Rational Adaptation in Lexical Prediction: The Influence of Prediction Strength.

Recent studies indicate that the processing of an unexpected word is costly when the initial, disconfirmed prediction was strong. This penalty was suggested to stem from commitment to the strongly predicted word, requiring its inhibition when disconfirmed. Additional studies show that comprehenders rationally adapt their predictions in different situations. In the current study, we hypothesized that since the disconfirmation of strong predictions incurs costs, it would also trigger adaptation mechanisms influencing the processing of subsequent (potentially) strong predictions. In two experiments (in Hebrew and English), participants made speeded congruency judgments on two-word phrases in which the first word was either highly constraining (e.g., "climate," which strongly predicts "change") or not (e.g., "vegetable," which does not have any highly probable completion). We manipulated the proportion of disconfirmed predictions in highly constraining contexts between participants. The results provide additional evidence of the costs associated with the disconfirmation of strong predictions. Moreover, they show a reduction in these costs when participants experience a high proportion of disconfirmed strong predictions throughout the experiment, indicating that participants adjust the strength of their predictions when strong prediction is discouraged. We formulate a Bayesian adaptation model whereby prediction failure cost is weighted by the participant's belief (updated on each trial) about the likelihood of encountering the expected word, and show that it accounts for the trial-by-trial data.

From pre-activation to pre-updating: A threshold mechanism for commitment to strong predictions.

Ample evidence suggests that during sentence processing comprehenders can "pre-activate" lexical/semantic knowledge stored in long-term memory. A relatively recent development suggests that in some cases a stronger form of prediction is employed, involving "pre-updating" the predicted content into the sentence's representation being built in working memory. The current study argues for an activation threshold mechanism by which pre-updating is initiated, within the routine processing stages of a word in a context. By combining a speeded cloze task with event-related potentials, we were able to analyze electrophysiological data measured prior to when participants were prompted to produce a completion, based on the participant's cloze response, reflecting their strongest prediction at that specific moment in time. A P600 effect reflecting pre-updating was observed in high (relative to low) constraint sentences, even in trials where the participant predicted a low cloze word. The results support a mechanism in which multiple predictions accumulate activations, "racing" toward a retrieval threshold. Once the activation level of a certain word passes the threshold, the word is integrated into the sentence representation in working memory. Pre-updating occurs if a certain prediction passes the retrieval threshold prior to its realization in the input.

Noisy is better than rare: Comprehenders compromise subject-verb agreement to form more probable linguistic structures.

Production and perception errors are common in everyday language use. Recent studies suggest that in order to overcome the flawed speech signal, comprehenders engage in rational noisy-channel processing, which can pull their interpretation towards more probable "near-neighbor" analyses, based on the assumption that an error may have occurred in the transmission of the sentence. We investigate this type of processing using subject/object relative clause ambiguity in Hebrew. In four self-paced reading experiments and a sentence completion experiment, we find that during online processing, readers apply elaborate knowledge regarding the distribution of structures in the language, and that they are willing to compromise subject-verb agreement to refrain from (grammatical but) highly improbable structures. The results suggest that the prior probability of alternative analyses modulates the interpretation of agreement.

Love thy neighbor: Facilitation and inhibition in the competition between parallel predictions.

Ample evidence suggests that during word recognition and production, simultaneously activated lexical and sublexical representations interact, demonstrating varied patterns of facilitation and inhibition in various tasks and measures. A separate line of research has led to a growing consensus that prediction during sentence processing involves activating multiple possible predictions. However, very little is known about the nature of the interactions between parallel predictions. The current study employed a speeded cloze task to probe competition between simultaneously activated predictions. We focused on the modal response (the most probable completion for a sentence) and its strongest competitor (the second most probable completion). Examining production latencies of the modal response, the results showed an interaction between competitor strength and the semantic relatedness between the competitor and the modal: when the two were related, the stronger the competitor was, the more it facilitated production; however, when the two were unrelated, the stronger the competitor was, the more inhibition it caused. These results contrast with the pattern observed for the influences of near and distant semantic neighbors on word recognition and production. However, we show that when the different nature of the tasks is taken into consideration, these patterns of interaction between parallel predictions can be accounted for by the interactive activation and competition (IAC) model used to account for previous neighborhood effects (Chen & Mirman, 2012).

