Docstoc

Recovery of Grammatical Morphology in Agrammatic Aphasia

Document Sample
Recovery of Grammatical Morphology in Agrammatic Aphasia Powered By Docstoc
					Treatment of Grammatical Deficits in
             Aphasia:
   Language and Brain Recovery
                      Cynthia K. Thompson
         Aphasia and Neurolinguistics Research Laboratory
Departments of Communication Sciences and Disorders, and Neurology
       Cognitive Neurology and Alzherimer’s Disease Center
                     Northwestern University
  Supported by the NIH (R01-DC01948-14; R01DC007213-02; R21DC06423-02)
Aims
   Examine patterns of morphosyntactic learning and
    generalization in agrammatic aphasia
     Experimental manipulations motivated by mutually supportive
      theory from both formal linguistic (language representation)
      theory and language processing accounts
     Does language recover in linguistically principled ways?

   Investigate the processing mechanisms that support
    recovery.
       When recovery occurs, are normal language processing
        routines engaged?
   Examine the neural correlates of recovery
       Are there signature patterns associated with successful and
        unsuccessful learning and generalization?
Broca’s Aphasia with Agrammatism
         Primary Deficits
               Production (and comprehension)
                of complex syntax, e.g.,
                noncanonical sentences (Caplan &
                Hildebrandt, 1998; Caramazza &
                Zurif, 1976; Saffran et al., 1980;
                Grodzinsky 2000 and others)
               Production of grammatical
                morphology (Arabatzi & Edwards,
                2000, 2002; Bastiannse 1995;
                Bastiannse & Thompson, 2002;
                Friedman & Grodzinsky, 1997;
                Hagiwara 1995; Lee 2003; Miceli &
                Caramazza, 1988; Thompson et
                al., 2002)
      Syntactic Structures
Movement or filler-gap structures
   Movement: linguistic construct; Principles
    and Parameters/The Minimalist Program
   Filler-gap: psycholinguistic process

Wh-movement
   Object relatives
   Object clefts

   Object extracted wh-questions

NP-movement
   Subject raising
   Passives
            CP
          ty
         DP      C'
                                     Wh-movement
       (who) ty                      structure
               C      TP
             (did) ty
                    DP       TÕ
                  (thief) ty
                          T     VP
                               ty
                          t DP      V'
                                  ty
                              t V      DP
                               (chase)
                                       t




Whoi did the thief chase(ti) [trace/gap]
                 TP
                                            N-movement
            ty                              structure
           DP        T'
         (artist) ty
                   T     VP
                 (was) ty
                        VÕ PP
                    ty    (by the artist)
                     V DP
                 (chase d)
                             t




The artisti was chased (ti) by the thief
                                CP
                                ty
                                         C'
Spec, CP                            ty
Wh-movement landing site
(object relatives, object wh-
                                    Co          TP
questions)                                    ty
         Spec, TP
                                                     T'
         NP-movement landing site
                                                ty
         (subject raising, passi ves)           To
            Cross -modal lexical priming (CMLP)
(Zurif et al., 1993; Swinney et al., 1996; Thompson et al., 1999;
                  but see Blumstein et al., 1998)

  Pete saw the doctor i who the supervisor 1 called ti 2 to . . . .


  At points 1 and 2 Ss see:
   related word
   unrelated word
   pseudoword
                                         patient

Normals: RT for related; 2 < 1

Broca’s: RT for related; 2 = 1
         Impaired gap-filling
 Eyetracking: object wh-questions
 Dickey & Thompson, in press


This is a story about a boy and a girl.
One day they were playing at school.
The girl was pretty so the boy kissed the girl.
They were both embarrassed after the kiss.




Did the boy kiss the girl at school that day?

