Semantic Role Research Articles

Positional encoding of morphemes in visual word recognition.

Reading morphologically complex words requires analysis of their morphemic subunits (e.g., play + er); however, the positional constraints of morphemic processing are still little understood. The current study involved three unprimed lexical decision experiments to directly compare the positional encoding of stems and affixes during reading and to investigate the role of semantics during the position encoding of morphemes. Experiment 1 revealed that transposed compound words were harder to reject than their controls (e.g., dreamday vs. shadeday), whereas there was no difference between transposed suffixed words and their controls (e.g., fulpain vs. adepain). Experiment 2 replicated the results for transposed compound words of the first experiment and further showed that there was no difference between transposed prefixed words and their controls (e.g., qualifydis vs. qualifymis). Experiment 3 investigated the role of semantic transparency in morpheme transposition effects and revealed a larger morpheme transposition effect for semantically transparent transposed compound words (e.g., cuptea vs. taptea) than for semantically opaque transposed compound words (e.g., linedead vs. deskdead). These results bring to light important differences in the positional encoding of stems and affixes, suggesting that prefixes and suffixes are recognized in a position-dependent manner compared to the position-independent encoding of embedded stems and that morpheme transposition effects are guided by semantics. The current findings call for more clearly specified theoretical models of visual word recognition that reflect the distinct positional constraints of stems and affixes, as well as the influence of semantics on morphological processing. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

On Emotional States and their Syntactic Representation

The study focuses on the analyses of syntactic structure of some predicates and state predicatives denoting positive emotion: veselya (to make someone happy, in a good moon), veselya se (to frolic, to party), vessel sam (to be happy), veselo (mi) e (to be in a good mood), veselba e (it is fun). Our aim is to provide a syntactic model representing argument structure of predicates and predicatives from this thematic group. The syntactic realization of arguments, semantic role hierarchy, and specifics concerning the referentiality of the nominal phrases in argument position are examined. The analyses show differences both in the semantic structure of predicates and predicatives and in the syntactic realization of their arguments. In contrast to other emotional predicates, veselya and veselya se have only nominal phrases as complements. While the analyzed verbs represent causative and dynamic predicates, respectively, the predicatives vessel, vesselo denote emotional states. Nominal phrases, as well as complement clauses can be used in complement positions. In this respect, the syntactic realization of predicatives is similar to the one of other emotional state predicatives, such as radosten sam (to be happy), radostno (mi) e (it is rejoicing).

Hybrid subjects in Spanish and Catalan: Halfway between Agents and Patients

We analyze an intransitive construction involving verbs like Spanish matarse ‘kill’ whose subjects appear to have both internal and external argument properties. Examples include Juan se mató en un accidente de coche ‘Juan got himself killed in a car accident’, in which the subject’s referent shows hybrid behavior between agent and patient as it needs to be engaged in an action leading to its accidental death. We propose that the subject’s internal and external argument properties can be accounted for if subjects can bear two semantic roles by virtue of being associated with more than one distinct head in the syntax (Pineda & Berro 2020). We argue that such intransitive uses involve a distinct argument structure from transitive reflexives despite sharing the same surface form, cf., El sospechoso del homicidio se mató al estar rodeado por la policía ‘The suspect killed himself when he was surrounded by the police’. The present account provides evidence that agents and external arguments do not always correlate since some verb classes can have identical surface form, despite involving underlyingly distinct argument alignment.

Frames and semantic roles in metaphorical mappings

Abstract Conceptual Metaphor Theory describes metaphorical mappings as correspondences between domains. Nevertheless, the interrelations of the mapped elements still need to be characterised. This study analyses metaphorical mappings from one situational frame in the cooking domain in American English and Peninsular Spanish. The aims are to elucidate to what extent each language primes particular semantic roles in the boiling frame for metaphorisation and explore the possible cultural implications of these cross-linguistic differences. The qualitative analysis reveals the contrastive relevance of semantic roles in frame mappings. In addition, the frequencies of boil and hervir metaphorical senses determine the degree of entrenchment and salience of metaphors cross-linguistically. The results suggest that the English boiling frame has a considerably broader scope as a source than the Spanish hervir frame. Additionally, the paper contrasts the relative saliency of the semantic roles mapped in those metaphors shared by both languages. Finally, cultural implications are discussed.

