Linguistically-Informed Specificity and Semantic Plausibility for Dialogue Generation

2019 ◽  
Author(s):  
Wei-Jen Ko ◽  
Greg Durrett ◽  
Junyi Jessy Li
Keyword(s):  
Semantic Plausibility

scholarly journals The Language Network is Recruited but Not Required for Nonverbal Event Semantics

2021 ◽  
pp. 1-26
Author(s):  
Anna A. Ivanova ◽  
Zachary Mineroff ◽  
Vitor Zimmerer ◽  
Nancy Kanwisher ◽  
Rosemary Varley ◽  
...  
Keyword(s):  
Semantic Processing ◽  
Judgment Task ◽  
Semantic Task ◽  
Line Drawings ◽  
Patient Interactions ◽  
Language Difficulties ◽  
Individual Concepts ◽  
Semantic Plausibility ◽  
Language Areas ◽  
Language Network

The ability to combine individual concepts of objects, properties, and actions into complex representations of the world is often associated with language. Yet combinatorial event-level representations can also be constructed from nonverbal input, such as visual scenes. Here, we test whether the language network in the human brain is involved in and necessary for semantic processing of events presented nonverbally. In Experiment 1, we scanned participants with fMRI while they performed a semantic plausibility judgment task versus a difficult perceptual control task on sentences and line drawings that describe/depict simple agent–patient interactions. We found that the language network responded robustly during the semantic task performed on both sentences and pictures (although its response to sentences was stronger). Thus, language regions in healthy adults are engaged during a semantic task performed on pictorial depictions of events. But is this engagement necessary? In Experiment 2, we tested two individuals with global aphasia, who have sustained massive damage to perisylvian language areas and display severe language difficulties, against a group of age-matched control participants. Individuals with aphasia were severely impaired on the task of matching sentences to pictures. However, they performed close to controls in assessing the plausibility of pictorial depictions of agent–patient interactions. Overall, our results indicate that the left frontotemporal language network is recruited but not necessary for semantic processing of nonverbally presented events.

scholarly journals Effects of semantic plausibility, syntactic complexity and n-gram frequency on children's sentence repetition

2020 ◽  
pp. 1-25
Author(s):  
Kamila POLIŠENSKÁ ◽  
Shula CHIAT ◽  
Jakub SZEWCZYK ◽  
Katherine E. TWOMEY
Keyword(s):  
Language Processing ◽  
Sentence Processing ◽  
Age Groups ◽  
Linguistic Knowledge ◽  
Syntactic Complexity ◽  
Sentence Repetition ◽  
N Gram ◽  
Semantic Plausibility ◽  
Post Hoc

Abstract Theories of language processing differ with respect to the role of abstract syntax and semantics vs surface-level lexical co-occurrence (n-gram) frequency. The contribution of each of these factors has been demonstrated in previous studies of children and adults, but none have investigated them jointly. This study evaluated the role of all three factors in a sentence repetition task performed by children aged 4–7 and 11–12 years. It was found that semantic plausibility benefitted performance in both age groups; syntactic complexity disadvantaged the younger group but benefitted the older group; while contrary to previous findings, n-gram frequency did not facilitate, and in a post-hoc analysis even hampered, performance. This new evidence suggests that n-gram frequency effects might be restricted to the highly constrained and frequent n-grams used in previous investigations, and that semantics and morphosyntax play a more powerful role than n-gram frequency, supporting the role of abstract linguistic knowledge in children's sentence processing.

scholarly journals When Structure Competes with Semantics: Reading Chinese Relative Clauses

2018 ◽  
Vol 4 (1) ◽  
Author(s):  
Peiyun Zhou ◽  
Yun Yao ◽  
Kiel Christianson
Keyword(s):  
Relative Clause ◽  
Relative Clauses ◽  
Role Reversal ◽  
Dominant Factor ◽  
Clause Structure ◽  
Thematic Role ◽  
Online Processing ◽  
Reading Chinese ◽  
Semantic Plausibility ◽  
Object Relative

