Learning homophones in context: Easy cases are favored in the lexicon of natural languages

Abstract

Even though ambiguous words are common in languages, children find it hard to learn homophones, where a single label applies to several distinct meanings (e.g., Mazzocco, 1997). The present work addresses this apparent discrepancy between learning abilities and typological pattern, with respect to homophony in the lexicon. In a series of five experiments, 20-month-old French children easily learnt a pair of homophones if the two meanings associated with the phonological form belonged to different syntactic categories, or to different semantic categories. However, toddlers failed to learn homophones when the two meanings were distinguished only by different grammatical genders. In parallel, we analyzed the lexicon of four languages, Dutch, English, French and German, and observed that homophones are distributed non-arbitrarily in the lexicon, such that easily learnable homophones are more frequent than hard-to-learn ones: pairs of homophones are preferentially distributed across syntactic and semantic categories, but not across grammatical gender. We show that learning homophones is easier than previously thought, at least when the meanings of the same phonological form are made sufficiently distinct by their syntactic or semantic context. Following this, we propose that this learnability advantage translates into the overall structure of the lexicon, i.e., the kinds of homophones present in languages exhibit the properties that make them learnable by toddlers, thus allowing them to remain in languages.

Why do human languages have homophones?

“…This suggests that the pressure is not against the existence of homophony per se, but rather, could reflect a constraint on the extent to which any given wordform can be saturated with distinct, unrelated meanings. Assigning too many unrelated meanings to the same signal could impede communication or learning (Casenhiser, 2005; though see Dautriche et al, 2018), and may thus be selected against. Such a pressure against oversaturation is roughly analogous to what others have termed diversification (Zipf, 1949) or a pressure for clarity (Piantadosi et al, 2012).…”

“…Such a model should also predict which potential homophones will be selected against (and what form this selection process takes, i.e., whether it is via the avoidance of homophony-inducing mergers (Wedel et al, 2013;Yin & White, 2018) or something else) and which will be preserved. Homophones should be more likely to survive in a lexicon if their meanings are systematically made sufficiently discriminable by context (Dautriche et al, 2018). A better understanding of this process would also yield insights into which sources of contextual information human speakers and comprehenders routinely sample and deploy for disambiguation, and therefore influence language change.…”

“…Part of the answer appears to be that human comprehenders are adept at disambiguating ambiguous input using various contextual cues (Levinson, 2000;Wasow et al, 2005;Ferreira, 2008;Piantadosi et al, 2012). In the case of homophones, a wide array of cues to meaning are available, including the syntactic structures that words are embedded in (Dautriche et al, 2018), gestures that accompany speech (Holle & Gunter, 2007;Holler & Beattie, 2003;Kidd & Holler, 2009), and statistical aspects of linguistic context (Aina et al, 2019). The human capacity for disambiguation thus creates a tolerant environment for ambiguous wordforms--explaining why as languages evolve, homophones might not be strictly selected against.…”

Miller's Monkey Updated: Communicative Efficiency and the Statistics of Words in Natural Language

“…", and conclude that communication features in the design of language as a cognitive system. The PTG study has been influential (see also Dautriche, Fibla, Fievet, and Christophe (2018); Mahowald, Fedorenko, Piantadosi, and Gibson (2013); Piantadosi, Tily, and Gibson (2011) for similar studies), and the communicative function served by context, which is critical for PTG's study of lexical ambiguity, has been extended to other domains of language including morphology and syntax (see Gibson et al 2019 for a review).…”

4-5-year-old children adapt to the reliability of conflicting sources of information to learn novel words

“…Appendix for the items), because no distractor is available. No feedback is given to the participants to validate their interpretation after their choice, in any item, induction or test, as was the case in other similar previous studies(Havron et al, 2019;Yurovsky et al, 2016 andDautriche et al, 2018).…”