Neurocognitive correlates of category ambiguous verb processing: The single versus dual lexical entry hypotheses.

Word-class ambiguous words engender greater processing time and fMRI (BOLD signal) activation than unambiguous ones. Theoretical accounts of this phenomenon suggest that words with multiple meanings (1) are associated with multiple lexical entries and thus require greater selection demands, or (2) undergo computationally expensive grammatical processes that convert words from one word-class to another. Using an fMRI grammaticality judgment task, we tested these accounts by examining word-class ambiguous polysemic (e.g., brush) and homonymic (e.g., bear) verbs, and unambiguous verbs (e.g., bake). Results showed that ambiguous verbs evoked longer response times and greater neural activation in the left inferior frontal and parietal gyri. However, homonymic verbs also showed increased left inferior frontal and temporal neural activations compared to polysemic verbs. This indicates that rather than having multiple lexical representations like homonyms, polysemic verbs may share a core representation with their noun counterparts.

Predictive Pre-updating and Working Memory Capacity: Evidence from Event-related Potentials.

It was recently proposed that lexical prediction in sentence context encompasses two qualitatively distinct prediction mechanisms: "pre-activation," namely, activating representations stored in long-term memory, and "pre-updating," namely, updating the sentence's representation, built online in working memory (WM), to include the predicted content [Lau, E. F., Holcomb, P. J., & Kuperberg, G. R. Dissociating N400 effects of prediction from association in single-word contexts. Journal of Cognitive Neuroscience, 25, 484-502, 2013]. The current study sought to find evidence for pre-updating and test the influence of individual differences in WM capacity on the tendency to engage in this process. Participants read strongly and weakly constraining sentences. ERPs were measured on the predictable noun as well as on the preceding verb, where the prediction is generated. Increased P600 amplitude was observed at the verb in the strongly constraining sentences, reflecting integration of the predicted upcoming argument, thus providing evidence for pre-updating. This effect was greater for participants with higher WM capacity, indicating that the tendency to engage in pre-updating is highly affected by WM capacity. The opposite effect was observed at the noun, that is, for participants with higher WM span, a greater decrease in P600 amplitude in the strongly constraining sentences was observed, indicating that the integration of a pre-updated word was easier. We discuss these results in light of previous literature and propose a plausible architecture to account for the interplay between pre-activation and pre-updating, mediating the influence of factors such as WM capacity.

Lexical inhibition due to failed prediction: Behavioral evidence and ERP correlates.

During sentence processing, comprehenders form expectations regarding upcoming material, and may even predict a specific word. Previous event-related potential (ERP) studies have shown that disconfirmed predictions elicit a post-N400-positivity (PNP) with two distinct distributions. A frontal-PNP (f-PNP) is elicited when an unexpected but congruent word appears instead of a highly predictable word, whereas an anomalous word elicits a posterior-PNP. The current study tested the hypothesis that during the processing of a sentence, the predicted word is inhibited to enable the integration of unexpected but congruent material, and that this inhibitory process is reflected in the f-PNP component. In contrast, anomalous continuations, which are not compatible with the preceding context, do not induce inhibition. Experiment 1 used cross-modal lexical priming to test inhibition patterns of predicted words, demonstrating inhibition when integration of a congruent-unexpected word was needed, but not when an anomaly was encountered. Experiment 2 showed that the inhibition observed in Experiment 1 is specific to the predicted word and does not stem from competition between two congruent continuations. In Experiment 3 we recorded ERPs using the same materials, and found that the f-PNP component is elicited under the same conditions giving rise to behavioral inhibition, and that the two are correlated, thus providing preliminary support for the hypothesis that this component reflects an inhibitory process. (PsycINFO Database Record

Working Memory in the Processing of Long-Distance Dependencies: Interference and Filler Maintenance.

