45
Linguistic approaches to aphasia Stephen Wilson UCLA April 22, 2002

Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Linguistic approaches to aphasia

Stephen WilsonUCLA

April 22, 2002

Page 2: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Outline

• Syntactic theoretic accounts of agrammatism(Grodzinsky, etc.)

• The ‘declarative/procedural model’ (Ullman, Pinker, etc.)

Page 3: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Trace Deletion Hypothesis

• First proposed by Grodzinsky (1984).• Relatively minor developments in the theory

over the last 15 years.• A recent version in Grodzinsky (2000).• Claims the sole function of Broca’s area is to

perform computations involving traces.

Page 4: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Ideas about localization

• 19th century “connectionists”: Broca’s area for speaking, Wernicke’s area for comprehension.

• 1970s: Broca’s area for syntax (c.f. Caramazza& Zurif, 1976), Wernicke’s area for semantics, lexicon.

• Mid 1980s on: Increasingly circumscribed role for Broca’s area (only some syntax).

Page 5: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

TDH: Overview

Broca’s area/Broca’s aphasia involve:• Trace deletion in comprehension• “Tree pruning” in production• “Clear evidence for near-normal

performance in most other domains of syntax” (Grodz, 2000, p. 4), also no lexical impairment

Page 6: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

TDH: Differentiating constructions

Above chanceThe girl pushed the boy.The girl who pushed the boy

was tall.Show me the girl who pushed

the boy.It was the girl who pushed the

boy.The boy was interested in the

girl.The woman was uninspired by

the man.

ChanceThe boy was pushed by the

girl.The boy who the girl pushed

was tall.Show me the boy who the girl

pushed.It was the boy who the girl

pushed.The woman was unmasked by

the man.

Page 7: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

TDH mechanisms 1• Traces, according to GB (Chomsky, 1981) and

related theories, are silent syntactic entities found in original and intermediate positions of moved elements.

• They reflect the pretheoretical distinction between “basic/canonical” and “derived/ noncanonical” structures, e.g.– The cat chased the mouse.– The mouse was chased by the cat.

• All linguistic theories have some way of capturing the fact that this pair of sentences are related.

Page 8: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

TDH mechanisms 2

• Thematic roles (aka θ-roles, etc.) encode the relation of arguments with the verb, e.g.– Agent, Theme, Experiencer, Goal, etc.

• In GB, they are assigned by the verb to its arguments in local configurations (Baker, 1987).

• In the case of sentences involving movement, θ-roles are assigned to the trace and transmitted to the argument.

• According to the TDH, when traces are deleted, aphasic patients cannot assign θ-roles to moved constituents.

Page 9: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

TDH mechanisms 3

• When an argument fails to receive a θ-roles, it is assigned Agent by a default, non-linguistic strategy.

• The distinction between actives and passives:The boy pushed the girl.

Normal: Agent ThemeAgramm: AGENT Theme

The girl was pushed by the boy.Normal: Theme AgentAgramm: AGENT Agent

• Competition between the two agents results in chance performance.

• In old versions, the subject of actives was assigned its θ-role by the verb. But since the VP-interal subject hypothesis (Kitagawa, 1986) all subjects are thought to move, so its θ-role is now assigned by the strategy.

Page 10: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

TDH structures “accounted for”

The boy who pushed the girl was tall. (above chance)The boy who the girl pushed was tall. (chance)The girl was admired by the boy. (below chance)Taro-ga Hanako-o nagutta. (above chance)Hanako-o Taro-ga nagutta. (chance)Taro-ga Hanako-ni nagu-rare-ta. (chance)Okaasan-ga musuko-ni kaze-o hik-are-ta. (above chance)Zhuei gou de mau hen da. (chance)

chase dog that cat very bigMau Zhuei de gou hen xiao. (above chance)

cat chase that dog very small

Page 11: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Problems with the TDH

• Most people agree that there are serious problems with the TDH.

• However, a surprisingly large number of people seem to think that it’s on the right track – they concern themselves with proposing modified versions, e.g. Hickok &Avrutin (1995), Mauner et al. (1993).

Page 12: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Data it is supposed to account for but doesn’t really

• Passives of exp-theme psychological verbs:The girl was admired by the boy. (below chance)

• Object relatives:The boy who the girl pushed was tall.

Page 13: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Other data it can’t account for

• Above-chance comprehension of truncated passives, e.g. The mouse was chased (Beretta, et al., 1999)

• If patients are given a scenario with two agents, they don’t choose it (Beretta & Munn, 1998).

• Good performance on passives in Dutch (Friederici &Graetz, 1987).

• Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980).

• Patients do better with passives on theme-exp psych verbs, e.g. The woman frightened the man (Pinango, 1998).

