Paragrammatism

The (deviant) substitution (or omission) of grammatical morphemes in spontaneous speech

(substitution makes it different to agrammatism)

Abnormal grammatical sequences

Paragrammatism
Around as a term since 1914!

Originally seen as a problem of word order

But always seen as different to Agrammatism

Paragrammatism

It is normally associated with fluent aphasias such as Wernickes aphasia

And is seen in language production

Paragrammatism

Difficulties in comprehension are also a feature of fluent aphasia

Paragrammatism

If a Wernickes aphasic patient has good communicational and interactional skills they can sometimes disguise the comprehension deficit

Paragrammatism

Wernickes aphasic patients may be unaware of the comprehension difficulty especially in the early days after onset

Paragrammatism

Fluent aphasic patients find meaning dependent on syntax more difficult to understand, and find understanding complex sentences more difficult then simple sentences

Paragrammatism
Production: Subjects may make mistakes; attempting to say one thing but saying another. Target: what they want (or are instructed) to say Response: what they actually say

Paragrammatism

Testing See Edwards (2005: 98-104)

Paragrammatism
Testing A patient is asked to look at a picture and describe that picture in one sentence. 20 pictures in all. The patients spontaneous speech is fluent and he displays both well and ill formed sentences. The language produced by the patient can be compared to the language of non-aphasics users performing the same task

Paragrammatism
Target
The man is running The man is walking The man is painting the woman The boy is hitting the girl

Response
The man running Walking down the street Painting the picture The lady box

Paragrammatism
Out of four sentences, three of the four lexical verbs, running, walking, painting, are produced while the verb hitting in sentence (4) is substituted for a semantically related verb box, which is like hit in meaning. There were also substitutions and omissions made with subject NPs e.g. the man in sentence (2, 3) is omitted, and the NP in object position in sentence (4) is substituted e.g. girl replaced by lady.

Paragrammatism
Although tense inflection is not seen in these examples we cannot assume it is missing altogether as the aspectual inflection -ing is present.

The preservation of ing has been documented as a feature of non-fluent aphasia and has not been generally recognised as a feature of fluent aphasia.

Paragrammatism
All four sentences are missing the constituent under T (Tense node). Three of the sentences have the correct verb while the fourth does not.
Substitutions suggest the underlying node is intact omissions suggest the underlying node is unavailable or missing.

The sentences produced by the patient showed that the patient understood the pictures and the task.

Paragrammatism

Fluent aphasic speakers produce fewer subordinate clauses than the normal controls.

However they also produced more canonical sentence structures than normal controls.

Paragrammatism

Fluent aphasic speakers produce fewer subordinate clauses than the normal controls. However they also produced more canonical sentence structures than normal controls. So, avoiding complex structures.

Paragrammatism
This greater prevalence of canonical structures in fluent aphasics speech was shown to be the case across a wide range of languages.

The overall finding of a reduction in sentence complexity in fluent aphasic speakers of different languages tells us that this is a feature of fluent aphasia and is not a language specific feature.

Paragrammatism

Edwards and Bastiaanse (1998) showed that fluent aphasic speakers although able to produce main and subordinate clauses used significantly fewer than the control group.

Paragrammatism

So something is making it difficult for them to deal with complex structures

Paragrammatism

Testing can take place using elicited speech Or using spontaneous speech

Paragrammatism

Connected fluent aphasic speech samples are assumed to be more naturalistic. Therefore more representative of aphasic language abilities.

Paragrammatism
Patients are recorded in a variety of situations;
Speaking with relatives, carers, speech and language therapists, and other researchers. The samples vary in length, and manner. The aphasic patients know they are being recorded. The samples gathered are compared to normal non-aphasic individuals. Speech samples are usually about two minutes long or three hundred words.

Paragrammatism

The task can fairly straightforward such as re-telling the cinderella story!

Paragrammatism
Hagoort and colleagues (1999), suggest that syntactic, semantic, and pragmatic information all play a role in determining the meaning of a sentence. Therefore sentences given in a test situation with no additional contextual information may be understood less than the same sentence within a communicative process where extra syntactic, semantic, and pragmatic information is available (Edwards, 2005: 157).

Paragrammatism

Wernickes aphasics were aware of the syntactic and pragmatic information but were unable to integrate it. Wernickes aphasics have deficits with semantic combinatorial operations, and also difficulties in lexical-semantic integration processing.

Paragrammatism
Brocas aphasics are said to have reduced lexical activation while Wernickes aphasics are said to have increased lexical activation. Both under-activation and over-activation can result in a deficit for combining lexical, syntactic, and pragmatic information in thematic structures.

