QTUNA: A Corpus for Understanding How Speakers Use Quantification

Abstract

A prominent strand of work in formal semantics investigates the ways in which human languages quantify over the elements of a set, as when we say "All A are B", "All except two A are B", "Only a few of the A are B" and so on. Our aim is to build Natural Language Generation algorithms that mimic humans' use of quantified expressions. To inform these algorithms, we conducted on a series of elicitation experiments in which human speakers were asked to perform a linguistic task that invites the use of quantified expressions. We discuss how these experiments were conducted and what corpora they gave rise to. We conduct an informal analysis of the corpora, and offer an initial assessment of the challenges that these corpora pose for Natural Language Generation. The dataset is available at: https: //github.com/a-quei/qtuna.

Computational Modelling of Quantifier Use: Corpus, Models, and Evaluation

“…In this section, we explain how the experiment was set up and how the resulting corpus was analysed. An initial introduction and analysis of this experiment can be found in Chen, van Deemter, Pagliaro, Smalbil, and Lin (2019b).…”

Generating Quantified Descriptions of Abstract Visual Scenes

“…We will summarise the main findings from these experiments before proposing and experimentally comparing algorithms that seek to mimic the aforementioned corpus. More details can be found in Chen et al (2019) 1 . The QTUNA experiment was set up in order to study how speakers use sequences of QEs to describe a visual scene.…”

“…Analysis of this corpus (see Chen et al (2019)) taught us the following lessons: 1) Speakers use more vague quantifiers (e.g., most, few) as domain size increases; 2) Speakers also use more under-specifications in larger domains, describing large domain "with a broad brush"; 3) The average length of quantified descriptions is not significantly larger in larger domains than in smaller ones; 4) Speakers tends to start describing the high level information of the whole scene, before going into detail about parts; 5) Speakers tend to mention shape before the colour. Quantified Expression Generation.…”