Page 14: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

The default strategy

• Where does it come from?• Is it just chance that it reflects canonical word

order in English?• Some of the empirical failures of the TDH

are caused by the default strategy.• Many linguists otherwise sympathetic to

Grodzinsky’s general approach take issue with the default strategy.

Page 15: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

A TDH offshoot

• Double dependency hypothesis (DDH) proposed byMauner et al., 1993, defended by Beretta & Munn (1998), Beretta et al. (1999), etc.

• Agrammatic patients fail when there is more than one dependency between a moved element and its trace.

• No default strategy.• Captures roughly the same data, also fails on

roughly the same data, e.g. truncated passives.

Page 16: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Agrammatism as a syndrome

• Assumption of TDH & many related theories: allagrammatics pattern essentially alike.

• Should Broca’s aphasia be functionally or anatomically defined?

• Grodzinsky, et al. (1999) carry out a meta-analysis attempting to show that all Broca’s aphasics fit the “agrammatic profile”.

• This claim has been disputed, or, I think it is fair to say, falsified.

Page 17: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Berndt, Caramazza, etc. onagrammatism as a syndrome

• Berndt et al. (1996) found that only one third ofBroca’s aphasics display the “agrammatic profile”

• They criticize Grodzinsky et al. (1999) for dubious selection procedures.

• Berndt points out that Goodglass (1993) found 5 of 7 Broca’s aphasics performed better than chance on passives, with 3 of them at 100%

• It’s hard to find agrammatics: Beretta & Munn (1998) screened 15 to find 6; Hagiwara (1993) screened 70 to find 10.

Page 18: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Caplan on agrammatism

• He has tested hundreds if not thousands of patients, finding that deficits are not consistent, severity is the only real determinant of co-occurrence.

• Caplan et al. (1996) report a patient with a lesion confined toBroca’s area and 100% perfect performance on passives.

• One of the more reasonable comments from a neurolinguistwith Chomskian leanings:– “Perhaps the lesson aphasia is teaching us is that brain damage does

not selectively disrupt the categories postulated in Chomsky’s model of syntactic structure (which, of course, is not to say that such categories do not exist).” (Caplan, 2000, p. 27)

Page 19: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Broca’s area and Broca’s aphasia• The link is simply assumed without argument by Grodzinsky and

many others.• Dronkers et al. (1992) find only about half of patients with lesions

to Broca’s area have persistent Broca’s aphasia.• Many patients also have WAB-determined Broca’s aphasia without

lesions to Broca’s area.• Broca’s area has numerous non-linguistic functions (episodic

memory encoding, gesture recognition, mirror drawing, object manipulation, etc.), mostly determined with fMRI.

• Broca’s aphasics have numerous other linguistic deficits, e.g. lexicon (Goodglass, 1993), phonological processing, maybe even regular morphology (Ullman et al, 1997) and non-linguistic deficits, e.g. failure to comprehend environmental sounds (Saygin, et al., 2001).

Page 20: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Does poor performance on passives imply lack of traces?

• Dick et al. (2001) had college students interpret actives, passives, subject clefts and object clefts under acoustically degraded conditions.

• They were able to effectively simulate the “agrammatic profile”.

Page 21: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Dick et al. (2001) results

• Presumably we don’t want to say that low-pass filtering and compression induces a transient trace deletion mode…

Page 22: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Explaining the data

Competition model (CM), e.g. Bates et al. (1991)• Various cues to meaning (e.g. word order,

agreement, case marking).• Cues contribute to different extents in

different languages.• Some types of cue are more prone to suffer in

aphasia (and other adverse conditions).

Page 23: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Partitioning the data with the CM

Relatively impaired

The boy was pushed by the girl.

It was the boy who was pushed by the girl.

The boy who the girl pushed was tall.

Zhuei gou de mau hen da.chase dog that cat very big

Relatively spared

The girl pushed the boy.

It was the girl who pushed the boy.

The girl who pushed the boy was tall.

Mau Zhuei de gou hen xiao.cat chase that dog very small

Page 24: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Grodzinsky on such accounts

• Discussed on pp. 55, 57.• Three structures which are supposed to create

problems:– The woman was pushed by the man vs. The woman was

loved by the man. (frequency issues?)– Which elephant did the giraffe sniff? vs. Who did the

giraffe sniff? (Hickok & Avrutin, 1995, but Thompson et al. (1999) failed to replicate).

– The woman was pushed by the man vs. Every woman was pushed by a man. (very dubious finding)

Page 25: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Quantifiers (of mice and men)• Empirical claim: Sentences like “Every mouse is paid by the man”

are comprehended better than “The mouse is paid by the man”• Theoretical explanation: non-referential phrases are not subject to

the strategy.• Serious empirical problems with yes/no task used by Balogh &

Grodzinsky (1999):• Characters: three men and three mice• Story: The mouse has just won some money from playing bingo.

