Export citation Request permission. References Hide All. Anagnostopoulou , E. Cross-linguistic and cross-categorical variation of datives. In Stavrou , M.
- Hello: We Speak the Truth.
- Frontiers | How language production shapes language form and comprehension | Psychology.
- Publications – Ferreira Lab?
Festshrift for Dimitra Theophanopoulou-Kontou, pp. Bowers , J. In Alexiadou , A. Cambridge : Cambridge Scholars Publishing. Clahsen , H. Antecedent priming at trace positions: evidence from German scrambling. Journal of Psycholinguistic Research , 28 , — Felser , C. Processing wh—dependencies in a second language: A cross—modal priming study. Second Language Research , 23 , 9 — Georgala , E. Short object shift and ditransitive structure in Greek.
Klepousniotou , E. The processing of lexical ambiguity: Homonymy and polysemy in the mental lexicon.
Brain and Language , 81 , — Love , T. The influence of language exposure on lexical and syntactic language processing. Experimental Psychology , 50 , — Coreference processing and levels of analysis in object-relative constructions: Demonstration of antecedent reactivation with the cross-modal priming paradigm. Journal of Psycholinguistic Research , 25 , 5 — Journal of Psycholinguistic Research , 36 , — Marinis , T. Using on-line processing methods in language acquisition research. In Blom , E. Amsterdam : John Benjamins.
On-line processing of wh-questions in children with G-SLI and typically developing children.
Sentence Processing: Toward a Neurobiological Approach - Oxford Handbooks
International Journal of Language and Communication Disorders , 42 , — McKee , C. Children's application of binding during sentence processing. Language and Cognitive Processes , 8 , — Miller , K. Facilitating the task for second language processing research: A comparison of two testing paradigms. Applied Psycholinguistics , 36 , — Intermediate traces and intermediate learners. Studies in Second Language Acquisition , 37 , — Nakano , Y. Antecedent priming at trace positions in Japanese long-distance scrambling. Journal of Psycholinguistic Research , 31 , — Nicol , J.
- Bestselling Series.
- Edited by Greig I. de Zubicaray and Niels O. Schiller?
- Franklin Chang - sentprodmodel;
- Mon premier amour (Littérature Française) (French Edition).
- Neurophysiological dynamics of phrase-structure building during sentence processing?
Reconsidering Reactivation. In Gerry , T. Hillsdale, N. The role of structure in coreference assignment during sentence comprehension. Half the items had a subject relative clause, and the other half an object relative clause. Thirty additional sentences were used as fillers, which consisted of 15 plausible and 15 implausible sentences.
The sentences were pre-tested to check if participants would indeed be sensitive to the manipulation of the attractor plausibility. They were always given two choices, and the choices for the experimental sentences were between a grammatical John got a meal from the takeaway. Outliers made up 3.
The mean accuracy and reaction time for each condition are summarised in Table 3. I compared 2AFC judgments of the interpretations when the attractor was plausible to when the attractor was implausible see Fig 2. The former was calculated by subtracting a standardized percentage of ungrammatical interpretation responses in c from the standardized percentage of grammatical interpretation responses in a , and the latter by subtracting the standardized percentage of incorrect responses in d from the standardized percentage of correct responses in b. A further linear mixed effects model was constructed with the lme4 package [ 59 ] in R to evaluate the effect of Cue Validity, the effect of Attractor Plausibility and the interaction of Cue Validity by Attractor Plausibility.
The model included random intercepts and slopes for Cue Validity and Attractor Plausibility by subjects and by items. The effects were regarded as significant when the associated absolute t -value exceeded 2 [ 60 ].
The effect of Attractor Plausibility was not significant. The items were divided into four counterbalanced lists, so that each list contained one condition per item and had the same number of sentences from each condition, and the same number of Plausible and Implausible sentences. The sentences were randomised so that no more than three sentences from the same condition appeared consecutively. The experimental procedure was the same as in Experiment 1.
The experiment took approximately 50 minutes. The data processing procedure was identical to that in Experiment 1, except that a standard pre-stimulus baseline to 0 ms was used because there were no observable differences early in the epoch or in the traditional pre-stimulus baseline period. An analogous statistical analysis from Experiment 1 was conducted on the time window from — ms on frontal and posterior channels in line with the prediction of an N effect arising from the manipulation of Attractor Plausibility.
A linear mixed-effects model which evaluated the ERP amplitude predicted by the fixed effects of Cue Validity Grammatical vs. Ungrammatical and Attractor Plausibility Plausible vs. Implausible , and by the interaction of the two, was built. The model additionally included random intercepts and slopes by participants and by items.
Random slopes by Cue Validity and Plausibility were not included because the model with these random slopes did not converge. Visual inspection of the ERP data suggests that Invalid Cue critical phrases elicited greater negativity than grammatical equivalents when the attractor was Plausible, but not when the attractor was Implausible. This negativity was most prominent at frontal channels see Fig 3. ERPs elicited by each condition at FCz and Pz, and the scalp distribution of the effect of Cue Validity for the Implausible and the Plausible Attractor conditions in the — ms time window befitting the N modulation.