During the temporal delay between the filler and gap sites in long-distance dependencies, the "active filler" strategy can be implemented in two ways: the filler phrase can be actively maintained in working memory ("maintenance account"), or it can be retrieved only when the parser posits a gap ("retrieval account"). The current study tested whether filler content is maintained during the processing of dependencies. Using a self-paced reading paradigm, we compared reading times on a noun phrase (NP) between the filler and gap sites in object relative clauses, to reading times on an NP between the antecedent and ellipsis sites in ellipsis sentences. While in the former type of dependency a filler by hypothesis can be maintained, in the latter there is no indication for the existence of a dependency prior to the ellipsis site, and hence no maintenance. By varying the amount of similarity-based interference between the antecedent and integration sites, we tested the influence of holding an unresolved dependency on reading times. Significantly increased reading times due to interference were found only in the object relative condition, and not in the ellipsis condition, demonstrating filler maintenance costs. The fact that these costs were measured as an effect on similarity-based interference indicates that the maintained representation of the filler must include at least some of the features shared by the interfering NP.

What do pauses in narrative production reveal about the nature of word retrieval deficits in PPA?

Naming and word-retrieval deficits, which are common characteristics of primary progressive aphasia (PPA), differentially affect production across word classes (e.g., nouns, verbs) in some patients. Individuals with the agrammatic variant (PPA-G) often show greater difficulty producing verbs whereas those with the semantic variant (PPA-S) show greater noun deficits and those with logopenic PPA (PPA-L) evince no clear-cut differences in production of the two word classes. To determine the source of these production patterns, the present study examined word-finding pauses as conditioned by lexical variables (i.e., word class, frequency, length) in narrative speech samples of individuals with PPA-S (n=12), PPA-G (n=12), PPA-L (n=11), and cognitively healthy controls (n=12). We also examined the relation between pause distribution and cortical atrophy (i.e., cortical thickness) in nine left hemisphere regions of interest (ROIs) linked to word production. Results showed higher overall pause rates for PPA compared to unimpaired controls; however, greater naming severity was not associated with increased pause rate. Across all groups, more pauses were produced before lower vs. higher frequency words, with no independent effects of word length after controlling for frequency. With regard to word class, the PPA-L group showed a higher rate of pauses prior to production of nouns compared to verbs, consistent with noun-retrieval deficits arising at the lemma level of word production. Those with PPA-G and PPA-S, like controls, produced similar pause rates across word classes; however, lexical simplification (i.e., production of higher-frequency and/or shorter words) was evident in the more-impaired word class: nouns for PPA-S and verbs for PPA-G. These patterns are consistent with conceptual and/or lemma-level impairments for PPA-S, predominantly affecting objects/nouns, and a lemma-level verb-retrieval deficit for PPA-G, with a concomitant impairment in phonological encoding and articulation affecting overall pause rates. The greater tendency to pause before nouns was correlated with atrophy in the left precentral gyrus, inferior frontal gyrus and inferior parietal lobule, whereas the greater tendency to pause before less frequent and longer words was associated with atrophy in left precentral and inferior parietal regions.

How the brain processes different dimensions of argument structure complexity: evidence from fMRI.

Verbs are central to sentence processing, as they encode argument structure (AS) information, i.e., information about the syntax and interpretation of the phrases accompanying them. The behavioral and neural correlates of AS processing have primarily been investigated in sentence-level tasks, requiring both verb processing and verb-argument integration. In the current functional magnetic resonance imaging (fMRI) study, we investigated AS processing using a lexical decision task requiring only verb processing. We examined three aspects of AS complexity: number of thematic roles, number of thematic options, and mapping (non)canonicity (unaccusative vs. unergative and transitive verbs). Increased number of thematic roles elicited greater activation in the left posterior perisylvian regions claimed to support access to stored AS representations. However, the number of thematic options had no neural effects. Further, unaccusative verbs elicited longer response times and increased activation in the left inferior frontal gyrus, reflecting the processing cost of unaccusative verbs and, more generally, supporting the role of the IFG in noncanonical argument mapping.

The forgotten grammatical category: Adjective use in agrammatic aphasia.

