Getting to “Hearer-old”: Charting Referring Expressions Across Time

Abstract

When a reader is first introduced to an entity, its referring expression must describe the entity. For entities that are widely known, a single word or phrase often suffices. This paper presents the first study of how expressions that refer to the same entity develop over time. We track thousands of person and organization entities over 20 years of New York Times (NYT). As entities move from hearernew (first introduction to the NYT audience) to hearer-old (common knowledge) status, we show empirically that the referring expressions along this trajectory depend on the type of the entity, and exhibit linguistic properties related to becoming common knowledge (e.g., shorter length, less use of appositives, more definiteness). These properties can also be used to build a model to predict how long it will take for an entity to reach hearer-old status. Our results reach 10-30% absolute improvement over a majority-class baseline.

Characterizing Collective Attention via Descriptor Context: A Case Study of Public Discussions of Crisis Events

“…As more people became familiar with the locations mentioned in news coverage about the island (DiJulio, Muñana, and Brodie 2017), news headlines and articles referred to "San Juan" without extra contextual descriptors such as "the capital of Puerto Rico." This is consistent with a rational model of communication in which linguistic contextualization is used for entities that might otherwise be unknown or ambiguous (Prince 1992;Staliūnaitė et al 2018): as San Juan became increasingly salient through repeated mentions, readers could be expected to understand the reference without additional context. Figure 1 presents evidence from Twitter in favor of this hypothesis: the locations of San Juan (1a) and Myrtle Beach (1b) received fewer contextualizing descriptors following peaks in the volume of mentions in discussions of Hurricanes Maria and Florence respectively.…”

“…In a dataset of Facebook posts from public groups concerning Hurricane Maria relief, we find that location mentions received descriptors more often when the locations were not local to the group of discussion, suggesting that descriptors may be used to help explain new information to audiences. In a dataset of public Twitter posts related to five hurricane events, we find that the aggregate rate of descriptor phrases decreased following the peaks in these locations' collective attention, supporting prior findings on named entity references in professional newstext (Staliūnaitė et al 2018).…”

“…These predictions have received empirical support in corpus analyses: Siddharthan, Nenkova, and McKeown (2011) found that highly salient or important named entities in a summary of a news story are more likely to be understood as shared knowledge and are therefore less likely to need a descriptor phrase. Similarly, a diachronic analysis of newspaper articles showed that entities tended to receive fewer descriptors as they became more familiar over time (Staliūnaitė et al 2018). In laboratory settings, speakers have been found to modulate their use of contextualizing information depending on the expected entity salience to specific listeners (Galati and Brennan 2010), but such fine-grained analyses have not previously been explored with corpus data.…”

“…To extract sentence structure from text, we used dependency parsing, which decomposes sentences into directed acyclic graphs connecting words and phrases (Eisenstein 2019). Following Staliūnaitė et al (2018), we defined a set of dependencies to capture the natural language processing, with hundreds of thousands of tokens of labeled data for named entity recognition (Hovy et al 2006;Taulé, Martí, and Recasens 2008). The extension of this data acquisition pipeline to languages that lack substantial labeled data may pose a significant challenge (Rahimi, Li, and Cohn 2019 MODIFIER phrase type in a subclause (adjectival clause, appositional modifier, prepositional modifier, numeric modifier) and another set of dependencies to capture the COMPOUND type in a super-clause (nominal modifier, compound, appositional modifier).…”

Continual Adaptation for Efficient Machine Communication

“…terlocutors are able to establish ad hoc conventions based on this history (Clark and Wilkes-Gibbs, 1986;Clark, 1996), allowing for increasingly accurate and efficient communication. Speakers can remain understandable while expending significantly fewer words (Krauss and Weinheimer, 1964;Orita et al, 2015;Staliūnaitė et al, 2018;Hawkins et al, 2020a;Stewart et al, 2020). For example, consider a nurse visiting a bedridden patient at their home.…”