Recovery of Grammatical Morphology in Agrammatic Aphasia

1
Results



Complementizers
All grammatical morphemes

Verb Inflection
2
Verb Inflection Test
Narrative Production
MLU 3.9
5.8 gram sentences 26
40.5 nounverb ratio 2.3
1.8 correct morphemes 17
56
3
Generalization Patterns Agreement to Tense
YES (80) NO (n 20)
S3
S2
S7
S12
S4
4
Generalization Patterns Tense to agreement
YES (n 1 of 2) NO (n 1 of 2)
S10
S11
5
NO (n 0)
Generalization Patterns Complementizers to
Tense/Agreement
S4
S8
S9
6
NO (90)
Generalization Patterns Tense/agreement to
Complementizers
S7
YES (n 1)
S12
S2
S10
S11
7

Summary Generalization from C to
I Complementizers to Tense/Agreement Predicted
by the TPH but not CATE
CP
TP
wh
AgrP
that
NegP
ed
AspP
s
VP
V
ing
XP
0 Participants
8

Summary Generalization from I to
C Tense/Agreement to Complementizers Not
predicted by the TPH or CATE
CP
TP
wh
AgrP
that
NegP
ed
AspP
s
VP
V
ing
XP
1/8 Participants
9

Summary Generalization T to ARG Predicted by the
TPH Complexity?
CP
TP
wh
AgrP
that
NegP
ed
AspP
s
VP
V
ing
XP
1/2 Participants
10

Summary Generalization AGR to T Not predicted by
the TPH Complexity?
CP
TP
wh
AgrP
that
NegP
ed
AspP
s
VP
V
ing
XP
6/8 Participants
11

Summary Generalization T to AGR/AGR to
T Predicted by CATE
CP
TP
wh
AgrP
that
NegP
ed
AspP
s
VP
V
ing
XP
7/10 Participants
Supports (sort of) Bobalijk Thrainsson (1989)
Unsplit IP hypothesis agreement and tense under I
12
fMRI Tasks

• ?Verb production
• ?ER design
• Three conditions (each with 40 items)
• Nowadays verb s
• Yesterday verb ed
• Say verb
• 50 null events
• ?Pre-training of verb (base form)

13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
Successful learning and generalization
(AGRT)-Stem
Post Tx.
Pre Tx.
17
Successful learning and generalization
(AGRT)-Stem
Pre Tx.
Post Tx.
18
Successful learning and generalization
(AGRT)-Stem
Post Tx.
19
Negligible learning and generalization
Pre
Pre Tx.
Post
Post Tx.
Pre-post overlap
(AGRT)-Stem
20
What have we learned?

• Generalization follows path of linguistic
  knowledge
• Generalization within but not across CP
  structures
• Wh-movement to NP-movement no cross
  movement-type generalization
• Generalization from C to I and I to C little
  or no C to I or I to C generalization
• Support theories of syntax that make distinctions
  between CP and IP structures (Chomsky 1986
  Cinque 1999)
• Movement structures do not comprise a syntactic
  category
• Functional category members form subcategories
• Members licensed by CP are distinct from those
  licensed by IP
• Suggest that, for English, IP may be unsplit
  (Bobalijk Thrainsson, 1989)
• Improved access to morphosyntax reflected in
  recruitment of neural structures
• Emerging patterns differ for successful and
  unsuccessful learning and generalization

21
Thanks to
Patients and their family members Michael
Walsh-Dickey Borna Bornakdapour Darren
Gitelman Steve Fix Marsel Mesulam Lisa
Milman Todd Parrish Dirk den Ouden Lewis
Shapiro Janet OConnor David Swinney Resear
ch Assistants Diane Arcuri Kyla Garabaldi
Janet Choy Jiyeon Lee
Supported by the NIH Grants R01DC01948-13
R01DC007213-01 R21DC06423-02
22
Thank you