The pattern of ERPs on FCz is representative of the effects across frontal channels, while the pattern on Pz is representative of posterior channels. The model did not show any significant main effects. To resolve the interaction, another linear mixed-effects model testing an effect of Cue Validity for Plausible and Implausible Attractor conditions was run separately. To sum up, there was a greater frontal negativity for Invalid Cue versus Valid Cue critical phrases when the attractor was Plausible but not when the attractor was Implausible. The interaction was driven by the difference between the Plausible conditions—only when the attractor could plausibly combine with the embedded subject was there a modulation of the N This modulation likely reflects a disruption in processing, either at retrieval of the antecedent or during integration of the embedded subject and the antecedent.
As in Experiment 1, the interaction suggests that information in the recent sentence context and the information on the retrieval cues at the ellipsis site is compared or combined during sentence processing. A functional interpretation of the interaction is that the partial cue match between plausible attractors and an ungrammatical retrieval cue led to the Plausible attractor being momentarily considered as an antecedent of the ellipsis in the absence of a sufficient antecedent being elicited.
Alternatively, the effect might reflect that fact that the system is sensitive to semantic similarity, even across syntactically illicit relations, and this sensitivity is only measurable when no grammatical antecedent is elicited. The pattern of results from Experiments 1 and 2 showed evidence suggesting that information available at retrieval, that is, both information in memory and retrieval cues, interacts.
The interaction shaped processing outcomes such that the pattern of results across both experiments is consistent with an architecture where the contents of memory are linearly combined with cues at retrieval to elicit the antecedent, providing a representation for interpretation in a new sentence position and referential relation. Both of these findings suggest that information at retrieval, in the form of integrated cues, must have interacted with the contents of memory in order to produce the observed pattern of results.
Furthermore, across the two experiments, the manipulation of linguistic features of the attractor controlled the expression of the interference effect in terms of ERP componentry: manipulation of morphosyntax led to a modulation of the P component, while manipulation of plausibility led to an N effect, but the nature of the modulation was similar—an interaction drive by conditions of comparable cue status both the Different Voice Attractor and the Plausible Attractor partially matched their respective Invalid Cues across levels of linguistic analysis.
This fact in and of itself is novel in the literature—I am aware of no other study that reports a form of replicated effect expressed on two different ERP components [ 20 ] where interference effects were observed both a sustained negativity and a P effect were in the same experiment, although among different groups of participants].
Sentence Processing (Current Issues in the Psychology of Language)
The paucity of cases in the literature where two different components show similar interference effects may also stem from the fact that there are not many ERP studies that use a similar paradigm, nor many on ellipsis. I am not aware of another ERP study which manipulates morphosyntax or plausibility of syntactically illicit attractor representations during dependency resolution other than [ 19 , 20 ]. The expression of the effects may also have interesting implications for the functional interpretation of ERP components, but the current functional interpretations of the N and P components are predicated on experimental circumstances that are so distant from this ellipsis paradigm that I hesitate to speculate about componentry differences further.
The difference in elicited componentry can arguably be said to be a function of the level of linguistic analysis over which cues and the contents of memory interact. The distribution of the effects was frontal in both experiments, regardless of the different components that were elicited, and in contrast with most language-elicited P and N effects. In sentence processing experiments, frontal P effects have been associated with resolving who did what to whom in the face of multiple referents in a discourse [ 58 ] and with processing morphosyntactic agreement violations that have consequences for the possible reanalysis or repair of "who did what to whom" following the violation [ 61 ].
In addition to featuring morphosyntactic violations, the Ungrammatical conditions in Experiment 1 also featured a form of referential insufficiency or failure, whereby the cues to the antecedent were not diagnostic. This situation might be similar to the processing circumstances in [ 58 ] where multiple referents were available, but further computation was needed determine who did what to whom in the face of ambiguity. In recognition memory, frontal Ns have been elicited in paradigms that vary the familiarity of a stimulus [ 62 ], but these effects have been argued to reflect a form of conceptual priming that varies during the processing of old versus new stimuli, rather than familiarity per se [ 63 ].
In some sense, ellipsis calls on a previously processed "old" representation in a "new" way, but given that it is not clear that frontal Ns are functionally or even descriptively separate from more posteriorly distributed Ns [ 45 ]. The anteriority of the results in Experiment 2 may result from the partial match between an insufficient or ungrammatical cue and an attractor. The partial match might lead to sub-threshold activation of a representation in a similar way that judgments of familiarity can be said to be a lesser form of representational activation compared to full recognition or recollection.
Despite these caveats, the findings reported here can be interpreted such that they make two substantive claims about how cues are integrated during language comprehension. Most models of recognition memory operate on global matching of cues to memory at retrieval such that all available cues are utilized [ 64 ]. Since no special status is given as a function of location or temporal distribution, retrieval cues are likely to be combined dynamically as information processing goes forward, though the computational mechanisms that govern this process for language processing are only beginning to be investigated [ 5 , 11 , 13 ].