| 1. |
- Ayub Khan, Fahima, et al.
(författare)
-
Investigating code-switching and disfluencies in bilingual dialogue
- 2022
-
Ingår i: Proceedings of the 26th Workshop on the Semantics and Pragmatics of Dialogue - Poster Abstracts.
-
Konferensbidrag (refereegranskat)abstract
- This paper investigates the relationship between disfluency and code-switching in bilingual dialogue. We examine a corpus of 41 bilingual (Spanish-English) conversations and test the hypothesis that code-switching can be a response to negative evidence of grounding in the form of disfluencies. We find that there is a statistically significant relation between disfluencies and code-switching. Particularly, disfluencies have a positive effect on within turn code-switching.
|
|
| 2. |
- Dobnik, Simon, 1977, et al.
(författare)
-
In Search of Meaning and Its Representations for Computational Linguistics
- 2022
-
Ingår i: Proceedings of the 2022 CLASP Conference on (Dis)embodiment, Gothenburg and online 15–16 September 2022 / Simon Dobnik, Julian Grove and Asad Sayeed (eds.). - : Association for Computational Linguistics. - 2002-9764. - 9781955917674
-
Konferensbidrag (refereegranskat)abstract
- In this paper we examine different meaning representations that are commonly used in different natural language applications today and discuss their limits, both in terms of the aspects of the natural language meaning they are modelling and in terms of the aspects of the application for which they are used.
|
|
| 3. |
|
|
| 4. |
- Maraev, Vladislav, 1986, et al.
(författare)
-
Dialogue act classification is a laughing matter
- 2021
-
Ingår i: Proceedings of the 25th Workshop on the Semantics and Pragmatics of Dialogue, Potsdam & The Internet, 20–22 September 2021 / Ellen Breitholtz, Kallirroi Georgila and David Schlangen (eds.). - Potsdam : Semdial. - 2308-2275.
-
Konferensbidrag (refereegranskat)abstract
- In this paper we explore the role of laughter in attributing communicative intents to utterances, i.e. detecting the dialogue act performed by them. We conduct a corpus study in adult phone conversations showing how different dialogue acts are characterised by specific laughter patterns, from both the speaker and from the partner. Furthermore, we show that laughs can positively impact the performance of Transformer-based models in a dialogue act recognition task. Our results highlight the importance of laughter for meaning construction and disambiguation in interaction.
|
|
| 5. |
- Noble, Bill, et al.
(författare)
-
Classification Systems: Combining taxonomical and perceptual lexical meaning
- 2022
-
Ingår i: Proceedings of the 3rd Natural Logic Meets Machine Learning Workshop (NALOMA III), 8–12 August, 2022, Galway, Ireland / Aikaterini-Lida Kalouli, Stergios Chatzikyriakidis (Editors). - Stroudsburg, PA : Association for Computational Linguistics. - 9781959429395
-
Konferensbidrag (refereegranskat)abstract
- Lexical meaning includes both perceptual and logical aspects. We present a method for combining a taxonomy with perceptual classifiers, and show that in the few-shot setting, it out-performs other methods of injecting taxonomical information in image classification. We use this method to define witness conditions for types in a rich type system with probabilistic type judgments and suggest how such a type system can be used as the basis for a new type of hybrid NLU architecture.
|
|
| 6. |
- Noble, Bill, et al.
(författare)
-
Conditional Language Models for Community-Level Linugistic Variation
- 2022
-
Ingår i: Proceeding of the 2022 Workshop on NLP and Computational Social Science, November 7, 2022 / David Bamman, Dirk Hovy, David Jurgens, Katherine Keith, Brendan O'Connor, Svitlana Volkova (Editors). - Stroudsburg : Association for Computational Linguistics. - 9781959429203
-
Konferensbidrag (refereegranskat)abstract
- Community-level linguistic variation is a core concept in sociolinguistics. In this paper, we use conditioned neural language models to learn vector representations for 510 online communities. We use these representations to measure linguistic variation between communities and investigate the degree to which linguistic variation corresponds with social connections between communities. We find that our sociolinguistic embeddings are highly correlated with a social network-based representation that does not use any linguistic input.
|
|
| 7. |
- Noble, Bill, et al.
(författare)
-
Coordinating taxonomical and observational meaning: The case of genus-differentia definitions
- 2022
-
Ingår i: Proceedings of the 26th Workshop on the Semantics and Pragmatics of Dialogue - Full Papers. - Dublin, Ireland : SEMDIAL.
-
Konferensbidrag (refereegranskat)abstract
- Genus-differentia definitions exhibit the dual nature of lexical semantic meaning—they incorporate both “hard” X is a Y relations between words, as well as “soft” aspects of meaning which can be supported or challenged by observation. Modeling such definitions as contributions in dialogue requires that we accommodate the fluidity of linguistic resources, while respecting the dual nature of the relations that hold between lexical items. In this paper, we use a Probabilistic Type Theory with Records (ProbTTR) to characterise genus-differentia definitions by describing the update they license to the common ground of a dialogue.
|
|
| 8. |
- Noble, Bill, et al.
(författare)
-
Describe Me an Auklet: Generating Grounded Perceptual Category Descriptions
- 2023
-
Ingår i: Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, December 6-10, 2023, Singapore / Houda Bouamor, Juan Pino, Kalika Bali (Editors). - : Association for Computational Linguistics. - 9798891760608
-
Konferensbidrag (refereegranskat)abstract
- Human speakers can generate descriptions of perceptual concepts, abstracted from the instance-level. Moreover, such descriptions can be used by other speakers to learn provisional representations of those concepts. Learning and using abstract perceptual concepts is under-investigated in the language-and-vision field. The problem is also highly relevant to the field of representation learning in multi-modal NLP. In this paper, we introduce a framework for testing category-level perceptual grounding in multi-modal language models. In particular, we train separate neural networks to **generate** and **interpret** descriptions of visual categories. We measure the *communicative success* of the two models with the zero-shot classification performance of the interpretation model, which we argue is an indicator of perceptual grounding. Using this framework, we compare the performance of *prototype*- and *exemplar*-based representations. Finally, we show that communicative success exposes performance issues in the generation model, not captured by traditional intrinsic NLG evaluation metrics, and argue that these issues stem from a failure to properly ground language in vision at the category level.
|
|
| 9. |
- Noble, Bill, et al.
(författare)
-
Large-scale text pre-training helps with dialogue act recognition, but not without fine-tuning
- 2021
-
Ingår i: Proceedings of the 14th International Conference on Computational Semantics, June 17-18, 2021, Groningen, The Netherlands (Online). - Stroudsburg, PA : Association for Computational Linguistics. - 9781954085190
-
Konferensbidrag (refereegranskat)abstract
- We use dialogue act recognition (DAR) to investigate how well BERT represents utterances in dialogue, and how fine-tuning and large-scale pre-training contribute to its performance. We find that while both the standard BERT pre-training and pretraining on dialogue-like data are useful, task-specific fine-tuning is essential for good performance
|
|
| 10. |
|
|
