Error patterns of Greek aphasic speakers in sentence completion and grammaticality judgment.

Presented at the 48th Annual Meeting of the Academy of Aphasia. Athens, Greece, October 24–26 2010.

Spyridoula Cheimariou,1 Spyridoula Varlokosta,2 Alexandra Economou,3 Maria Kakavoulia,4 & Athanassios Protopapas1,5
1 Program in Basic and Applied Cognitive Science, University of Athens
2 Department of Linguistics, University of Athens
3 Department of Psychology, University of Athens
4 Department of Communication, Media, and Culture, Panteion University
5 Institute for Language and Speech Processing / R.C. “Athena”

Introduction: Varlokosta, Valeonti, Kakavoulia, Lazaridou, Economou & Protopapas (2006) found selective deficits in verb inflection by Greek aphasic speakers. However, materials in their study were not balanced across conditions, confounding functional category with putative processing load. A recent replication with balanced materials suggests that deficits are evenly distributed across functional categories and not selective. Here we extend the analysis of the study with balanced materials, examining distinct categories of errors.
Participants: Participants included 10 individuals clinically diagnosed with aphasia and a control group of 10 non-impaired individuals matched in sex, age and education.
Materials and procedure: A sentence completion and a grammaticality judgement task were administered, each including three conditions: subject-verb agreement, tense, and aspect. (See further details about the design and interaction in Varlokosta et al., 2006). Ten verbs were used to construct a total of 40 sentences in each condition. The new materials were balanced across functional categories in phrase length (number of characters: M=48, SD=6.3; words: M=8.6, SD=1.1) and number of words preceding the verb (M=4.9, SD=0.6).
Results: As reported elsewhere, aphasic speakers made more errors than controls in each functional category condition, in both sentence completion and grammaticality judgement. There was no difference between categories in sentence completion for the aphasic group. For both groups there were fewer errors in agreement than tense or aspect in grammaticality judgment. Sentence completion responses were classified as form or lexical errors, i.e., incorrect inflectional morpheme or incorrect lexeme, respectively. (Both types are simultaneously possible on each item). Both groups made more form errors than lexical errors (both p<.0005, by GLM mixed-effects FIML analysis). Patients made more lexical and more form errors than controls (both p<.0005). The group difference did not interact with condition for lexical errors but was larger in agreement than tense (p<.0005) or aspect (p=.003) for morphological errors. Grammaticality judgement errors were classified as acceptances of incorrect sentences versus rejections of correct sentences. Patients made more acceptance and more rejection errors than controls (both p<.0005). The group difference did not interact with condition for acceptance errors but was smaller in tense (p=.034) for rejection errors. Patients made more acceptance errors than rejection errors overall (p=.021); this difference was greater for tense and aspect (p<.0005). Similarly, control participants made more acceptance errors in tense and aspect (both p<.0005). The attached table displays the overall proportion of errors of each type relative to the number of test items per category, for each task, group, and functional category condition.
Table 1. Proportion of errors as a percentage of the total number of sentences per condition, for each group.
Sentence completion
AgreementTenseAspect
TotalLexicalFormTotalLexicalFormTotalLexicalForm
Aphasics
M 30.0 14.3 20.3 35.0 15.0 24.0 36.0 12.3 27.5
SD 19.9 19.2 13.7 24.5 16.2 23.8 9.5 10.0 11.7
Controls
M 2.0 1.8 0.8 8.5 0.3 8.3 7.0 0.3 6.8
SD 2.3 2.3 1.2 16.1 0.8 16.2 5.9 0.8 6.0
Grammaticality judgment
AgreementTenseAspect
TotalAcceptRejectTotalAcceptRejectTotalAcceptReject
Aphasics
M 19.0 22.0 16.0 42.0 66.3 17.8 39.3 57.5 21.0
SD 17.4 24.4 12.0 10.1 18.2 17.5 12.0 23.9 18.4
Controls
M 2.0 2.8 1.3 8.1 11.8 4.5 9.8 18.8 0.8
SD 2.4 3.0 3.2 6.3 11.1 4.0 7.1 14.6 1.2
Note: For sentence completion, totals may be less than the sum of lexical and form errors because both types of errors may appear on the same item (mixed errors), counting as one in the total.

Discussion: Our findings show similar patterns of error distributions for patients with aphasia and control participants, taking into account the large and expected overall differences between the two groups. The data show no evidence for a selective deficit, therefore the results are not compatible with structural approaches to agrammatism. Rather, our findings appear compatible with processing accounts.

References:
Varlokosta, S., Valeonti, N., Kakavoulia, M., Lazaridou, M., Economou, A., & Protopapas, A. (2006). The breakdown of functional categories in Greek aphasia: Evidence from agreement, tense, and aspect. Aphasiology, 20 (8), 723–743.