Semantic Argument Structure Research Articles

Dependency and Span, Cross-Style Semantic Role Labeling on PropBank and NomBank

The latest developments in neural semantic role labeling (SRL) have shown great performance improvements with both the dependency and span formalism/styles. Although the two styles share many similarities in linguistic meaning and computation, most previous studies focus on a single style. In this article, we define a new cross-style semantic role label convention and propose a new cross-style joint optimization model designed around the most basic linguistic meaning of a semantic role. Our work provides a solution to make the results of the two styles more comparable and allowing both formalisms of SRL to benefit from their natural connections in both linguistics and computation. Our model learns a general semantic argument structure and is capable of outputting in either style. Additionally, we propose a syntax-aided method to uniformly enhance the learning of both dependency and span representations. Experiments show that the proposed methods are effective on both span and dependency SRL benchmarks.

Towards patent text analysis based on semantic role labelling

Mining patent texts can obtain valuable technical information and competitive intelligence which is important for the development of technology and business. The current patent text mining approaches suffer from lack of effective, automatic, accurate and wide-coverage techniques that can annotate natural language texts with semantic argument structure. It is helpful for text mining to derive more meaningful semantic relationship from semantic role labelling (SRL) results of patents. This paper uses Word2Vec to learn word real-valued vector and design features related to word vector to train SRL parser. Based on the SRL parser, two patent text mining methods are then given: patent topic extraction and automatic construction of patent technical effect matrix (PTEM). Experiments show that semantic role labelling help achieve satisfactory results and saves manpower.

Towards patent text analysis based on semantic role labelling

Mining patent texts can obtain valuable technical information and competitive intelligence which is important for the development of technology and business. The current patent text mining approaches suffer from lack of effective, automatic, accurate and wide-coverage techniques that can annotate natural language texts with semantic argument structure. It is helpful for text mining to derive more meaningful semantic relationship from semantic role labelling (SRL) results of patents. This paper uses Word2Vec to learn word real-valued vector and design features related to word vector to train SRL parser. Based on the SRL parser, two patent text mining methods are then given: patent topic extraction and automatic construction of patent technical effect matrix (PTEM). Experiments show that semantic role labelling help achieve satisfactory results and saves manpower.

Effects of syntactic and semantic argument structure on sentence repetition in agrammatism: Things we can learn from particles and prepositions

Background: Sentence production impairment in aphasia has been attributed to several possible sources that are not mutually exclusive. Linguistic accounts often attribute the difficulty to the complexity of a verb's syntactic and/or semantic argument structure. Cognitive processing accounts emphasise the reduced processing capacity observed in agrammatic aphasia, which in turn has been attributed to reduced semantic short-term memory (STM) or slowed processing. Aims: In this study we used verb particles and prepositions to investigate effects of differences in syntactic and semantic argument structure on sentence repetition in aphasia. We predicted that verb particles and sentences containing verb-particle constructions would be easier to repeat than prepositions and prepositional transitive sentences, as the former have a less-complex semantic and syntactic argument structure than the latter. Also, semantic and phonological spans were assessed to determine if a reduction in either capacity correlates with repetition ability. Methods & Procedures: Participants were eight individuals with chronic aphasia. The experimental task was repetition of transitive sentences balanced for length and lexical content containing either verb particles or prepositional object structures. Accuracy of sentence repetition and repetition of verb particles and prepositions within sentences was examined. We calculated the effect of structural complexity on the sentence repetition task as the difference between proportion correct of verb-particle constructions and prepositional transitives. Semantic and phonological STM spans and word spans were also assessed and correlated with this measure of the structural complexity effect on sentence repetition. Outcomes & Results: Verb-particle sentences were repeated correctly significantly more often than prepositional transitive sentences, and within those sentences verbal particles were repeated correctly significantly more often than prepositions. The effect was strongly associated with fluency scores: it was present in participants with low fluency scores, but not in those with high fluency scores. The phonological, but not the semantic, STM probe span measure correlated with both the difference in accurate repetition of verb-particle and prepositional transitive sentences and the particles and prepositions within those sentences. Conclusions: Results indicate that differences in argument structure of particle and preposition constructions influence sentence repetition in agrammatic aphasia. The finding that lower fluency scores are associated with poorer performance on more complex structures suggests that this effect is associated with agrammatism. The impact of these structural distinctions between particles and prepositions should be taken into account during development of treatment stimuli for those with agrammatism.

