Learning, Memory, and Syntactic Bootstrapping: A Meditation

Abstract

Lila Gleitman's body of work on word learning raises an apparent paradox. Whereas work on syntactic bootstrapping depends on learners retaining information about the set of distributional contexts that a word occurs in, work on identifying a word's referent suggests that learners do not retain information about the set of extralinguistic contexts that a word occurs in. I argue that this asymmetry derives from the architecture of the language faculty. Learners expect words with similar meanings to have similar distributions, and so learning depends on a memory for syntactic environments. The referential context in which a word is used is less constrained and hence contributes less to the memories that drive word learning.

Public language, private language, and subsymbolic theories of mind

“…Gleitman et al (Gleitman, 1990; Gleitman et al, 2005; Landau & Gleitman, 1985; Lidz, 2020; Lidz et al, 2003a, 2003b) have developed a detailed and robustly confirmed theory of how children learn the meanings of novel lexical items. The mere fact that English speakers say “dog,” rather than “perro” or “chien,” shows that the mapping between word meaning and pronunciation must be learned on the basis of the child's experience.…”

“…Gleitman argues that this is made possible through syntactic bootstrapping , a process by which language learners narrow down the range of descriptions of the scene to associate with these verbs by attending to the sentential contexts in which these words are used. Lidz (2020) gives the example of a child learning the difference between contact verbs (“hit,” “kick,” …) and change of state verbs (“break,” “melt,” …). Many scenes will be truly described by verbs from both of these classes—a scene in which something is broken is likely one in which something contacts it—which poses a problem for the learner.…”

Editor's Introduction: 2017 Rumelhart Prize Issue Honoring Lila R. Gleitman

“…Lidz () pushes syntactic bootstrapping in a different way. He poses the following puzzle: Gleitman and colleagues have proposed two different kinds of solution to the problem of learning the meaning of words from observation.…”

What language models can tell us about learning adjectives

“…In other words, children subcategorize verbs by paying attention to the number of arguments and the type(s) of complement (see discussion in Lidz 2020). Beyond the verbal domain, research has indicated that distributional cues are also helpful for learning other grammatical categories, such as adjectives (see Syrett 2007;Syrett & Lidz 2010;Becker 2015;Gotowski 2022).…”