An ongoing debate in Chinese psycholinguistics is whether subject-relative clauses or object-relative clauses are more difficult to process. The current study asks what happens when structure and plausibility are pitted against each other in Chinese relative clause processing. Chinese relative clause structures and semantic plausibility were manipulated to create both plausible and implausible versions of subject- and object-relative clauses. This method has been used in other languages (e.g., English) to elicit thematic role reversal comprehension errors. Importantly, these errors—as well as online processing difficulties—are especially frequent in implausible versions of dispreferred (noncanoncial) structures. If one relative clause structure in Chinese is highly dispreferred, the structural factor and plausibility factor should interact additively. If, however, the structures are relatively equally difficult to process, then there should be only a main effect of plausibility. Sentence reading times as well as analyses on lexical interest areas revealed that Chinese readers used plausibility information almost exclusively when reading the sentences. Relative clause structure had no online effect and small but consistent offline effects. Taken together, the results support a slight preference in offline comprehension for Chinese subject-relative clauses, as well as a central role for semantic plausibility, which appears to be the dominant factor in online processing and a strong determinant of offline comprehension.

scholarly journals A Probabilistic Model of Semantic Plausibility in Sentence Processing

2009 ◽  
Vol 33 (5) ◽  
pp. 794-838 ◽  
Author(s):  
Ulrike Padó ◽  
Matthew W. Crocker ◽  
Frank Keller
Keyword(s):  
Sentence Processing ◽  
Probabilistic Model ◽  
Semantic Plausibility

Plausibility and subcategorization preference in children's processing of temporarily ambiguous sentences: Evidence from self-paced reading

2002 ◽  
Vol 55 (1) ◽  
pp. 75-96 ◽  
Author(s):  
Matthew J. Traxler
Keyword(s):  
Subordinate Clause ◽  
Semantic Plausibility ◽  
Age Range ◽  
Matrix Verb

Three self-paced reading experiments investigated children's processing of temporarily ambiguous sentences. Across the three experiments, subcategorization preference of a verb in a subordinate clause and the semantic plausibility of the misanalysis were manipulated. Reading times in the temporarily ambiguous region and following syntactic disambiguation indicated that children in the age range tested (8 years, 11 months to 12 years, 11 months) routinely misanalyse sentences of the type tested, and their tendency to misanalyse the sentences does not depend on the subcategorization preferences of the initial verb. Additional correlational analyses suggested that subcategory information did affect the degree of difficulty that readers experienced processing the critical noun and matrix verb.

scholarly journals The language network is recruited but not required for non-verbal event semantics

2019 ◽  
Author(s):  
Anna A. Ivanova ◽  
Zachary Mineroff ◽  
Vitor Zimmerer ◽  
Nancy Kanwisher ◽  
Rosemary Varley ◽  
...  
Keyword(s):  
Semantic Processing ◽  
Judgment Task ◽  
Control Task ◽  
Semantic Task ◽  
Line Drawings ◽  
Patient Interactions ◽  
Perceptual Control ◽  
Semantic Plausibility ◽  
Temporal Language ◽  
Language Network

AbstractThe ability to combine individual meanings into complex representations of the world is often associated with language. Yet people also construct combinatorial event-level representations from non-linguistic input, e.g. from visual scenes. Here, we test whether the language network in the human brain is involved in and necessary for semantic processing of nonverbal events. In Experiment 1, we scanned participants with fMRI while they performed a semantic plausibility judgment task vs. a difficult perceptual control task on sentences and line drawings that describe/depict simple agent-patient interactions. We found that the language network responded robustly during the semantic task but not during the perceptual control task. This effect was observed for both sentences and pictures (although the response to sentences was stronger). Thus, language regions in healthy adults are engaged during a semantic task performed on pictorial depictions of events. But is this engagement necessary? In Experiment 2, we tested two individuals with global aphasia, who have sustained massive damage to perisylvian language areas and display severe language difficulties, against a group of age-matched control participants. Individuals with aphasia were severely impaired on a task of matching sentences and pictures. However, they performed close to controls in assessing the plausibility of pictorial depictions of agent-patient interactions. Overall, our results indicate that the left fronto-temporal language network is recruited but not necessary for semantic processing of nonverbal events.