Semantic-Aware Contrastive Learning With Proposal Suppression for Video Semantic Role Grounding

Semantic-Aware Contrastive Learning With Proposal Suppression for Video Semantic Role Grounding

Meta-learning framework with updating information flow for enhancing inductive prediction

Inductive relation prediction, which handles unseen entities at the reasoning stage, has the potential to complete continuously expanding knowledge graphs. Existing inductive approaches predominantly focus on providing alternatives predicted by the subgraph structure of the target. However, many approaches neglect the correlation between semantic relationships, which does not mine additional implicit information. In this paper, we propose MUIF, a meta-learning framework with an updated information flow that addresses these challenges. Further, MUIF aims to enhance the role of relational semantics in inductive prediction, while emphasizing the importance of distinguishing relational paths. We propose an updated information flow mechanism for strengthening the relational pass, and we use a contrastive strategy to represent strongly associated paths. Further, meta-learning is applied to transfer the updated total embedding, thereby enhancing prediction accuracy. Further, the Hits@10 of MUIF can reach up to 97.03% in the link prediction experiments, thereby demonstrating the excellent performance of the proposed approach.

TDeLTA: A Light-Weight and Robust Table Detection Method Based on Learning Text Arrangement

The diversity of tables makes table detection a great challenge, leading to existing models becoming more tedious and complex. Despite achieving high performance, they often overfit to the table style in training set, and suffer from significant performance degradation when encountering out-of-distribution tables in other domains. To tackle this problem, we start from the essence of the table, which is a set of text arranged in rows and columns. Based on this, we propose a novel, light-weighted and robust Table Detection method based on Learning Text Arrangement, namely TDeLTA. TDeLTA takes the text blocks as input, and then models the arrangement of them with a sequential encoder and an attention module. To locate the tables precisely, we design a text-classification task, classifying the text blocks into 4 categories according to their semantic roles in the tables. Experiments are conducted on both the text blocks parsed from PDF and extracted by open-source OCR tools, respectively. Compared to several state-of-the-art methods, TDeLTA achieves competitive results with only 3.1M model parameters on the large-scale public datasets. Moreover, when faced with the cross-domain data under the 0-shot setting, TDeLTA outperforms baselines by a large margin of nearly 7%, which shows the strong robustness and transferability of the proposed model.

Frame Semantic Role Labeling Using Arbitrary-Order Conditional Random Fields

This paper presents an approach to frame semantic role labeling (FSRL), a task in natural language processing that identifies semantic roles within a text following the theory of frame semantics. Unlike previous approaches which do not adequately model correlations and interactions amongst arguments, we propose arbitrary-order conditional random fields (CRFs) that are capable of modeling full interaction amongst an arbitrary number of arguments of a given predicate. To achieve tractable representation and inference, we apply canonical polyadic decomposition to the arbitrary-order factor in our proposed CRF and utilize mean-field variational inference for approximate inference. We further unfold our iterative inference procedure into a recurrent neural network that is connected to our neural encoder and scorer, enabling end-to-end training and inference. Finally, we also improve our model with several techniques such as span-based scoring and decoding. Our experiments show that our approach achieves state-of-the-art performance in FSRL.

Video Event Extraction with Multi-View Interaction Knowledge Distillation

Video event extraction (VEE) aims to extract key events and generate the event arguments for their semantic roles from the video. Despite promising results have been achieved by existing methods, they still lack an elaborate learning strategy to adequately consider: (1) inter-object interaction, which reflects the relation between objects; (2) inter-modality interaction, which aligns the features from text and video modality. In this paper, we propose a Multi-view Interaction with knowledge Distillation (MID) framework to solve the above problems with the Knowledge Distillation (KD) mechanism. Specifically, we propose the self-Relational KD (self-RKD) to enhance the inter-object interaction, where the relation between objects is measured by distance metric, and the high-level relational knowledge from the deeper layer is taken as the guidance for boosting the shallow layer in the video encoder. Meanwhile, to improve the inter-modality interaction, the Layer-to-layer KD (LKD) is proposed, which integrates additional cross-modal supervisions (i.e., the results of cross-attention) with the textual supervising signal for training each transformer decoder layer. Extensive experiments show that without any additional parameters, MID achieves the state-of-the-art performance compared to other strong methods in VEE.