Who did the boy kiss [trace] that day at school?
                                                                                         Object wh-questions
                                                                        0.5


                                                                                     Control participants
                                                                                                                                                 Looks to girl
                                              Proportion of fixations
                                                                        0.4




                                                                        0.3      *            *          *
                                                                        0.2
                                                                                                                                                 Looks to boy
                                                                        0.1




                                                                         0


                                                                                 V           Gap       Location

Who did the boy kiss                                                                      [gap]        at school

                                                                              Aphasic participants                                      Aphasic participants
                                  4 0. 0 0%                                   Correct Responses                                        Incorrect Responses
                                                                                                                        0.5
        Proportion of fixations




                                                                                     *
                                                  *
                                  3 5. 0 0%




                                                                                                                        0.4
                                                                                                                                             Interaction
                                  3 0. 0 0%




                                  2 5. 0 0%




                                  2 0. 0 0%




                                  1 5. 0 0%
                                                                                                                        0.3

                                                                                                                        0.2
                                                                                                                                                  *
                                  1 0. 0 0%




                                                                                                                        0.1
                                  5 . 0 0 %




                                  0 . 0 0 %




                                                                                                                         0
                                              V                                   Gap       Location                            V          Gap        Location

. . . . . . . kiss                                                             [gap]       at school              . . . . . . . kiss    [gap]         at school
             Results                                                                                                                                            Complex Sentences
                                                                                                                                                               Wh- and NP-Movement



  Wh- and NP-Movement Structures                                                                                                      100           Object Cleft Training




                                                                                                             Wh-Movement Structures
                                                                                                                                       80


                                                                                                                                       60                                                                        who-questions
                                                                                                                                                                                                                 object clefts
                          CP                                                                                                           40


                                                                                                                                       20

             SPEC
                                     C'                                                                                                 0
                                                                                                                                            1   5     9   13   17   21      25   29   33   37   41   45     49




                        COMP                   IP                                                                                     100
                                                                                                                                                                                 Subject Raising Training

   Landing si te for                                                                                                                   80




                                                                                                            NP-Movement Structure
   wh-elements in
   wh-movement;                                                                                                                        60
   a non-argument                SPEC                        I'                                                                                                                                                  passives
   posi tion -- never                                                                                                                                                                                            subject raising
                                                                                                                                       40
   assi gned a
   theta-role at d-
   structure                                                                                                                           20
                                               INFL                    VP
           Landing si te for
                                                                                                                                        0
           auxilary verb in
                                                                                                                                            1   5     9   13   17    21     25   29   33   37   41   45     49
           certain question
           types; d-structure                                                                                                                                       Probe Sessions
           posi tion of                                     V                     CP
           complementizer head
                                      d-structure location
           (e.g. that, for)
                                      of verb tense Main verb

                   Posi tion of the subject,
                                                      llocation
                                                                     SPEC                    C'
                                                                                                        Generalization Patterns
                   landing si te for NP-
                   movement; an argument
                   posi tion -- may be
                   assi gned a theta-role at
                                                      Landing si te for wh-
                                                      movement in embedded
                                                      questions
                                                                                                                                     wh-movement ------> wh-movement
                                                                                  COMP             IP

                                                                                                                                      NP-movement ------> NP-movement
                   d-structure
                                                                  d-structure posi tion of
                                                                  complementizer head (e.g.             
                                                                  that, for) in embedded clauses


                                                                                                                                     wh-movement --//--> NP-movement
                                                                                                                                     NP-movement --//--> wh-movement
(Ballard & Thompson 1999; Jacobs & Thompson 2000; Thompson & Shapiro, l994;
U of I (2006) Thompson et al. 1997; Thompson et al., 2003).
Thompson et al., l993;
              Generalization
Functional relation between trained and
 untrained language structures -->
Generation of new forms -->
Development of representations and
 computations that are relevant to new
 responses -->
Response generalization
       Wh-movement generalization
                Trained      Untrained             Untrained
Subject         Wh-Structure Wh-Structure          NP-Structure
DLB             Object clefts => Who-questions -/-> Passives
MD*             Object clefts => Who-questions -/-> Passives
P1µ             Object clefts => Who-questions -/-> Passives
P2µ             Object clefts => Who-questions -/-> Passives
P3µ             Object clefts => Who-questions -/-> Passives
MRW             Object clefts => Who-questions -/-> Passives
LDW             Object clefts => Who-questions -/-> Passives
HH**            Object clefts => Who-questions -/-> Passives
P4µ             Object clefts -/-> Who-questions -/-> Passives
P5µ             Object clefts -/-> Who-questions -/-> Passives
AH**            Who-questions -/-> Object clefts -/-> Passives
JHW             Who-questions -/-> Object clefts -/-> Passives
GKW             Who-questions -/-> Object clefts -/-> Passives
FPB             Who-questions -/-> Object clefts -/-> Passives
KD**            Who-questions -/-> Object clefts -/-> Passives
CHB             Who-questions -/-> Object clefts -/-> Passives
PR*             Who-questions ==> Object clefts -/-> Passives


       Ballard & Thompson 1999; Thompson et al., 1997, 1998
      Prominent Generalization Pattern
                Complexity

The judge saw the senator who the reporter attacked ___.

