Zainab Sabra


2025

pdf bib
Deciphering Implicatures: On NLP and Oral Testimonies
Zainab Sabra
Proceedings of the first International Workshop on Nakba Narratives as Language Resources

The utterance of a word does not intrinsically convey its intended force. The semantic of utterances is not shaped by the precise references of the words used. Asserting that “it is shameful to abandon our country” does not merely convey information; rather, it asserts an act of resilience. In most of our exchanges, we rarely utilize sentences to describe reality or the world around us. More frequently, our statements aim to express opinions, to influence, or be influenced by others. Words carry more than just their syntax and semantics; they also embody a pragmatic normative force. This divergence between literal and conveyed meaning was depicted in the literature of philosophy of language as the difference between sentence meaning and speaker meaning. Where the former is the literal understanding of the words combined in a sentence, the latter is what the speaker is trying to convey through her expression. In order to derive the speaker meaning from the sentence meaning, J.L. Austin (the author of How To Do Things with Words) relied on conventions, whereas H.P. Grice (the author of Logic and Conversations) relied on conventional and non conventional implicatures. This paper aims to decipher how we can infer speaker meaning from sentence meaning and thereby capture the force of what has been articulated, focusing specifically on oral testimonies. I argue that oral testimonies are forms of speech acts that aim to produce normative changes. Following this discussion, I will examine various natural language processing (NLP) models that make explicit what is implicit in oral testimonies with its benefits and limitations. Lastly, I will address two challenges, the former is related to implicatures that are not governed by conventions and the latter is concerned with the biases inherent in hermeneutical approaches.
Search
Co-authors
    Venues
    Fix data