HiFormer: Hierarchical transformer for grounded situation recognition

The prevalence of monitoring video is critical to public safety, but existing Object Detection and Action Recognition models are overwhelmed by camera operators, unable to identify relevant events. In light of this, Grounding Situation Recognition (GSR) provides a practical solution to recognize the events in a surveillance video. That is, GSR can identify the noun entities (e.g., humans) and their actions (e.g., driving), and provide grounding frames for involved entities. Compared with Action Recognition and Object Detection, GSR is more in line with human cognitive habits, better allowing law enforcement agencies to understand the predictions. However, the crucial issue with most existing frameworks is the neglect of verb ambiguity, that is, superficially similar verbs but have distinct meanings (e.g. buying v.s. giving). Many existing works propose a two-stage model, which first blindly predicts the verb, and then uses this verb information to predict semantic roles. These frameworks ignore the importance of noun information during verb prediction, making them susceptible to misidentifications. To address this problem and better discern between ambiguous verbs, we propose HiFormer, a novel hierarchical transformer framework with direct and comprehensive consideration of similar verbs for each image, to more accurately identify the salient verb, semantic roles, and the grounding frames. Compared with the state-of-the-art models in Grounded Situation Recognition (SituFormer and CoFormer), HiFormer shows an advantage of over 35% and 20% on the Top-1 and Top-5 verb accuracy respectively, as well as 13% on the Top-1 Noun accuracy.

Information extraction tool text2alm: From narratives to action language system descriptions and query answering

In this work we design an information extraction tool text2alm capable of narrative understanding with a focus on action verbs. This tool uses an action language ALM to perform inferences on complex interactions of events described in narratives. The methodology used to implement the text2alm system was originally outlined by Lierler, Inclezan, and Gelfond (In IWCS 2017 – 12th International Conference on Computational Semantics – Short Papers (2017)) via a manual process of converting a narrative to an ALM model. We refine that theoretical methodology and utilize it in design of the text2alm system. This system relies on a conglomeration of resources and techniques from two distinct fields of artificial intelligence, namely, (i) knowledge representation and reasoning and (ii) natural language processing. The effectiveness of system text2alm is measured by its ability to correctly answer questions from the bAbI tasks published by Facebook Research in 2015. This tool matched or exceeded the performance of state-of-the-art machine learning methods in six of the seven tested tasks. We also illustrate that the text2alm approach generalizes to a broader spectrum of narratives. On the path to creating system text2alm, a semantic role labeler text2drs was designed. Its unique feature is the use of the elements of the fine grained linguistic ontology VerbNet as semantic roles/labels in annotating considered text. This paper provides an accurate account on the details behind the text2alm and text2drs systems.

Artificial language creation as a tool for assessing children’s metalinguistic abilities

The author aims to analyse two projects of the artificial language Uniling. They were created by students of years 5 to 11 within the linguistic framework of the "Formula de Integreco" training camp in June 2017 and January 2018. The research methods included structural analysis of the artificial language, as well as its creation process in comparison with Russian, i. e. the native language for all participants in the experiment. Research results: the specific conditions for artificial language development enable us to trace the stages of its accelerated "evolution" and complexity buildup at all levels. The latter becomes possible by shifting the initial focus away from observing the economy principle in language and raising randomly occurring exceptions to the status of a rule. Such rules are further extrapolated to other same-level units of the artificial language. This indicates that children attach importance to the concept of linguistic norm, as well as the possibility of clear grammatical differentiation when the grammatical meaning is secondary relative to the lexical-conceptual one. Children give no less significance to the inner form of words, which makes it possible to create lexemes employing the means of the artificial language itself without relying on the material of natural substrate languages. Moreover, the inner form helps to outline the systemic connections within the thematic groups. However, the hypernym/hyponym relations are obviously weak as generic concepts are not found in the Uniling lexis. At the same time, the aesthetics and euphony notions often come to the fore, which is reflected in the expression plane of artificial language lexemes and in the idea of the pragmatic potential of the language. This is most clearly seen at the syntax level due to the introduction of particular semantic role markers not typical of the Russian language (benefactives and malefactives). Conclusions: artificial language analysis enables us to identify the strengths and weaknesses of children’s metalinguistic representations, which can be considered in the process of teaching Russian at school.