Electrophysiological correlates of complement coercion.

This study examined the electrophysiological correlates of complement coercion. ERPs were measured as participants read and made acceptability judgments about plausible coerced sentences, plausible noncoerced sentences, and highly implausible animacy-violated sentences ("The journalist began/wrote/astonished the article before his coffee break"). Relative to noncoerced complement nouns, the coerced nouns evoked an N400 effect. This effect was not modulated by the number of possible activities implied by the coerced nouns (e.g., began reading the article; began writing the article) and did not differ either in magnitude or scalp distribution from the N400 effect evoked by the animacy-violated complement nouns. We suggest that the N400 modulation to both coerced and animacy-violated complement nouns reflected different types of mismatches between the semantic restrictions of the verb and the semantic properties of the incoming complement noun. This is consistent with models holding that a verb's semantic argument structure is represented and stored at a distinct level from its syntactic argument structure. Unlike the coerced complement noun, the animacy-violated nouns also evoked a robust P600 effect, which may have been triggered by the judgments of the highly implausible (syntactically determined) meanings of the animacy-violated propositions. No additional ERP effects were seen in the coerced sentences until the sentence-final word that, relative to sentence-final words in the noncoerced sentences, evoked a sustained anteriorly distributed positivity. We suggest that this effect reflected delayed attempts to retrieve the specific event(s) implied by coerced complement nouns.

Production and Perception from Perspectice of Focus

This paper investigates the effect of semantic argument structure on the comprehension and production of sentences by observing the prosodic realizations of English secondary predications. Specifically, the goal of this study is to show how the theory of predication, argument structure, and focus semantically interact to account for similarities and differences between English resultative and depictive predications. To address this issue, production and comprehension tests were performed. In the fried focus domain (verb phrase), subjects were asked to utter and to comprehend ambiguous sentences in the context monologues. The experimental results were generally consistent with general linguistic analyses: In the resultative constructions, secondary subject NPs tend to be accented, as in other argument-head constructions, while in the depictive constructions, secondary predicates tend to have accents, as in other adjunct-head constructions. (Dongduk Women’s University)

Incorporation in Universal Grammar: a case study in theoretical reductionism

Over the past three decades there has been a pervasive tendency in linguistic theorizing, especially in syntactic theory, towards theoretical reductionism; that is, theorists have attempted either to reduce complex phenomena to a single unitary phenomenon or to explain phenomena in terms of a small number of unifying principles of some kind. Reduction of complex phenomena or principles to simpler, better understood ones is a natural part of scientific explanation, but the danger in this course lies in 'overreduction' to the point where the phenomena or the principles become distorted; determining the correct reduction is an important consideration in explanatory scientific theories. The reduction of linguistic phenomena can be seen first in the reduction of the scope of linguistics from language to grammar (Chomsky, I 98 i) and, second, in the assumption of the autonomy of syntax. Complex semantic and pragmatic features of natural language are either excluded from the stipulated domain of linguistics or are treated as syntactic. A good example of this is the postulation of D-structure in Government-Binding (GB) theory and the initial stratum in Relational Grammar (RG) as a 'thematically pure' representation of the semantic argument structure of a verb. That is, semantic role distinctions are represented formally in syntactic structure by either structural configurations or grammatical relations, and rules affecting these arguments apply to their syntactic representations, thereby avoiding any direct reference to their semantic representations, which are often dispensed with altogether. The opposite of this is the Generative Semantics position, which maintained that

Argument structure frames: A lexical complexity metric?

The number of semantic argument structure frames associated with a verb has been reported to influence ease of processing during language comprehension. The present experiments tested the generality of the argument structure complexity effect with three dependent measures: eye-fixation times, naming latencies, and lexical decision latencies. Two eye-movement experiments and two experiments using cross-modal tasks failed to provide evidence supporting the argument structure complexity effect. The present experiments indicated that results reflecting verbs' argument structure complexity are not generalizable.