Paragrammatism
Brocas aphasics are said to have reduced lexical activation while Wernickes aphasics are said to have increased lexical activation. Both under-activation and over-activation can result in a deficit for combining lexical, syntactic, and pragmatic information in thematic structures. The former because lexical entries would only map on weakly to the thematic roles, and the latter because the overly activated system would be unable to select the appropriate thematic structures from all of those activated.

Paragrammatism

A Lexically driven account suggests that word finding difficulties, or failure to activate the appropriate word would therefore impair syntax and sentence formulation.
The syntactic frame required by a verb for instance would not be completely available so the paragrammatic would be unable to properly construct the sentence.

Paragrammatism
In the study by Faroqi-Shah, and Thompson (2002) the relation between lexical retrieval and sentence formulation was examined.

Seven Wernickes aphasic speakers and seven Brocas aphasic speakers participated, and narrative speech samples of the Cinderella story were obtained from each participant using different pictures depicting actions accompanied by varying lexical information i.e. nouns and verbs, to elicit the speech samples.

Paragrammatism
The study examined the production of reversible and non-reversible actives and passives in Brocas and Wernickes aphasia. Wernickes aphasic subjects showed difficulty with producing and comprehending passive sentences, in particular passive sentences with reversible roles e.g. the man was hugged by the woman, just like agrammatic aphasics.

So it looks as if paragrammatism and agrammatism arent as different as was once thought!

Paragrammatism

Brocas aphasic subjects produced more morphological errors than Wernickes aphasics who were more impaired in objectnaming.

Paragrammatism

They both showed some similarities across the tasks i.e. grammatical morpheme errors, role reversal errors, preposition errors, unrelated sentences, and non-sentences. The most common error noted was the difficulty with reversible passives.

Paragrammatism

In conclusion the authors failed to find a direct effect of lexical cues on the success of sentence production both in Brocas and Wernickes aphasia, suggesting the relationship between lexical and syntactic mechanisms is far more complex than a lexically driven account implies.

Paragrammatism
They also note the underlying sentence production deficits seem to be different for both Brocas and Wernickes aphasic subjects. In Wernickes aphasia, they suggest the difficulty is an accessing and self-monitoring failure. In Brocas aphasia, they suggest the difficulty is in retrieving grammatical morphemes.
(Faroqi-Shah and Thompson, 2002: 412-426)

Paragrammatism
Fluent aphasics and normal controls Qualitatively similar, quantitatively different Errors more frequent in fluent aphasics But do they have the same cause?
Butterworth and Howard (1987)

Paragrammatism

The word can refer to a syntactic error or to an aphasic syndrome Paragrammatism can co-occur with Agrammatism Most work has been done on Agrammatism

Paragrammatism
Types of explanations
Syntactical disturbance Lexical selection impairment

Monitoring failure

Paragrammatism
Types of explanations
Syntactical disturbance Impairment on the use of function words and bound grammatical morphemes

Paragrammatism
Types of explanations
Lexical selection impairment No problem in the syntax, but problems in word finding

Paragrammatism
Types of explanations
Monitoring failure

Loss of the ability to monitor and correct ones speech output This should be accompanied by poor comprehension of others speech

Paragrammatism
Types of explanations

Control impairment Just normal errors, at a greater rate!

 11,000 words from 5 fluent aphasics 12,000 words from 4 non-aphasics

open class lexical errors closed class lexical errors errors of grammatical inflexion constructional errors (constituent order violations etc.) residual errors.

Paragrammatism
Types of explanations Syntactical disturbance: Problems in inflection, lack of complex constructions

Lexical selection impairment: Neologisms, mistakes in open class terms, mistakes correlated with rate of neologisms, no association with poor comprehension Monitoring failure: poor comprehension and correlation with neologisms

Paragrammatism
Types of explanations
Control impairment: Greater frequency than normals, but of the same type, no correlation with neologisms or comprehension

Butterworth and Howard

Samples of at least 1300 words of unrehearsed conversational speech were recorded for five aphasics and four normal subjects

Butterworth and Howard

Errors were divided into five broad categories: (1) Open class lexical errors - where the substitution, omission or addition of a single noun, verb, adjective or some adverbs rendered the sentence ungrammatical; (2) Closed class lexical errors where the substitution, omission or addition of a single preposition, pronoun, determiner, quantifier, conjunction or some adverbs (e.g., not, only) rendered the sentence ungrammatical ; (3) Inflectional errors - where the substitution, omission or addition of an inflectional affix rendered the sentence ungrammatical;

(4) Constructional errors - where the order of words or other determinable grammatical process yielded an ungrammatical sentence. This includes the putative blending of two or more sentences;
(5) Residue - other ungrammatical strings not falling into categories l-4.