The mouse wants to pay off some of his debts. He gives some money to the first man, then the second man, and finally to the third man.

• Sentence to judge: Every mouse is paid by the man.• Problem: semantics of “every” is enough to figure out the roles.• Stats were also “creative”: actives at 89%, quantified passives at

85%, normal passives at 73% (said to be chance).

Page 26: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

The CM is adequate

• A model like the CM seems to be able to account for most if not all of the reliable data on tendencies in agrammatism, without having to make recourse to theory-specific devices such as traces.

• Kay (2000): “the data presented in Grodzinsky’starget article suggest that Broca’s aphasics may rely on a small number of parsing strategies based on the most frequent construction types in their language.” (p. 38)

Page 27: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

CM predictions

• The CM account differs from practically every restrictive model in that it asserts a role for the “weakness” of morphological cues.

• The TDH, DDH, etc. all assume the problem is solely with movement.

• Therefore only the CM predicts that modulating the strength of the morphological cues should modulate performance, e.g.:– The cat is being chased vs. The cat was chased.– The cat is beaten vs. The cat is thrashed.

Page 28: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Tree pruning

• Production is different to comprehension, instead of traces being deleted, the top of the tree is “pruned”.

• Supposed to account for problems with– questions in English– Tns but not Agr in a Hebrew-speaking patient

and other patients speaking other languages.

Page 29: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Some problems with tree pruning

• Bickerton (2000, p. 25): “flies in the face of countless examples of agrammatic speech in which “everything below the Tense node” is as disorganized as anything above it”

• Predicts that in a V-to-C language like German, patients should produce final, non-agreeing verbs, but this is not the case (Penke, 1998).

• The proposed deficit is so different to the one proposed for comprehension that it is no longer possible to talk about a deficit in central representations.

Page 30: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Grammaticality judgment

• Grodzinsky & Finkel (1998): “If traces are indeed deleted from syntactic representations as a consequence of focal brain damage, no computation of ungrammaticality can be carried out properly in constructions were traces are involved.” (p. 284)

• They argue that both Broca’s and Wernicke’s patients show problems judging sentences involving phrasal movement but not involving other kinds of violations, including violations involving a different kind of movement: head movement.

• Cited approvingly by Chomsky (2000) as evidence for a neural distinction between different kinds of movement.

Page 31: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

G&F (1998) stimuliXP-traces

1. NP movementIt seems likely that John will win.*John seems that it is likely to win.It seems that John is likely to win.John seems likely to win.2. Wh-movement/that-traceWhich woman did David think John saw?*Which woman did David think that saw John?Which woman did David think that John saw?Which woman did David think saw John?3. SuperiorityI don’t know who said what.*I don’t know what who saw.4. Adjunct/complementWhen did John do what?*What did John do when?

No XP-traces

5. Filled gapsWho did John see?*Who did John see Joe?Who saw John?*Who John saw Joe?6. Bad complementsThe children threw the football over the fence.*The children sang the football over the fence.The children sang.*The children threw.7. Place of auxiliaryThey could leave town.*Have they could leave town?Could they leave town?They could have left town. Etc.8. NegationJohn has not left the office.*John did not have left the office.John did not sit.*John sat not.

Page 32: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

G&F (1998) problems

• Stimuli not controlled for anything, many things differ between the sets of sentences.

• In the theory assumed by G&F, practically every sentence involves traces, so it’s hard to explain how the allegedly traceless sentences can be judged.

• Many of the patients were actually above chance on the sentences involving traces, which was obscured by a lack of appropriate statistical tests.

• Substantial variation between subjects.

Page 33: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Work in progress

• We are currently running an experiment to challenge G&F’s claims.

• Normal controls under both regular and perceptually degraded conditions will be used to obtain objective measures of difficulty.

• We have also included “easy” judgments involving trace violations and “hard” judgments involving other violations:– *Susan the car which drives is red.– *What did you buy bananas and?– *Could have they already left?– *Patrick does already what he is told.

Page 34: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Restrictive accounts and the big picture

• If Broca’s area only does traces, where is the rest of syntax?

• N.b. G&F (1998) also claim that the deficit exhibited by Wernicke’s patients is similarly restricted.

• If you overlap the lesions of all the patients in the G&F study, there’s not much left out.

• A problem for all restrictive accounts (e.g. DDH): WHY would random brain damage always cause the same kind of highly specific disruption?

• Inevitable conclusion: syntax is highly distributed.

Page 35: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

The declarative/procedural model

• A huge body of work argues for a neural distinction between grammar and the lexicon.

