대화함축(Conversational implicature)에 대한 연구

The detection of hate speech in social media is a crucial task. The uncontrolled spread of hate has the potential to gravely damage our society, and severely harm marginalized people or groups. A major arena for spreading hate speech online is social media. This significantly contributes to the difficulty of automatic detection, as social media posts include paralinguistic signals (e.g. emoticons, and hashtags), and their linguistic content contains plenty of poorly written text. Another difficulty is presented by the context-dependent nature of the task, and the lack of consensus on what constitutes as hate speech, which makes the task difficult even for humans. This makes the task of creating large labeled corpora difficult, and resource consuming. The problem posed by ungrammatical text has been largely mitigated by the recent emergence of deep neural network (DNN) architectures that have the capacity to efficiently learn various features. For this reason, we proposed a deep natural language processing (NLP) model—combining convolutional and recurrent layers—for the automatic detection of hate speech in social media data. We have applied our model on the HASOC2019 corpus, and attained a macro F1 score of 0.63 in hate speech detection on the test set of HASOC. The capacity of DNNs for efficient learning, however, also means an increased risk of overfitting. Particularly, with limited training data available (as was the case for HASOC). For this reason, we investigated different methods for expanding resources used. We have explored various opportunities, such as leveraging unlabeled data, similarly labeled corpora, as well as the use of novel models. Our results showed that by doing so, it was possible to significantly increase the classification score attained.

/pdf/challenges-of-hate-speech-detection-in-social-media-4y0det7zh1.pdf

Challenges of Hate Speech Detection in Social Media

Slurring is a kind of hate speech that has various effects. Notable among these is variable offence. Slurs vary in offence across words, uses, and the reactions of audience members. Patterns of offence aren’t adequately explained by current theories. We propose an explanation based on the unjust power imbalance that a slur seeks to achieve. Our starting observation is that in discourse participants take on discourse roles. These are typically inherited from social roles, but only exist during a discourse. A slurring act is a speech-act that alters the discourse roles of the target and speaker. By assigning discourse roles the speaker unjustly changes the power balance in the dialogue. This has a variety of effects on the target and audience. We show how these notions explain all three types of offence variation. We also briefly sketch how a role and power theory can help explain silencing and appropriation. Explanatory power lies in the fact that offence is correlated with the perceived unjustness of the power imbalance created.

/pdf/slurs-roles-and-power-1t6v3bm3at.pdf

Slurs, roles and power

Understanding figures of speech: Dependency relations and organizational patterns

The ironic operation: Revisiting the components of ironic meaning

Two rival accounts of irony claim, respectively, that pretence and echo are independently sufficient to explain central cases. After highlighting the strengths and weaknesses of these accounts, I argue that an account in which both pretence and echo play an essential role better explains these cases and serves to explain peripheral cases as well. I distinguish between “weak” and “strong” hybrid theories, and advocate an “integrated strong hybrid” account in which elements of both pretence and echo are seen as complementary in a unified mechanism. I argue that the allegedly mutually exclusive elements of pretence and echo are in fact complementary aspects enriching a core-structure as follows: by pretending to have a perspective/thought F, an ironic speaker U echoes a perspective/thought G. F is merely pretended, perhaps caricaturised or exaggerated, while G is real/possible.

/pdf/pretence-and-echo-towards-an-integrated-account-of-verbal-371b1296nu.pdf

Pretence and Echo: Towards an Integrated Account of Verbal Irony

Compound figures are a rich, and under-explored area for tackling fundamental issues in philosophy of language. This paper explores new ideas about how to explain some features of such figures. We start with an observation from Stern (2000) that in ironic-metaphor, metaphor is logically prior to irony in the structure of what is communicated. Call this thesis Logical-MPT. We argue that a speech-act-based explanation of Logical-MPT is to be preferred to a content-based explanation. To create this explanation we draw on Barker’s (2004) expressivist speech-act theory, in which speech-acts build on other speech-acts to achieve the desired communicative effects. In particular, we show how Barker’s general ideas explain metaphor as an assertive-act, and irony as a ridiculing-act. We use Barker’s notion of proto-illocutionary-acts to show how metaphorical-acts and ironic-acts can build one on the other. Finally, we show that while an ironic-act can build on a metaphorical-act, a metaphorical-act cannot build on an ironic-act. This restriction on how they can be composed establishes Logical-MPT via a different route.

Compound figures: priority and speech-act structure

This paper argues that we need to re-think the semantics/pragmatics distinction in the light of new evidence from embedding of irony. This raises a new version of the old problem of ‘embedded impli...

/pdf/embedding-irony-and-the-semantics-pragmatics-distinction-1e0xl7t9be.pdf

Embedding Irony, and the Semantics/Pragmatics Distinction

https://philpapers.org/archive/POPGFU.pdf

Mihaela Popa-Wyatt

Papers

Slurs, roles and power

Pretence and Echo: Towards an Integrated Account of Verbal Irony

Compound figures: priority and speech-act structure

Embedding Irony, and the Semantics/Pragmatics Distinction

Go Figure: Understanding Figurative Talk