Mechanisms underlying syntactic comprehension deficits in vascular aphasia: new evidence from self-paced listening

Abstract

Sixty-one people with aphasia (pwa) and 41 matched controls were tested for the ability to understand sentences that required the ability to process particular syntactic elements and assign particular syntactic structures. Participants paced themselves word-by-word through 20 examples of 11 spoken sentence types and indicated which of two pictures corresponded to the meaning of each sentence. Sentences were developed in pairs such that comprehension of the experimental version of a pair required an aspect of syntactic processing not required in the corresponding baseline sentence. The need for the syntactic operations required only in the experimental version was triggered at a “critical word” in the experimental sentence. Listening times for critical words in experimental sentences were compared to those for corresponding words in the corresponding baseline sentences. The results were consistent with several models of syntactic comprehension deficits in pwa: resource reduction, slowed lexical and/or syntactic processing, abnormal susceptibility to interference from thematic roles generated non-syntactically. They suggest that a previously unidentified disturbance limiting the duration of parsing and interpretation may lead to these deficits, and that this mechanism may lead to structure-specific deficits in pwa. The results thus point to more than one mechanism underlying syntactic comprehension disorders both across and within pwa.

Keywords:

• syntactic comprehension disorders
• aphasia
• on-line measures

Supplemental data

Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/02643294.2015.1058253.

Notes

1. To avoid confusion in the case of researchers potentially interested in performing meta-analyses, we note that the participants in this study are the same as those reported in Caplan et al. (2013a, 2013b) and at meetings of several scientific societies (Academy of Aphasia, Neurobiology of Language, CUNY sentence processing). The end-of-sentence data have been reported in depth in those papers; the on-line data have not been presented fully but have been discussed in relation to the short term/working memory performances of the participants. The descriptions of people with aphasia, the behavioural methods, and the end-of-sentence data are the same as in prior publications and are repeated here for the reader's convenience. The description and analysis of the on-line self-paced listening data are entirely novel and unavailable in the literature.

2. Because the relative clause verb was the sentence-final word in CO sentences, where responses to the sentence were also made (see below), we did not compare the relative clause verbs in CO and CS sentences. In sentences with reflexives and pronouns, the critical words in both the experimental and baseline sentences were sentence-final, and could be compared. In sentences that contained both a relative clause and a pronoun or reflexive, there were two critical words: the verb of the relative clause when the contrast was between the object- and the corresponding subject-extracted relative clause (both with either a pronoun or a reflexive), and the pronoun or reflexive when the comparison was between the pronoun or reflexive and the third NP in the corresponding baseline sentence with the same type of relative clause (object- or subject-extracted).

3. If words were not truncated when the button press occurred, a participant could simply press the button rapidly many times, resulting in hearing the entire sentence without pauses. Just and Carpenter (1980) found that this occurred in self-paced reading with non-disappearing cumulative text.

4. In this study, negative corrected listening times occurred on 57% of all words in controls and 59.5% of all words in pwa. Negative corrected listening times are due to two factors. The first is that words can be recognized at their uniqueness points, which can occur prior to their termination in multisyllabic words. The second is that contextual influences on word occurrences lead to word recognition before the end of a word and often before its uniqueness point. In this study, most words were monosyllabic, and their uniqueness point was the last phoneme. Contextual constraints are not likely to be derived pragmatically or from statistical properties of the language in this study but can arise from the fact that we used a limited number of words. The presence of negative corrected listening times does not create a problem for the interpretation of results because, in all sentences except those with pronouns and reflexives, the critical words in experimental and baseline sentences were the same, making the effect of the processes that led to negative corrected listening times the same in the critical comparisons. In the case of pronouns and reflexives, both word properties and contextual effects favour early recognition of pronouns and reflexives over full NPs, biasing against the longer self-paced listening times we found in pronouns and reflexives.

5. One problem is that many studies have tested only pwa with Broca's aphasia or agrammatism, and it is unclear whether other pwa have the same behaviours as those seen in this group. A second problem is that there currently is no explanation of why agrammatism or Broca's aphasia should be linked to syntactic comprehension disorders, or to particular types of syntactic comprehension disorders. Agrammatism could be causally related to a syntactic comprehension disorder because that the functional deficit(s) that produce(s) agrammatism affect syntactic comprehension (an “overarching” deficit: Zurif, 1978), but no one has been able to specify a functional deficit that leads to both expressive agrammatism and the disorders of syntactic comprehension that have been identified. Loss of a particular syntactic element could explain a deficit in both production and comprehension, but modern analyses that postulate such deficits propose different disturbances in production and comprehension (e.g, Grodzinsky, 2000b, for discussion ). Alternatively, the lesions that produce agrammatism might affect the brain areas that support syntactic comprehension. However, the lesions that produce Broca's aphasia or agrammatism are heterogeneous and often large (Mohr et al., 1978; Vanier & Caplan, 1990) and can spare Broca's area (Vanier & Caplan, 1990); thus, the hypothesis that agrammatic pwa or pwa with Broca's aphasia have syntactic comprehension disorders is not the same as the hypothesis that lesions in Broca's area cause syntactic comprehension disorders and requires analysis of lesions, not of pwa with Broca's aphasia. A third problem is that researchers have not agreed about whether the pwa to whom claims apply are agrammatic pwa, pwa with Broca's aphasia, or both. Zurif et al. (1989) argued that the Trace Deletion Hypothesis (actually, a variant of the hypothesis) applies to agrammatic patients (more specifically, patients who have “a mix of nonfluency, closed class item omission, and syntactic simplification”). Zurif and Pinango (1999) argued that the hypothesis does not apply to agrammatic aphasics but to Broca's aphasics. Grodzinsky (2000) excluded patients who are classified as Broca's aphasics on the basis of apraxia of speech or dysarthria. Zurif and Pinango (1999) at one point mention “agrammatic Broca's” aphasics, perhaps intending to claim that being both agrammatic and a Broca's aphasic is necessary for inclusion.

6. A fourth problem with examining pwa with Broca's aphasia or agrammatism for particular patterns of syntactic comprehension is how to make the diagnosis of Broca's aphasia or agrammatism. Researchers have disagreed about this, with important consequences for theories (see Caplan, 2001, for discussion). Our approach to this problem was based on principles presented by Caplan (1988), who argued that a minimal requirement of research that makes claims about syntactic comprehension disorders in groups of pwa such as those with agrammatism is to use objective measures of performance that can be scored objectively and that are preserved for analysis by other researchers as the basis for classification.

7. The reason for the significant effect of sentence type in agrammatic pwa in these last two analyses was partly due to residual corrected listening times being very fast for the third NPs in the baseline sentences in agrammatic pwa—these residual corrected listening times trended to be faster in agrammatic pwa than in controls (p = .08). However, there was also a trend for residual corrected listening times to be slower for reflexives for agrammatic pwa than for fluent aphasics (p = .07).

8. ANOVAs of the residuals of these regressions in the groups of pwa described above did not show any significant effects in any group of sentence types.

9. We note that this may be true of some normal individuals as well, complicating the interpretation of self-paced listening times, eye fixation durations, and other measures of incremental processing in normal individuals.

Additional information

Funding

This research was supported by National Institute of Deafness and Communication Disorders (NIDCD) [grant number DC00942] to David Caplan.