        It was the senator who the reporter attacked ___.


                        Who did the reporter attack ___.
      TP
  t    y
SPEC     T'
(man) ty
      T     VP
     (saw) ty
                  V'
           t    t y                     Object Relative Structure
                V      DP
                     ty
               t (artist) (j) CP                                     TP
                              ty                                   ty
                        SPEC       TP                           SPEC         T'
                      (who)(i)(j) ty                            (it)      ty
                               SPEC     T'                               T      VP
                               (thief) ty                              (was) ty
                                   (chased ) VP                                     V'
                                             ty                                   t y                               Object Cleft Structure
                                                    V'                            V      DP
                                                  ty                                   ty
                                             t    V      DP                      t (artist) (j) CP
                                                                                                ty
                                                  t      t(i)                             SPEC       TP
                                                                                        (who)(i)(j) ty
                                                                                                 SPEC     T'
                                                                                                 (thief) ty
                                                                                                     (chased ) VP
                                                                                                               ty
                                                                                                                      V'
                                                                                                                    ty
                                                                                                               t    V      DP

                                                                                                                    t      t(i)
Treatment:   least to most complex   vs   most to least complex
              Wh-questions 1st            Object relatives 1st




                       Thompson et al., 2003
                            IP
Object relatives
                                      I’
                    SPEC
                                               VP

                   the man INFL                     V’           Object clefts
                           saw                            NP

                                  t
                                       V                         CP

                                                N
                                           t
                                                                          C’
                                      the artist (j)      SPEC

                                                         who(i)(j)                IP
                                                                 COMP
                                                                                            I’
                                                                        SPEC
                                                                                                     VP

                                                                      the thief INFL                      V’
                                                                               chased

                                                                                        t
                                                                                                               NP
                                                                                                 V

                                                                                                               t(i)
Object extracted who/what questions                                                              t
                             100
                              90




Proportion of Participants
                              80
                              70
                              60
                              50
                              40
                              30
                              20
                              10
                               0

                                   Sim ple to
                                   C om ple x   C om ple x to
                                                  Sim ple
      Processing mechanisms of recovery
  Anomaly detection: Participants listened to sentences, pressed a
  button when the sentence began to sound odd (cf. Boland, et al.)

      The girl put on a shirt that her mother picked ____ for her.
      The girl put on a shirt that her mother fried ____ for her.

 1                                                     2500
0.8                                                    2000
0.6                                                    1500
                                            'picked'
0.4                                                    1000
                                            'fried'                                              RT, in msecs
0.2                                                     500
 0                                                        0
      Controls    Treated     Untreated                       Controls    Treated Untreated
                 aphasic Ps   aphasic Ps                                 aphasic Ps aphasic Ps

        Percent Rejected                                           Reaction times

                                     (Dickey & Thompson, 2004)
Neural mechanisms of Recovery - fMRI

    Decrease                 Increase                   Decrease
                            QuickTime™ and a
                        TIFF (LZW) decompressor
                     are neede d to see this picture.




Thompson, Bonakdarpour, Fix, Parrish, Gitelman, & Mesulam, in progress
      Syntax Training and Generalization
Generalization --> linguistically related
No generalization --> linguistically unrelated
Complex sentence learning results in
 generalization to simpler sentences iff they are in a
 subset relation to one another
  •   ESL (Eckman 1988)
  •   Phonological development (Gierut in press)
  •   Lexical semantics (Kiran & Thompson)
  •   Computer simulations (Plaut 1986; Eckman 1991)
Some evidence that treatment results in “normal”
 processing routines
Treatment changes can be mapped onto the brain
   Greater or lesser activation
Generalization Patterns in Grammatical
 Morphology: Functional Categories
                         CP



           SPE C
                                     C'
                                                                                                               Morphology licensed by CP
                       COMP                       IP                                                              Complementizers
  La ndi ng site for
  wh -el eme nts i n
  wh -movemen t;
                                  SPE C                        I'
                                                                                                                    if, whether, that
  a no n-argumen t


