Working memory constraints on syntactic processing: an electrophysiological investigation.

Event-related potentials (ERPs) and reaction times (RTs) were used to study how the processing of sentences with morphosyntactic violations is constrained by working memory (WM) capacity. The available WM capacity was varied by three orthogonal manipulations: (1) syntactic complexity; (2) additional WM load; and (3) verbal WM span. The processing of the morphosyntactic violations was reflected in longer RTs in ungrammatical compared with grammatical sentences, and in an anterior negativity and a centroparietal positivity in the ERPs. While the behavioral grammaticality effect was not influenced by the WM manipulations, the ERP effects were. The amplitude of the anterior negativity was modulated by the combination of complexity and load, and by WM span. The onset of the centroparietal positivity was delayed in the high-load condition, and for the low-span group. ERPs over the course of the sentences showed a frontal negative slow wave under high WM load, largest for the low-span group. The finding that online syntactic processing is related to WM span and to additional WM load does not support the theory that there is a WM capacity specific for syntactic processing.

Error monitoring in speech production: a computational test of the perceptual loop theory.

A theory of speech monitoring, proposed by Levelt (1983), assumes that the quality of one's speech is checked by the speech comprehension system. This system inspects one's own overt speech but would also inspect an inner speech plan ("the inner loop"). We have elaborated and tested this theory by way of formalizing it as a computational model. This model includes a new proposal concerning the timing relation between planning the interruption and the repair: the proposal that these two processes are performed in parallel. We attempted to simulate empirical data about the distribution of error-to-cutoff and cutoff-to-repair intervals and the effect of speech rate on these intervals (these intervals are shorter with faster speech). The main questions were (1) Is an inner monitor that utilizes the speech perception system fast enough to simulate the timing data? (2) Can the model account for the effects of speech rate on these intervals? We conclude that including an inner loop through the speech comprehension system generates predictions that fit the empirical data. The effects of speed can be accounted for, given our proposal about the time course of planning interruption and repair. A novel prediction is that the error-to-cutoff interval decreases with increasing position in the phrase.

Prearticulatory and postarticulatory self-monitoring in Broca's aphasia.

The present study examined to what extent patients with Broca's aphasia and healthy controls rely upon prearticulatory and postarticulatory monitoring processes for detecting and repairing errors in speech production. Monitoring skills were investigated in a speaking situation with normal auditory feedback, a speaking situation with white noise, and a situation in which errors had to be detected in other-produced speech. The results demonstrated that the Broca's aphasics repaired a lower percentage of errors than the controls in the situation with normal auditory feedback, whereas their performance in the noise-masked condition was comparable. In contrast to the controls, the aphasics did not suffer from the presence of white noise. In addition, the proportion of covert repairs was higher for the Broca's aphasics than for the healthy controls. These findings indicate that Broca's aphasics concentrate primarily on prearticulatory monitoring. Possible explanations for this strong reliance on prearticulatory monitoring processes are discussed.

Agrammatic production of subject-verb agreement: the effect of conceptual number.

Three experiments tested the hypothesis that the deficit underlying agrammatic sentence production difficulties can be characterized as a limitation of computational resources and that these resources are not restricted to syntactic processing. This hypothesis was tested by eliciting subject-verb agreement errors in a sentence fragment completion paradigm. Sentence fragments were complex noun phrases, containing a subject (head) noun and a modifying prepositional phrase, containing a "local" noun. We varied the number of "tokens" a singular head noun referred to. Therefore, in one condition, grammatical and conceptual number of the head noun mismatched, whereas these numbers were the same in another condition. In Experiments 1 and 2, we observed an effect of this variable (i.e., more agreement errors when conceptual number was plural and grammatical number singular) in normal controls. Broca's aphasics, on the other hand, showed no effect. Experiment 3 consisted of a sentence/picture matching test. This test showed that the lack of effect with Broca's aphasics cannot be attributed to a comprehension deficit. We argue that these results are incompatible with the notion of a limitation in resources specific for syntactic processing. Instead, we interpret this as the result of a trade-off: Broca's aphasics lack computational resources to take into account both grammatical and conceptual information in morphosyntactic processing and rely on grammatical information only.