• We’ll focus on Ullman et al. (1997), Ullman (2001), who have argued that grammar is anterior and tied to the procedural memory system, whereas the lexicon is posterior and tied to the declarative memory system.

• The debate has mostly centered around regular vs. irregular morphology, especially the English past tense.

Page 36: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Some history 1

• Linguists have always found it useful to distinguish regular and irregular morphology, and grammar and lexicon.

• It used to be generally assumed that these involved categorically different psychological mechanisms.

• Rumelhart & McClelland’s (1986) seminal connectionist model of the English past tense challenged this assumption, by learning both regulars and irregulars with the same mechanism, and matching the acquisition data.

Page 37: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Some history 2

• Pinker & Prince (1988) challenged R&M’s model on detailed empirical grounds.

• Connectionists responded with new and improved models (e.g. Plunkett & Marchman, 1993, Joanisse& Seidenberg, 1999).

• A cottage industry developed (Pinker, 1999).• Ullman et al. (1997) made the link to the

procedural/declarative distinction made in the learning and memory literature.

Page 38: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Patient groups and the English past tense

• Ullman et al. (1997) tested five groups of patients on a past tense generation task: Alzheimer's (AD), Parkinson's (PD), Huntington's (HD), anterior aphasics (AA) and posterior aphasics (PA).

• The data are quite interesting although the methods seem sloppy in places, this study needs to be replicated.

Page 39: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Ullman et al. (1997) results 1

• AD: temporal lobe degeneration, better on regulars. (89% vs. 60%)

• PD: basal ganglia degeneration, hypokinesia, better on irregulars. (88% vs. 80%)

• HD: basal ganglia degeneration leading to chorea, over-regularization (digged) and hyper-inflection (lookeded). (no difference in means but different error distribution)

Page 40: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Ullman et al. (1997) results 2

• Anterior aphasics better on irregulars (52% vs. 20%) though only one had a lesion restricted to the frontal lobe/BG.

• Posterior aphasics better on regulars (83% vs. 71%) though again only one had a lesion restricted to Wernicke’s area.

Page 41: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Issues with Ullman et al. (1997)

• Most regular/irregular differences were small and need to be replicated.

• AD failure on irregulars has been replicated but may reflect perseveration.

• Patterson (2002) claims that when phonological complexity is controlled, the irregular advantage for anterior aphasics disappears.

Page 42: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Neuroimaging

Hemodynamic responses to syntactic and lexical/semantic violations detected by fMRI.Hemodynamic responses averaged over 14 subjects in a functional magnetic resonance imaging (fMRI) study. a | Syntactic violations elicited greater blood-oxygen-level-dependent (BOLD) activations than semantic violations, primarily in bilateral superior frontal gyrus, corresponding to Brodmann areas (BA) 6 and 8, including the supplementary motor area. Additional activations were observed in the leftinsula and right anterior superior temporal sulcus. b | Semantic anomalies yielded a different pattern of activation, with substantially more temporal and temporoparietalinvolvement than syntactic anomalies, in the angular gyribilaterally (BA 39), the right middle temporal gyrus (BA 21), and left hippocampus and parahippocampal gyrus. Additional activations were found in dorsolateral prefrontal cortex and medial foci.

Page 43: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

What if it’s true? 1

• Ullman (2001) vacillates between reasonable but weak statements such as “the model does not assume that these two memory systems are the only systems that underlie lexicon and grammar” (p. 718) and unsubstantiated claims such as “the frontal cortex and basal ganglia are ‘domain general’ … but contain parallel, domain-specific circuits.” (p. 718)

• Some attempts have been made by connectionists to model dissociations within a single network (e.g.Joanisse & Seidenberg, 1999).

Page 44: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

What if it’s true? 2• A differential reliance on frontal/BG vs. posterior

structures is not incompatible with models which claim there is no clear-cut distinction between grammar and lexicon, although the details remain to be specified.

• Continuity between grammar and lexicon is argued for by many researchers of all stripes:– Neurolinguists (Bates & Goodman, 1999)– Functionally-oriented linguists (Goldberg, 1995, Croft,

2000)– Formally-oriented linguists (Jackendoff, 1997, Albright &

Hayes, 2002)

Page 45: Stephen Wilson UCLA · 2005-10-04 · • Poor performance on locatives, e.g. The dog is behind the cat (Schwartz, et al., 1980). • Patients do better with passives on theme-exp

Conclusions• Most syntactic approaches to aphasia go well

beyond the data in terms of the claims they make and the mechanisms they invoke.

• It is still interesting to look at the structures on which patients tend or tend not to fail, and to try to understand them in terms of a well-articulated theory.

• There is some evidence for differential involvement of different parts of the brain in different aspects of linguistic processing, but nothing incompatible with the linguistic and psycholinguistic evidence for a continuum between grammar and lexicon.