                                                                                                               Morphology licensed by IP
  po si ti on --
  ne ver assi gne d
  a theta-rol e a t
  d-structu re
                                                  INFL                     VP
         La ndi ng site for
         au xi la ry verb in
         ce rtai n qu esti on
                                                                                                                  Tense
         type s; d-stru cture

                                                                                                                     regular past tense [-ed]
         po si ti on o f                                      V                        CP
         co mpl emen ti ze r
         he ad (e.g. that, for)         d-structu re l ocatio n
                                        of verb tense Mai n ve rb
                                                          l lo cati on
                 Po si ti on o f the su bj ect,
                 l and in g si te for NP-
                                                                            SPE C                    C'
                                                                                                                  Agreement
                 movemen t; an argu me nt                  La ndi ng site for wh -

                                                                                                                     third person singular [-s]
                 po si ti on -- ma y be                    movemen t i n embe dded
                 assi gn ed a theta-rol e a t              qu esti ons
                 d-structu re                                                           COMP              IP
                                                                       d-structu re po si tio n of
                                                                       co mpl emen ti ze r he ad (e.g.
                                                                       that, for) i n embe dde d
                                                                       cl au se s




                                          (Benedet, et al. 1998; De Villiers 1978; Friedmann 1998,
                                            Milman, et al. 2004; Rochon et al., 2001 and others)
  CP-level more impaired than IP-level
    (Japanese Agrammatic Speakers (Hagiwara 1995))

                           CP



More impaired   that, if          IP



     Less impaired     -ed, -s                VP
                Tree Pruning Hypothesis
C, T, or Agr may be underspecified; underspecified nodes cannot project higher
                        Hebrew Agrammatic Speakers




      Friedmann and colleagues (1997, 1998, 2001, 2002)
       after Jean-Yves Pollock’s (1989) split IP hypothesis
                      Complexity Account
C’ unrelated to I’ except for status as a functional category member


                   Morphology licensed by CP
                       Complementizers




    Morphology licensed by IP          Morphology licensed by IP
              Tense                           Agreement
                        Questions
   Do functional category members recover as
    a class?
       Generalization from C’ to I’ and/or I’ to C’?
   Or is recovery constrained the linguistic
    status of category members?
       Generalization from T to ARG and/or AGR to T?
          Members licensed by the same node
          Presumably rely on similar language processing
           routines?
                              Generalization Patterns
       CP
                              Predicted by Tree Pruning
                 TP
wh
     that             AgrP
            ed               NegP
                                    AspP
                 s
                                           VP
                                                V’
                              ing
                                                     XP
                                      Generalization Patterns
         CP
                                      Predicted by Complexity
                   TP
  wh
       that             AgrP
              ed               NegP
                                       AspP
                    s
                                              VP
                                                   V’
                                ing
                                                        XP



• Agreement below tense phrase (Pollock 1989)
• Agreement above tense phrase (Chomsky, 1993; Ouhalla, 1990)
• Unsplit IP; agreement and tense under I (Bobaljik & Thrainsson, 1998)
                 Cross-Over Design
            P                         P

Group 1
            r
            e
                                      o
                                      s      Embedded multiple baseline
                   Treatment
                                             • across participants
                                      t
            T
            e   (max 15 session per   T

                                             • across behaviors (n=3)
            s        structure)       e
            t                         s
                                      t
            I
            P                         P                        P
            r                         r                        o
Group 2     e                         e                        s

            T                         T
                                           Treatment           t

            e                         e    (max 15 session     T
            s                         s     per structure)     e
            t                         t                        s
                                                               t
            I                         II

    Production Probes         Production Probes       Production Probes
        Narratives               Narratives               Narratives
    Verb Inflection Test                              Verb Inflection Test
      Grammaticality                                    Grammaticality
       Judgement                                         Judgement
           fMRI                                              fMRI
     Production Probe
  Verb Inflection Elicitation

   Yesterday
Tense [-ed] (n=20)




   Nowaday
   s
Agreement [-s] (n=20)
                        weigh
        Production Probe Complementizers
            Sentence production priming task (n = 20)


know                              +
see
wonde             kno                paint
r                 w
      Prime: They know if/that/whether the woman is painting the child.
care
think

                                   +

                      see                paint
         Target: They see if/that/whether the woman is painting the child.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:13
posted:11/13/2010
language:English
pages:31