Butterworth and Howard

Errors were divided into five broad categories: (1) Open class lexical errors - where the substitution, omission or addition of a single noun, verb, adjective or some adverbs rendered the sentence ungrammatical; Category: In these, the speaker substitutes a word of the wrong syntactic category, for example, uses a noun where he should have used an adjective Subcategory: In these, a word with the correct part of speech - major category - is chosen, but its subcategorisation is inappropriate in context. There will become a time Omission: These are errors where a noun, verb or adjective has been omitted. Addition: (only one example, and in one of the controls speech) They were all had pleasant sandy beaches.

Butterworth and Howard

(2) Closed class lexical errors where the substitution, omission or addition of a single preposition, pronoun, determiner, quantifier, conjunction or some adverbs (e.g., not, only) rendered the sentence ungrammatical Category Substitutions: This is where, for example, a pronoun is substituted for a preposition, or an auxiliary verb for a determiner. Subcategory substitutions: In these, although a word from the correct major category has been selected, e.g., a preposition, it is from the wrong subcategory in the syntactical context. Itd take me at least 5 minutes to my house down to the bus station

Omissions: Here a closed class item is omitted. Oh thats __ bag, isnt it?
Addition: I am also on a tenant

Butterworth and Howard

(3) Inflectional errors - where the substitution, omission or addition of an inflectional affix rendered the sentence ungrammatical A solitary mice was chased by the cat. Both the mouses were chased by the cat.

Butterworth and Howard

(4) Constructional errors - where the order of words or other determinable grammatical process yielded an ungrammatical sentence. This includes the putative blending of two or more sentences The most striking paragrammatisms are those which cannot be explained in terms of an error of single word or inflexion, and seem to result from mistakes, not in the choice of items to go into a sentence, but in the process of constructing the sentence itself

Sentence blends - Isnt look very dear, it Isnt very dear, is it + Doesnt look very dear, does it Tag errors But its silly, arent they, Now its gone nasty, isnt it? Illegal NPs in relative clause gaps The boy whom the girl kissed him was angry Pronoun headed relative clause in object position The woman there was very jealous of her who was new in the office.

Syntactic Disturbance
In general, as we have tried to demonstrate for each category of error, those constructions involved in patients paragrammatisms are also found in the paragrammatisms of our normal control subjects. Thus we have to conclude that there is little support for the idea that aphasic paragrammatisms arise as a consequence of some permanent loss or corruption of grammatical rules or grammatical knowledge.

Lexical Selection Impairment

Although nearly half of all paragrammatisms can be accounted for in terms of a selection failure concerning a single word, the other half have no account under this hypothesis. Moreover, to the extent that neologisms indicate a deficit in lexical selection, there is no significant correlation between the rate of neologising and the rate of paragrammatisms.

Monitoring Failure
According to Levelt (1983) a speaker monitors the correctness of his speech (not necessarily only after it has been emitted) using the same mechanisms as are involved in the comprehension of the speech of others.

Monitoring Failure
A wide range of comprehension abilities are, therefore, associated with paragrammatic speech, and at least two patients had good sentence comprehension. It is worth noting that all patients produced the same kinds of paragrammatic error, and there is no quantitative correlation between the degree of comprehension impairment and the incidence of paragrammatism. Moreover, the kinds of paragrammatism found in the patients were found also in our controls who, though not tested, can be presumed to have relatively intact comprehension.

Control Impairment
The control impairment hypothesis predicts that errors will be of the same kinds as are found in the speech of normal subjects, and that constructions involved in error will also be found correctly employed. These predictions are well supported by the data.

Control Impairment
Globally, the independence of error types follows from the independence of the modules, and the presence of all aphasic error types in the comparison corpus will be a consequence of transient control malfunctions rather than permanent malfunctions in the systems themselves. So a temporary lack of control at any stage of the derivation of a sentence can result in a paragrammatism.

Control Impairment
So a temporary lack of control at any stage of the derivation of a sentence can result in a paragrammatism.
The type of error results from the part of the derivation where the lack of control takes place. Semantic Lexical Syntactic Phonological/Phonetic

Integration theory
The integration of lexical and grammatical information fails

Integration theory
The integration of lexical and grammatical information fails The claim is that verbs are harder to retrieve in Paragrammatism

And significantly harder to retrieve where there is also lots of syntactic work to do

 The structures used by fluent aphasic speakers are grammatically less complex and less elaborated than normal In highly inflectional languages, like Italian and Finnish, inflectional errors have been mentioned as well
Bartiaanse (2010)

Lexical Verbs
First of all, there are lexical verbs, which are considered to be open-class words: they have a meaning and can be used to refer to actions, events et cetera. Lexical verbs can be finite, when they are inflected for tense and agreement (he walks, we walked), and non-finite, when they are combined with another verb (to walk; walking, we have walked).

Copular Verbs
Second, there are the copular verbs, the most frequent of which is to be.

