Download Willems_2011_Neuropsychologia_suppl_material

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Participant
JCB
SA
PR
ST
Age (years post-onset)
64 (14)
64 (17)
65 (9)
73 (9)
Lexical-semantic assessments
ADA spoken word picture matching
58/66
60/66
61/66
55/66
60/66
62/66
66/66
61/66
80/160
123/160 121/160 94/160
(chance = 16.5)
ADA written word picture matching
(chance = 16.5)
ADA spoken synonym matching (chance =
80)
ADA written synonym matching (chance =
137/160 121/160 145/160 101/160
80)
PALPA 54 spoken picture naming
8/60
0/60
0/60
0/60
PALPA 54 written picture naming
37/60
24/60
2/60
0/60
Syntactic assessments
Comprehension of spoken reversible
32/100
49/100
38/100
55/100
48/100
42/100
49/100
62/100
19/40
26/40
21/40
28/40
sentences (chance = 50)
Comprehension of written reversible
sentences (chance = 50)
Written grammaticality judgements
(chance = 20)
Verbal working memory assessment
PALPA 13 digit span (recognition)
2 items
3 items
4 items
3 items
1
Supplementary Table S1. Results of standardised language tests for each aphasic
participant. Tests are taken from the Action for Dysphasic Adults (ADA) Auditory
Comprehension Battery [1] and the Psycholinguistic Assessments of Language
Processing in Aphasia (PALPA) [2] or were devised for the purposes of this study.
Spoken and written word comprehension was assessed by word-picture matching and
synonym judgment tests, with the latter permitting evaluation of lower-imageability
non-picturable words. Patients showed greater impairment in comprehension of low
imageability words than high (errors on spoken high imageability pairs: JCB 13; SA 8:
PR 7; ST 18; errors on written high imageability pairs: JCB 0; SA 2: PR 0; ST 13; errors
on spoken low imageability pairs: JCB 32; SA 9: PR 14; ST 17; errors on written low
imageability pairs: JCB 14; SA 19: PR 8; ST 15); Word-picture matching tests required a
stimulus word (spoken or written) to be matched to a corresponding picture in the
presence of visual, phonological/orthographic, and semantic distracters. Synonym
judgment tests involved decisions as to whether two words (spoken or written) had
similar meanings. Lexical retrieval was assessed through spoken and written picturenaming tests. Grammatical processing was evaluated by comprehension of reversible
spoken and written sentences, and a written grammaticality judgment test. The
reversible sentence comprehension tests required matching of a spoken or written
sentence to the corresponding pictured event in the presence of a distracter picture
showing the reversed roles of the protagonists (e.g., ‘‘The man killed the lion’’ and ‘‘The
lion killed the man’’). The stimulus sentences included equal numbers of active and
passive sentences to prevent use of an order-of-mention strategy in decoding the
sentence (e.g., assuming that the first-mentioned noun is the subject of the sentence).
Patients showed impairments on both active and passive sentence comprehension
(spoken sentence comprehension (active vs. passive correct) JCB 20 vs. 12, SA 32 vs. 17,
2
PR 24 vs. 14, ST 31 vs. 24; written sentence comprehension (active vs. passive correct)
JCB 34 vs. 14, SA 33 vs. 9 , PR 23 vs. 26, ST 33 vs. 29). The grammaticality judgments test
required the patient to decide whether a written sentence was grammatical. This task
was not performed in the phonological modality to avoid prosodic cues influencing
grammaticality judgments. The task examined sensitivity to flagrant violations of
English syntactic conventions e.g., ‘sentences’ with no verbs, sentences with postpositions vs. prepositions. Only simple sentence structures were evaluated i.e., a single
main clause. Each patient’s short-term phonological memory capacity was determined
by a digit span test. Span was established in a recognition paradigm in order to
eliminate speech production problems from the task. The patient decided whether two
auditorily presented strings of numbers were the same or different.
3
L
L
L
Supplementary Figure S1. Axial slices the anatomical scans of three of the four
patients (JCB, PR and SA). Note the extensive damage in the left hemisphere,
encompassing the whole peri-sylvian language network. No images are available for
patient PZ who cannot be scanned because of contra-indications for imaging.
4
References
1.
2.
Franklin, S., J.E. Turner, and A.W. Ellis, The ADA Auditory Comprehension Battery.
1992, York, UK: University of York.
Kay, J., R. Lesser, and M. Coltheart, Psycholinguistic Assessment of Language
Processing in Aphasia 1992, Hove, UK: Psychology Press.
5