Syntactic facilitation in agrammatic sentence production.

Recently, proposals have been made to relate processing difficulties in aphasic language performance to limitations in resources for grammatical processing (Carpenter et al., 1994; Hagiwara, 1995; Kolk, 1995; Martin & Romani, 1994). Such proposals may account for a defining characteristic of agrammatic sentence production: reduced syntactic complexity. Syntactic structures that require deep hierarchical processing or reversals of canonical word order make demands exceeding limited resources. In the present study, we investigate the possibility of counteracting hypothesized resource limitations by increasing the availability of relatively complex sentences (i.e., datives and passives). The phenomenon of "syntactic priming" has been observed in a number of studies with healthy adults (e.g., Bock, 1986). With respect to Broca's aphasia, we hypothesized that increased availability of a syntactic structure, due to syntactic priming, results in a lesser demand on (limited) resources for sentence production. We elicited speech from 12 Broca's aphasics and 12 control subjects in three different conditions: spontaneous speech, picture description without priming, and picture description with priming. In addition, we varied instructions, in order to determine the role of strategies. The main findings were that (a) Brocas show stronger syntactic priming effects than controls; (b) the effects are automatic rather than strategic; and (c) in conditions with priming, Brocas produce relatively complex sentences (e.g., passives). We discuss these results in relation to capacity theories.

Syntactic persistence in Dutch.

Three experiments are reported that showed effects of "structure priming," the tendency to repeat syntactic structure across successive sentences. These effects were demonstrated in a previously untested language, Dutch. All experiments studied spoken sentence production. Importantly, pre-experimental baselines were measured for all target structures in order to assess possible effects of frequency on the magnitude of priming effect. We obtained priming with dative sentences, including datives with medially placed prepositional phrases, a sentence type not tested before. In one experiment we obtained priming effects with passives, including passives with sentence-final verbs, which also have never been tested before. However, we failed to obtain priming effects with active transitives. A comparison with the baseline data suggested that priming is not related to baseline frequency. Further, the results allowed us to draw an empirical generalization: Structure priming is a relatively long-term event, lasting at least several trials. The implications of these findings are discussed.

On-line sensitivity to subject-verb agreement violations in Broca's aphasics: the role of syntactic complexity and time.

This study examined on-line sensitivity to subject-verb agreement violations in patients with Broca's aphasia and age-matched controls using a word monitoring paradigm. The agreement violations were couched in either simple or complex syntactic frames. In a first experiment, these syntactic frames were immediately followed by the noun phrase containing the target, whereas in the second experiment a 750-msec separation was introduced. The main finding of the first experiment was that patients with Broca's aphasia showed an agreement effect only for simple (i.e., conjoined) sentences but not for complex (i.e., embedded) ones, while controls showed the expected agreement effect for both. The results of the second experiment demonstrated further that the 750-msec delay in target presentation abolished the agreement effect in Broca's aphasics but not in normal controls. The findings are interpreted to suggest that Broca's suffer from a pathological limitation in parsing capacity, giving rise to a faster than normal decay of syntactic information.

The effects of task variation on the production of grammatical morphology in Broca's aphasia: a multiple case study.

Generally task variation is not taken into account in discussions about agrammatism in Broca's aphasia. Both free speech and speech under constrained conditions are classified as "agrammatic speech." In this study, we investigated the effects of task variation on the character of agrammatic speech. We employed three types of task situations, an interview and two picture description tasks, to elicit speech from 16 Dutch and 3 German Broca's aphasics. The two picture description tasks differed in the extent to which they necessitated the production of grammatical morphology. Various task-dependent changes in the character of agrammatic speech were observed. Overall, omission rates of grammatical morphemes went down and substitution rates tended to go up in picture description as compared to free conversation. Results are discussed by contrasting two explanatory concepts, task interaction and strategic variation.