Modal Verbs
Third, there are the modal verbs. Modal verbs are most often used as finite verbs, but in English, these are only inflected for tense, not for agreement (he can, he could). In Dutch, modal verbs are inflected for both tense and agreement, but, like in English, most modal verbs have irregular inflection paradigms. Therefore, the different forms of the Dutch modal verbs are assumed to be stored as separate entries, like copular verbs. Modal verbs can be used with and without a lexical verb (yes, we can; yes, he can swim). They are closed-class words.

Auxiliary Verbs
Fourth, there are the auxiliary verbs, which are always used in combination with other verbs. To have in combination with a participle refers to the past, to be in combination with a participle denotes the passive voice (he is pushed), and in combination with a gerund the progressive (he is pushing). What auxiliaries have in common with modals and copulas is that they are irregular and highly frequent. The difference is that the auxiliaries to have and to be are always used in combination with another verb that is inflected (either a participle or a gerund). Auxiliaries are closed-class words.

 Bastiaanse and Edwards (2004) studied the production of finite verbs and infinitives in a sentence completion task in (Dutch and English speaking) patients with Brocas and Wernickes aphasia. A picture of an action was presented with a sentence in which the verb was missing. In one test this was a finite verb (the boy__an apple; target eats), in another test it was an infinitive (the child wants to __ to his mum; target crawl). Surprisingly, the pattern of performance of the Broca and Wernickes patients was similar: both aphasia groups were significantly more impaired in the production of finite verbs.

 However, there was a qualitative difference: when Broca patients made errors in filling in finite verbs, they made inflectional errors (in the example above the boy eating an apple), whereas the Wernicke patients produced predominantly semantically related verbs which were correctly inflected for tense and agreement (in the example above the boy drinks an apple).

 Apparently, the need to inflect a verb for tense and agreement reduced the ability to retrieve the correct forms from the lexicon in this group.

 On the basis of the study of Bastiaanse and Edwards (2004), which showed that in an experimental setting the production of finite lexical verbs was more impaired than the production of non-finite lexical verbs, it was hypothesized that in fluent aphasic speech lexical access would diminish when the grammatical form was more complex due to an integration problem. This hypothesis is supported by the data: when lexical verbs have to be inflected for tense and agreement, the lexical diversity of the verbs decreases and the word frequency of produced verbs increases significantly.

 The strategy that these fluent aphasic speakers apply to circumvent this integration problem is to use more frequent verbs and to produce these verbs at a higher rate than normal. In this way, the threshold for a lexical verb is lowered and the verb is easier to retrieve. An extreme example of this multiple use of a finite lexical verb is observed in the following sample. The question he answered was can you tell me how your speech problems started? The finite lexical verbs are underlined.

 only sentences . only sentences . yes beginning that was nothing, basically .. yes, not even not even not even my name not even no numbers nothing and then I thought, well, yes . Then I thought . a day I thought then I thought two days well and not even newspapers then I thought okay you can with your legs and arms that is okay but only those sentences and a newspaper that [works not] and then I think well . then I thought again . one day again one day then I thought a week then I thought two. two weeks I thought then and odd and you can do everything. I want to story I want and I want very many friends. (want is a modal in Dutch)

 The low diversity and high frequencies of finite verbs imply that the forms are hard to produce. McAllister et al. showed that sentence production with unaccusative verbs was affected in mild to moderate fluent aphasia. So, both experimental studies demonstrate that complex verb (forms) hamper speech production in fluent aphasia. Sentences with unaccusative verbs are grammatically complex, because the subject of the sentence has a theme role, a role that is usually assigned to the direct object.

 In an experiment, semantic paraphasias are produced, while in spontaneous speech, the diversity of finite lexical verbs diminishes and the frequency of these forms is high

 However, the way the aphasic subgroups deal with the problem is different: agrammatic speakers make predominantly grammatical errors on a test and produce simple verb forms in spontaneous speech, whereas fluent aphasic speakers make lexical errors on a test and show little variety and a high frequency of their finite verbs in spontaneous speech.

 Finite verb forms are complex because the computations for tense and agreement have to be performed. In fluent aphasia, this grammatical complexity interferes with lexical retrieval.

 What speaks against the claim that the problems with finite verbs are due to a pure word retrieval problem is that the fluent aphasic patients in the present study do not have a verb retrieval problem in spontaneous speech: both the number and the diversity of non-finite verbs is normal.

 A pure grammatical impairment causing the reduced diversity is not very likely either: if this were the case, the proportion of finite verbs would have been lower, as in agrammatic speech These fluent aphasic speakers are capable of producing a normal variety of lexical verbs as long as these are not finite.

 They are also able to produce finite verbs. However, producing lexical finite verbs raises a problem, which is reflected by a reduced variety and a high frequency of these items.

Adaptation theory
Paragrammatic speakers not able to adapt to their syntactic impairment