BACKGROUND: In contrast to nouns and verbs, the use of adjectives in agrammatic aphasia has not been systematically studied. However, because of the linguistic and psycholinguistic attributes of adjectives, some of which overlap with nouns and some with verbs, analysis of adjective production is important for testing theories of word class production deficits in agrammatism. AIMS: The objective of the current study was to compare adjective use in agrammatic and healthy individuals, focusing on three factors: overall adjective production rate, production of predicative and attributive adjectives, and production of adjectives with complex argument structure. METHOD & PROCEDURES: Narratives elicited from 14 agrammatic and 14 control participants were coded for open class grammatical category production (i.e., nouns, verbs, adjectives), with each adjective also coded for its syntactic environment (attributive/predicative) and argument structure. OUTCOMES & RESULTS: Overall, agrammatic speakers used adjectives in proportions similar to that of cognitively healthy speakers. However, they exhibited a greater proportion of predicative adjectives and a lesser proportion of attributive adjectives, compared to controls. Additionally, agrammatic participants produced adjectives with less complex argument structure than controls. CONCLUSIONS: The overall normal-like frequency of adjectives produced by agrammatic speakers suggests that agrammatism does not involve an inherent difficulty with adjectives as a word class or with predication, or that it entails a deficit in processing low imageability words. However, agrammatic individuals' reduced production of attributive adjectives and adjectives with complements extends previous findings of an adjunction deficit and of impairment in complex argument structure processing, respectively, to the adjectival domain. The results suggest that these deficits are not tied to a specific grammatical category.

Training verb argument structure production in agrammatic aphasia: behavioral and neural recovery patterns.

INTRODUCTION: Neuroimaging and lesion studies indicate a left hemisphere network for verb and verb argument structure processing, involving both frontal and temporoparietal brain regions. Although their verb comprehension is generally unimpaired, it is well known that individuals with agrammatic aphasia often present with verb production deficits, characterized by an argument structure complexity hierarchy, indicating faulty access to argument structure representations for production and integration into syntactic contexts. Recovery of verb processing in agrammatism, however, has received little attention and no studies have examined the neural mechanisms associated with improved verb and argument structure processing. In the present study we trained agrammatic individuals on verbs with complex argument structure in sentence contexts and examined generalization to verbs with less complex argument structure. The neural substrates of improved verb production were examined using functional magnetic resonance imaging (fMRI). METHODS: Eight individuals with chronic agrammatic aphasia participated in the study (four experimental and four control participants). Production of three-argument verbs in active sentences was trained using a sentence generation task emphasizing the verb's argument structure and the thematic roles of sentential noun phrases. Before and after training, production of trained and untrained verbs was tested in naming and sentence production and fMRI scans were obtained, using an action naming task. RESULTS: Significant pre- to post-training improvement in trained and untrained (one- and two-argument) verbs was found for treated, but not control, participants, with between-group differences found for verb naming, production of verbs in sentences, and production of argument structure. fMRI activation derived from post-treatment compared to pre-treatment scans revealed upregulation in cortical regions implicated for verb and argument structure processing in healthy controls. CONCLUSIONS: Training verb deficits emphasizing argument structure and thematic role mapping is effective for improving verb and sentence production and results in recruitment of neural networks engaged for verb and argument structure processing in healthy individuals.

Syntactic and morphosyntactic processing in stroke-induced and primary progressive aphasia.

The paper reports findings derived from three experiments examining syntactic and morphosyntactic processing in individuals with agrammatic and logopenic variants of primary progressive aphasia (PPA-G and PPA-L, respectively) and stroke-induced agrammatic and anomic aphasia (StrAg and StrAn, respectively). We examined comprehension and production of canonical and noncanonical sentence structures and production of tensed and nontensed verb forms using constrained tasks in experiments 1 and 2, using the Northwestern Assessment of Verbs and Sentences (NAVS [57]) and the Northwestern Assessment of Verb Inflection (NAVI, Thompson and Lee, experimental version) test batteries, respectively. Experiment 3 examined free narrative samples, focusing on syntactic and morphosyntactic measures, i.e. production of grammatical sentences, noun to verb ratio, open-class to closed-class word production ratio, and the production of correctly inflected verbs. Results indicate that the two agrammatic groups (i.e., PPA-G and StrAg) pattern alike on syntactic and morphosyntactic measures, showing more impaired noncanonical compared to canonical sentence comprehension and production and greater difficulties producing tensed compared to nontensed verb forms. Their spontaneous speech also contained significantly fewer grammatical sentences and correctly inflected verbs, and they produced a greater proportion of nouns compared to verbs, than healthy speakers. In contrast, PPA-L and StrAn individuals did not display these deficits, and performed significantly better than the agrammatic groups on these measures. The findings suggest that agrammatism, whether induced by degenerative disease or stroke, is associated with characteristic deficits in syntactic and morphosyntactic processing. We therefore recommend that linguistically sophisticated tests and narrative analysis procedures be used to systematically evaluate the linguistic ability of individuals with PPA, contributing to our understanding of the language impairments of different PPA variants.

