Lexical distributional cues, but not situational cues, are readily used to learn abstract locative verb-structure associations
被引:10
|
作者:
Twomey, Katherine E.
论文数: 0引用数: 0
h-index: 0
机构:
Univ Lancaster, Lancaster LA1 4YW, England
ESRC Int Ctr Language & Commun Dev LuCiD, Lancaster, EnglandUniv Lancaster, Lancaster LA1 4YW, England
Twomey, Katherine E.
[1
,2
]
论文数: 引用数:
h-index:
机构:
Chang, Franklin
[2
,3
]
Ambridge, Ben
论文数: 0引用数: 0
h-index: 0
机构:
ESRC Int Ctr Language & Commun Dev LuCiD, Lancaster, England
Univ Liverpool, Liverpool L69 7ZA, Merseyside, EnglandUniv Lancaster, Lancaster LA1 4YW, England
Ambridge, Ben
[2
,3
]
机构:
[1] Univ Lancaster, Lancaster LA1 4YW, England
[2] ESRC Int Ctr Language & Commun Dev LuCiD, Lancaster, England
[3] Univ Liverpool, Liverpool L69 7ZA, Merseyside, England
Children must learn the structural biases of locative verbs in order to avoid making overgeneralisation errors (e.g., *I filled water into the glass). It is thought that they use linguistic and situational information to learn verb classes that encode structural biases. In addition to situational cues, we examined whether children and adults could use the lexical distribution of nouns in the post-verbal noun phrase of transitive utterances to assign novel verbs to locative classes. In Experiment 1, children and adults used lexical distributional cues to assign verb classes, but were unable to use situational cues appropriately. In Experiment 2, adults generalised distributionally-learned classes to novel verb arguments, demonstrating that distributional information can cue abstract verb classes. Taken together, these studies show that human language learners can use a lexical distributional mechanism that is similar to that used by computational linguistic systems that use large unlabelled corpora to learn verb meaning. (C) 2016 The Authors. Published by Elsevier B.V.