The role of prediction in construction-learning

Abstract

It is well-established that (non-linguistic) categorization is driven by a functional demand of prediction. We suggest that prediction likewise may well play a role in motivating the learning of semantic generalizations about argument structure constructions. We report corpora statistics that indicate that the argument frame or construction has roughly equivalent cue validity as a predictor of overall sentence meaning as the morphological form of the verb, and has greater category validity. That is, the construction is at least as reliable and more available than the verb. Moreover, given the fact that many verbs have quite low cue validity in isolation, attention to the contribution of the construction is essential.

28. Argument structure constructions

“…To examine whether constructional forms are predictive of sentence meaning in the naturally occurring input that children hear, Goldberg et al (2005) examined a corpus of child directed speech to investigate how consistently the meaning of a construction was encoded by the meaning of the verb used in the construction on the one hand, and the meaning of the construction itself on the other.…”

“…Children's ability to learn argument structure constructions themselves, that is to map novel constructional forms to novel meanings without being influenced (for better or worse) by patterns of language that the child already knows was recently investigated by Goldberg and colleagues in a number of studies that have produced evidence that children are in fact able to assign a novel meaning to a novel construction (e.g., Goldberg et al 2005;Casenhiser and Goldberg 2005;Boyd et al 2009). …”

“…Goldberg and colleagues (Goldberg et al 2007) present evidence that parallels evidence derived from non-linguistic category learning (Gentner and Medina 1998;Markman and Gentner 1993), suggesting that early presentation of stimuli with shared concrete similarity facilitates construction learning. Other work has demonstrated a facilitative effect on construction learning when exemplars follow a so-called Zipfian distribution (Zipf 1935) in which the frequency with which a verb occurs in a given construction accounts for the lion's share of tokens encountered by learners Goldberg et al 2004). A number of corpus-based studies (e.g., Gries and Wulff 2005;CameronFaulkner et al 2003), have suggested that natural language input tends to mirror this effect (see also Divjak and Caldwell-Harris this volume for a discussion of frequency effects), and evidence from non-linguistic category learning (Elio and Anderson 1984) has shown a facilitative effect for such an input distribution.…”

Licensing and blocking factors in the use of BEGIN verbs

“…Some cognitive approaches within CG (cf. Geeraerts & Cuyckens, 2010) maintain that grammar and the lexicon form a single whole and that linking rules are not required (Goldberg, 1995(Goldberg, , 2006Goldberg, Casenhiser & Sethuraman, 2005). Others, by contrast, vindicate the importance of verbal semantics in determining syntactic configurations (Croft, 2001;Boas, 2008aBoas, , 2008b.…”

Construction Grammar in the twenty-first century

“…For example, chapter 4 is based almost entirely on Goldberg, Casenhiser & Sethurman (2004, 2005 and . Similarly, a large part of chapter 6 focuses on prior research by Hare and Goldberg (1999), Bencini & Goldberg (2000), Chang, Bock & Goldberg (2003), and Goldberg, Casenhiser & Sethurman (2005). Goldberg's heavy reliance on prior work should not detract from the quality of its results.…”