The role of semantic plausibility in pronoun comprehension

2010 ◽  
Author(s):  
Rachel H. Messer ◽  
Shelia M. Kennison
Keyword(s):  
Semantic Plausibility

scholarly journals Sentence Recall in Latent and Anomic Aphasia: An Exploratory Study of Semantics and Syntax

2021 ◽  
Vol 11 (2) ◽  
pp. 230
Author(s):  
Christos Salis ◽  
Nadine Martin ◽  
Laura Reinert
Keyword(s):  
Real Time ◽  
Short Term Memory ◽  
Recall Performance ◽  
Syntactic Complexity ◽  
Control Groups ◽  
Short Term ◽  
Sentence Recall ◽  
Semantic Plausibility ◽  
Verbatim Recall ◽  
And Control

We investigated whether semantic plausibility and syntactic complexity affect immediate sentence recall in people with latent and anomic aphasia. To date, these factors have not been explored in these types of aphasia. As with previous studies of sentence recall, we measured accuracy of verbatim recall and uniquely real-time speech measures. The results showed that accuracy did not distinguish performance between latent aphasia and neurotypical controls. However, some of the real-time speech measures distinguished performance between people with latent aphasia and neurotypical controls. There was some evidence, though not pervasive, that semantic plausibility and syntactic complexity influenced recall performance. There were no interactions between semantic plausibility and syntactic complexity. The speed of preparation of responses was slower in latent aphasia than controls; it was also slower in anomic aphasia than both latent and control groups. It appears that processing speed as indexed by temporal speech measures may be differentially compromised in latent and anomic aphasia. However, semantic plausibility and syntactic complexity did not show clear patterns of performance among the groups. Notwithstanding the absence of interactions, we advance an explanation based on conceptual short-term memory as to why semantically implausible sentences are typically more erroneous and possibly also slower in recall.

Tone 3 in Pekinese

1967 ◽  
Vol 10 (3) ◽  
pp. 629-636 ◽  
Author(s):  
William S-Y. Wang ◽  
Kung-Pu Li
Keyword(s):  
Test Item ◽  
Native Speaker ◽  
Random Order ◽  
Tone Sequence ◽  
The Other ◽  
Pitch Contour ◽  
Tone Sandhi ◽  
Test Items ◽  
Phonological Features ◽  
Semantic Plausibility

This article treats the problem of tone sandhi in Pekinese phonology in terms of a perception experiment. The various shapes of tone 3 in different environments are illustrated by the use of spectrograms. The issue is whether the tone sequence 3–3 is homophonous with the sequence 2–3. There were 130 pairs of test items. The two members of each pair share the same phonological features except that of pitch contour. In other words, one member carries the tone sequence 2–3 while the other carries 3–3. Each test item was given twice, thus yielding a total of 520 items arranged in random order on a reading list from which tape recordings were made. Upon hearing an item on the tape, a subject is required to identify from each pair of 2–3 and 3–3 test items the member he hears. Should there be a phonemic distinction between the two types of tone sequences, a native speaker should have no difficulty in perceiving them as distinct. Then, it would be expected that the subjects could identify correctly in most cases, and that their responses should be reinforced by other phonetic properties and unbiased by semantic plausibility. However, the results of this experiment led to the opposite conclusion.

INCIDENTAL EXPOSURE AND L3 LEARNING OF MORPHOSYNTAX

2014 ◽  
Vol 36 (4) ◽  
pp. 611-645 ◽  
Author(s):  
Sarah Grey ◽  
John N. Williams ◽  
Patrick Rebuschat
Keyword(s):  
Word Order ◽  
Learning Effect ◽  
Judgment Task ◽  
Exposure Condition ◽  
Delayed Effects ◽  
Case Marking ◽  
Rule Knowledge ◽  
Incidental Exposure ◽  
Semantic Plausibility ◽  
Acceptability Judgment

Evidence of learning following incidental exposure has been found for aspects of nonnative syntax in adults (Rebuschat & Williams, 2006, 2012; Williams & Kuribara, 2008). However, little research has tested delayed effects of learning under an incidental condition or moved beyond word order. This study investigated learning of third language (L3) morphosyntax (word order and case marking) under an incidental exposure condition. Participants were second language Spanish learners exposed auditorily to a semiartificial language, Japlish, during a semantic plausibility judgment task. Performance following exposure to L3 Japlish was assessed with acceptability judgment and picture-matching tasks that were administered immediately after exposure and 2 weeks later. Results on immediate tests showed a significant learning effect on acceptability judgment but none for picture matching. At delayed testing, results showed maintenance of the learning effect on acceptability judgment and significantly improved performance on picture matching. Results additionally showed that many of the learners arrived at the correct Japlish word-order rule; fewer did so for case marking. Follow-up analyses indicated that this rule knowledge, as measured by debriefing information, was important for performance on both assessment tasks, particularly at delayed testing. The results suggest that learning of L3 structures can take place following incidental exposure, that such learning is durable and undergoes improvement, and that development of reportable rule knowledge may also play an important role.

Sign in / Sign up

Export Citation Format

Share Document