К вопросу о синтаксическом статусе субстантивов в русских предложениях тождества

A multi aspect study of bisubstantive sentences of characterizing semantics is necessary to solve complex problems of syntactic analysis. The grammatical specificity of bisubstantial sentences lies in the morphological identity of the predicative minimum. The study has theoretical and practical tasks: identify the structural, semantic and communicative organization of sentences, propose methodological recommendations for syntactic analysis of sentences. The material of the study is modern literature. Analytical and descriptive methods are used to analyze the linguistic material. The problems of teaching syntactic analysis of sentences are identified through experimental methods. The main conclusions of the study: syntactic analysis of bisubstantial sentences should consist of traditional structural and semantic parameters, roles and functions in the semantic structure and communicative organization of a sentence, take into account context, stylistic and pragmatic factors. Syntactic phenomena should be studied from the functional position at all stages and levels of teaching syntax of the modern Russian language.

Speaker Attribution in German Parliamentary Debates with QLoRA-adapted Large Language Models

The growing body of political texts opens up new opportunities for rich insights into political dynamics and ideologies but also increases the workload for manual analysis. Automated speaker attribution, which detects who said what to whom in a speech event and is closely related to semantic role labeling, is an important processing step for computational text analysis. We study the potential of the large language model family Llama 2 to automate speaker attribution in German parliamentary debates from 2017-2021. We fine-tune Llama 2 with QLoRA, an efficient training strategy, and observe our approach to achieve competitive performance in the GermEval 2023 Shared Task On Speaker Attribution in German News Articles and Parliamentary Debates. Our results shed light on the capabilities of large language models in automating speaker attribution, revealing a promising avenue for computational analysis of political discourse and the development of semantic role labeling systems.

Influence of social and semantic context in processing speech in noise

Social interactions occupy a substantial part of our life, and listening to others’ interactions is critical in understanding our social world. Although the role of semantics in speech comprehension has been studied, the role of social context, and its interaction with semantics, remain unknown. We conducted a series of four perceptual experiments to better understand the processing of multiple-speaker conversations from a third-person viewpoint, manipulating the social and semantic context of a conversation. We used a stimulus set consisting of two-speaker dialogues or one-speaker monologues (factor: social context) arranged in intact or sentence-scrambled order (factor: semantic context). Each stimuli comprised five sentences, with the fifth sentence embedded in multi-talker babbling noise. This fifth sentence was subsequently repeated without noise, with a single word altered or unchanged. Stimuli were presented to healthy young adults, asked whether the repeated sentence was same as or different from previous in-noise sentence. We found significant effects for both social and semantic contexts when processing a conversation. Our findings highlight that both semantic and social aspects of a conversation can modulate the processing of conversations. These results raise new questions regarding predictive or other mechanisms that may be at play when perceiving speech in social contexts.

In search of an optimal method for analyzing deep structures: frame semantics and classification of argumentative structures

The subject of the research is the search for the optimal method of analyzing deep structures using frame semantics. The study of semantic roles, similarities and differences in the approaches of both C. Fillmore and B. Levin – M. R. Hovav make it possible to analyze the structure of a sentence in more detail and accurately, identify deep cases and determine semantic relations between words. The study of these aspects is key to understanding language constructs and their interpretation. The study of various approaches makes it possible to identify both common features and unique features, which is key for a complete understanding of language constructions. An interest in text analysis in the field of artificial intelligence, machine learning and computational linguistics, and an understanding of the semantic relationships between words will help create more accurate and efficient text processing algorithms. One of the research methods is the semantic analysis of sentences based on corpus data. This method includes the study of various linguistic constructions in the context of their use in real texts, which allows us to identify common patterns and rules for the use of these linguistic units in different situations. The scientific novelty of the study lies in the fact that the authors have determined the similarity of the approaches on how to understand the surface and deep structures of language of Ch. Fillmore and B. Levin and M. Rappaport. Their work, despite differences in methodology and terminology, together allow for in-depth investigation of the relationship between the meanings of verbs and the structure of arguments. As a result of the study, the natural relationships between deep cases and semantic roles in sentences of various types are revealed, and key points that need to be taken into account when analyzing deep structures for a more accurate definition of the semantic roles of arguments are highlighted: frame semantics and thematic grids. Disagreements and alternative points of view contribute to the constant development and improvement of linguistic theories. Such debates eventually lead to a deeper understanding of the implementation of the arguments and open up opportunities for further research in this area. Both C. Fillmore and B. Levin and M. Rappaport have made significant contributions to understanding the surface and deep structures of language, although their approaches and terminology may differ.

