ca- guistic Theory, 124.
pacity for Processing Information. The Psy-
chological Review 63:8197. REFERENCE WORKS
NERC Report 1996; Network of European Refer- Collins English Dictionary, Hanks et al 1998.
ence Corpora Project Report. Pisa, Giardini. Collins Cobuild English Language Dictionary,
Sinclair J. 1991. Corpus, Concordance, Colloca- Sinclair et al 1987.
tion. Oxford: OUP. Todays English Dictionary, Sinclair et al 1995.
Sinclair J. 1996. The Search for Units of Mean-
ing. TEXTUS IX 1.


the Hungarians would continue their efforts to please on the football pitch.
made love parsimoniously. Gaspards efforts to revive their passion exhausts
to support poor people in their own efforts to work together. Development
at the plan and that farmers efforts to protect their livestock, crops
difficulties, and to increase their efforts to support Eastern Europe. At the
John Sinclair LEXICAL GRAMMAR 203

last Thursday and could hamper their efforts to gain access to Polly Pecks
arts companies, at their continuing efforts to raise additional funds from the
First Division, made more serious efforts to activate their attack. Smith
heroes for their - albeit futile - efforts to prevent the countryside from
office, due to their own hysterical efforts to keep their seats. <p> Yours
broke through police lines in their efforts to get on to the Garvaghy Road
new weapons. <p> Despite their efforts to kindle a debate in France - and
were likely to make strenuous efforts to help their employees get to
workers were overwhelmed in their efforts to deal with thousands of refugees
Assistens Kirkegaard. In their efforts to give the city still more green
close ranks against government efforts to control their work. Indeed, it
joined neighbours in their frantic efforts to save the family. Later,
while just possibly making some efforts to get their feet on the bottom
and WRC officials will suspend their efforts to sign players to contracts until
the attack yesterday switched their efforts to a murder inquiry. <p> Mrs
blunder desperately in their efforts to contain the uncontainable.

Gauta 2000 06 13
Parengta 2000 06 30



ia vartojamas teminas leksin gramatika, apimantis biau vien ar kit, o tai lm mintj struktros
dvi svarbiausias sudedamsias kalbos dalis, kuri ir reikms arba gramatikos ir leksikos atskyrim,
sveikoje yra kuriama reikm. Nuo seno jos buvo pirmenyb teikiant pirmajam. Pagal sigaljusi nuo-
atskiriamos ir prieinamos. is straipsnis rodo to- mon kalbos struktra, sintagminis jos modelis yra
kios dichotomijos nepagrstum ir keliam pavoj pirmapradis darinys konstruojant sakin, pagal jo
skaidant kalb sudedamsias dalis prarasti visu- atveriamas vietas vliau i paradigmins aies pa-
mos vaizd ir jame matomus modelius. Tradicin gal reikm yra parenkami leksiniai vienetai, upil-
dichotomija labiausiai griauna kompiuteri suteik- dantys jiems skirtas sintagmines kalbos struktros
t galimyb analizuoti didelius tekst kiekius ir ertmes. Taigi reikm beveik iimtinai paliekama
aptikti bei itirti leksins raikos dsningumus (le- paradigmatikai.
xical patterns), irykjanius i plika akimi neapvel- Tekstyn lingvistikoje yra daroma prielaida, kad
giamos labai didels konkrei apraik vairovs. reikm kalboje kuriama abiejose jos dimensijose,
Leksikos ir gramatikos atskyrimas suponuoja ir paradigminje, ir sintagminje, todl negalima
reikms ir struktros skirtybes. Nuo F.de Saussu- reikms atskirti nuo struktros, turinio nuo for-
reo laik suvokiant kalb kaip vertikalij para- mos. I ia kyla poreikis keisti kalbotyros terminus
digmini ir horizontalij sintagmini ryi vie- taip, kad jie neklaidint vartotoj ir atspindt
nov, struktra buvo labiau siejama su sintagmi- iandienin kalbos esms suvokim, taigi kalbti
niais, o reikm su paradigminiais ryiais. Ka- ne apie leksik ir gramatik, o apie leksin grama-
dangi abu kalbos ryi tipus yra sunku suvokti i tik, toki kalbos tyrim srit, kurioje abiem as-
karto, paprastai tyrintoj buvo orientuojamasi la- pektams bt skirtas vienodas dmesys.