The neural substrates of complex argument structure representations: Processing 'alternating transitivity' verbs.

This study examines the neural correlates of processing verbal entries with multiple argument structures using fMRI. We compared brain activation in response to 'alternating transitivity' verbs, corresponding to two different verbal alternates - one transitive and one intransitive - and simple verbs, with only one, intransitive, thematic grid. Fourteen young healthy participants performed a lexical decision task with the two verb types. Results showed significantly greater activation in the angular and supramarginal gyri (BAs 39 and 40) extending to the posterior superior and middle temporal gyri bilaterally, for alternating compared to simple verbs. Additional activation was detected in bilateral middle and superior frontal gyri (BAs 8 and 9). The opposite contrast, simple compared to alternating verbs, showed no significant differential activation in any regions of the brain. These findings are consistent with previous studies implicating a posterior network including the superior temporal, supramarginal and angular gyri for processing verbs with multiple thematic roles, as well as with those suggesting involvement of the middle and superior frontal gyri in lexical ambiguity processing. However, because 'alternating transitivity' verbs differ from simple intransitives with regard to both the number of thematic grids (two vs. one) and the number of thematic roles (two vs. one), our findings do not distinguish between activations associated with these two differences.

Neural correlates of processing passive sentences.

Previous research has shown that comprehension of complex sentences involving wh-movement (e.g., object-relative clauses) elicits activation in the left inferior frontal gyrus (IFG) and left posterior temporal cortex. However, relatively little is known about the neural correlates of processing passive sentences, which differ from other complex sentences in terms of representation (i.e., noun phrase (NP)-movement) and processing (i.e., the time course of syntactic reanalysis). In the present study, 27 adults (14 younger and 13 older) listened to passive and active sentences and performed a sentence-picture verification task using functional Magnetic Resonance Imaging (fMRI). Passive sentences, relative to active sentences, elicited greater activation in bilateral IFG and left temporo-occipital regions. Participant age did not significantly affect patterns of activation. Consistent with previous research, activation in left temporo-occipital cortex likely reflects thematic reanalysis processes, whereas, activation in the left IFG supports processing of complex syntax (i.e., NP-movement). Right IFG activation may reflect syntactic reanalysis processing demands associated with the sentence-picture verification task.

Electrophysiological responses to argument structure violations in healthy adults and individuals with agrammatic aphasia.

Sentence comprehension requires processing of argument structure information associated with verbs, i.e. the number and type of arguments that they select. Many individuals with agrammatic aphasia show impaired production of verbs with greater argument structure density. The extent to which these participants also show argument structure deficits during comprehension, however, is unclear. Some studies find normal access to verb arguments, whereas others report impaired ability. The present study investigated verb argument structure processing in agrammatic aphasia by examining event-related potentials associated with argument structure violations in healthy young and older adults as well as aphasic individuals. A semantic violation condition was included to investigate possible differences in sensitivity to semantic and argument structure information during sentence processing. Results for the healthy control participants showed a negativity followed by a positive shift (N400-P600) in the argument structure violation condition, as found in previous ERP studies (Friederici & Frisch, 2000; Frisch, Hahne, & Friederici, 2004). In contrast, individuals with agrammatic aphasia showed a P600, but no N400, response to argument structure mismatches. Additionally, compared to the control groups, the agrammatic participants showed an attenuated, but relatively preserved, N400 response to semantic violations. These data show that agrammatic individuals do not demonstrate normal real-time sensitivity to verb argument structure requirements during sentence processing.