How Is a “Kitchen Chair” like a “Farm Horse”? Exploring the Representation of Noun-Noun Compound Semantics in Transformer-based Language Models

Abstract Despite the success of Transformer-based language models in a wide variety of natural language processing tasks, our understanding of how these models process a given input in order to represent task-relevant information remains incomplete. In this work, we focus on semantic composition and examine how Transformer-based language models represent semantic information related to the meaning of English noun-noun compounds. We probe Transformer-based language models for their knowledge of the thematic relations that link the head nouns and modifier words of compounds (e.g., KITCHEN CHAIR: a chair located in a kitchen). Firstly, using a dataset featuring groups of compounds with shared lexical or semantic features, we find that token representations of six Transformer-based language models distinguish between pairs of compounds based on whether they use the same thematic relation. Secondly, we utilize fine-grained vector representations of compound semantics derived from human annotations, and find that token vectors from several models elicit a strong signal of the semantic relations used in the compounds. In a novel “compositional probe” setting, where we compare the semantic relation signal in mean-pooled token vectors of compounds to mean-pooled token vectors when the two constituent words appear in separate sentences, we find that the Transformer-based language models that best represent the semantics of noun-noun compounds also do so substantially better than in the control condition where the two constituent works are processed separately. Overall, our results shed light on the ability of Transformer-based language models to support compositional semantic processes in representing the meaning of noun-noun compounds.

Fictional worlds of fantasy and "magical" prose as a space for generating supertext

In this paper, the author addresses the features of the supertexts based on the modern Russian novels, namely postmodern works, the genre of which can be defined as fantasy. One of the areas of research is the interpretation of the author's myth, as well as understanding the role of myth in literature. The subject of the study is an end-to-end code that unites the prosaic macrocycles of modern authors as a supertext unity. The object of the study is the macrotext as a format of a literary work in the literature of modern Russian writers. The authors consider in detail such aspects of the topic as the cross-cutting theme of an alien / adopted / unrecognized / illegitimate child and its connection with the genus and family, and the implementation of this theme in novels combined into trilogies or macrocycles. Both general scientific research methods and specialized philological ones were used in the article, namely: the method of qualitative content analysis and the narrative method, the methods of idealization and deduction are secondary methods of working with the material. The leading method was the interpretation of literary texts, which were the source of the parktic research material. The main function of the myth is to reflect a certain image of reality, being an add-on above it, containing specific semantic models and roles distributed among the characters, and also an expression of the author's position. The author's myth can be described as a playful, mosaic, inverse construct. The conclusions presented in the article can be used in further analysis of the work of modern writers, as well as when working with the concept of "author's myth" and studying the transformation of myth (both archaic and social, cultural, etc.) in modern postmodern literature. A special contribution of the authors to the research of the topic is the consideration of this motif in its diversified interpretation in various works and the connection of this motif with the onomastic code of works as one of the principles of the formation of a supertext.

Features of Emotion Verbalization

The presented article is devoted to the study of the lexical and semantic role of emotions. Language serves not only to convey information, but also to express the inner world of the speaker. Emotions are multifaceted. Impressions in themselves are not considered linguistic phenomena, but learning the language of a speaking subject is one of the approaches to understanding human emotions. The study of the emotional world of a person involves the researcher’s appeal to a number of humanities: psychology, sociology, linguistics, philosophy. Cognitive science makes it possible to combine various theoretical platforms for a comprehensive study of this complex, multidimensional phenomenon, since a characteristic feature of modern linguistics is the search for linguistic modeling of a person in a close objectively existing relationship with other related human sciences.

The Computational Method for Supporting Thai VerbNet Construction

VerbNet is a lexical resource for verbs that has many applications in natural language processing tasks, especially ones that require information about both the syntactic behavior and the semantics of verbs. This article presents an attempt to construct the first version of a Thai VerbNet corpus via data enrichment of the existing lexical resource. This corpus contains the annotation at both the syntactic and semantic levels, where verbs are tagged with frames within the verb class hierarchy and their arguments are labeled with the semantic role. We discuss the technical aspect of the construction process of Thai VerbNet and survey different semantic role labeling methods to make this process fully automatic. We also investigate the linguistic aspect of the computed verb classes and the results show the potential in assisting semantic classification and analysis. At the current stage, we have built the verb class hierarchy consisting of 28 verb classes from 112 unique concept frames over 490 unique verbs using our association rule learning method on Thai verbs.