The production of grammatical morphology in Broca's and Wernicke's aphasics: speed and accuracy factors.

A group of Broca's aphasics (BA), Wernicke's aphasics (WA) and normal controls participated in a CLOZE experiment which required the oral production of various types of free and bound morphemes. Results provided support for the hypothesis that BA and WA share the same underlying impairment in the production of grammatical morphology. The relative difficulty of the various free and bound morpheme types was the same for BA and WA. This appeared to be the case not only in an analysis of the number of errors but also in an analysis of response times. The same analyses furthermore revealed no significant differences in the absolute levels of performance of BA and WA. Finally, it was found that BA and WA show the same relative contribution of within- and across-category substitutions of free morphemes. For bound morphemes, there was a slight difference between BA and WA, in that BA exclusively produced within-category substitutions while WA also produced some across-category substitutions.

On the Bradley hypothesis concerning agrammatism: the nonword-interference effect.

In 1980, D. C. Bradley, M. F. Garrett, and E. B. Zurif (in D. Caplan (Ed.), Biological studies of mental processes, Cambridge, MA: MIT Press) have proposed a lexical hypothesis on agrammatism. The theoretical background of this hypothesis is described. Then two experiments are reported with normal speakers on the so-called nonword-interference effect that has been one of the two major sources of experimental evidence for this hypothesis. The first experiment replicates the original finding that the interference effect is found when the nonword starts with a content word but not when it starts with a function word. In the second experiment, it is shown that this contrast is due to an artifact of the composition of the word list. When this list contains both content and function words, rather than only content words--as in Bradley's experiments and our Experiment I--both content and function nonwords show an equally sized interference effect. The consequences for the lexical hypothesis are discussed.

Strategy and impairment in sentence understanding by Broca's and Wernicke's aphasics.

15 Broca's aphasics and 14 Wernicke's aphasics, both German and Dutch speaking patients, were presented with a sentence-picture matching task. Both syntactic and semantic distractor pictures were used. Sentences were either reversible or non-reversible, had a topicalized or a non-topicalized word order and contained either of three types of prepositions that were labelled 'syntactic' (case-marking), 'obligatory' (subcategorized) and 'lexical' (locative). Results indicated (a) a significant difference between reversible and non reversible sentences for both Broca's and Wernicke's; (b) no significant differences between these two groups in the differences scores reversible minus non-reversible; (c) an above-chance performance on reversible sentences for both groups. To explain the pattern of differences between the various types of sentences, two schemes of interpretation are presented. The first scheme assumes that both Broca's and Wernicke's have lost basic (but different) syntactic abilities and that they can understand reversible sentences by non-syntactic strategies only. The second scheme assumes that there is no such loss, but that the impairment in Broca's and Wernicke's aphasia is such that the more syntactic analysis is required, the more errors are made. Patients are assumed to approach this task by trying to limit their syntactic analysis to the beginning of the sentence.

Metalinguistic judgments on sentence structure in agrammatism: a matter of task misinterpretation.

Judgments of sentence structure were studied in two non-agrammatic and four agrammatic patients. The subjects were required to sort the words from a sentence on the basis of how closely related they felt the words to be in that particular sentence. A hierarchical cluster analysis was then carried out on the data. The non-agrammatic patients produced typical "agrammatic" subjective trees in which function words and adjectives were weakly related to the rest of the sentence. The agrammatic patients produced structures that were fully normal. Two conclusions are drawn. First, this metalinguistic paradigm is inadequate as an indicator of an impairment in the mental representation of constituent structure. Second, the hypothesis that agrammatic speakers do not have such a representation is seriously weakened. Implications are discussed for the methodology of aphasia research as well as for theories on the nature of agrammatism.