Linguistic generalization on the basis of function and constraints on the basis of statistical preemption

Research output: Contribution to journalArticle

Standard

Linguistic generalization on the basis of function and constraints on the basis of statistical preemption. / Perek, Florent; Goldberg, Adele.

In: Cognition, Vol. 168, 11.2017, p. 276-293.

Research output: Contribution to journalArticle

Harvard

APA

Vancouver

Author

Bibtex

@article{1f4127ab4b914fa5bf171d8c5af605e7,
title = "Linguistic generalization on the basis of function and constraints on the basis of statistical preemption",
abstract = "A key question in language learning is what encourages and what constraintsgeneralization beyond what is witnessed in the input. Experiment 1 exposesparticipants to two novel word order constructions that differ in terms of theirsemantics: One construction exclusively describes actions that have a strongeffect; the other construction describes actions with a weaker but otherwisesimilar affect. One group of participants witnessed novel verbs onlyappearing in one construction or the other, while another group witnessed aminority of verbs alternating between constructions. Subsequent productionand judgment results demonstrate that participants in both conditionsextended and accepted verbs in whichever construction best described theintended message. Unlike related previous work, this finding is not naturallyattributable to prior knowledge of the likely division of labor between verbsand constructions. A second experiment included one verb (out of six) thatwas witnessed in a single construction to describe both strong and weakeffects, essentially preempting the use of the other construction. In this case,participants were much more lexically conservative with this verb and otherverbs, while they nonetheless displayed an appreciation of the distinctsemantics of the constructions with new novel verbs. Results indicate that theneed to better express an intended message encourages generalization, whilestatistical preemption constrains generalization by providing evidence thatverbs are restricted in their distribution.",
keywords = "Language acquisition, Artificial language learning, Novel construction learning, Statistical learning, Argument structure constructions, Generalization",
author = "Florent Perek and Adele Goldberg",
year = "2017",
month = nov
language = "English",
volume = "168",
pages = "276--293",
journal = "Cognition",
issn = "0010-0277",
publisher = "Elsevier",

}

RIS

TY - JOUR

T1 - Linguistic generalization on the basis of function and constraints on the basis of statistical preemption

AU - Perek, Florent

AU - Goldberg, Adele

PY - 2017/11

Y1 - 2017/11

N2 - A key question in language learning is what encourages and what constraintsgeneralization beyond what is witnessed in the input. Experiment 1 exposesparticipants to two novel word order constructions that differ in terms of theirsemantics: One construction exclusively describes actions that have a strongeffect; the other construction describes actions with a weaker but otherwisesimilar affect. One group of participants witnessed novel verbs onlyappearing in one construction or the other, while another group witnessed aminority of verbs alternating between constructions. Subsequent productionand judgment results demonstrate that participants in both conditionsextended and accepted verbs in whichever construction best described theintended message. Unlike related previous work, this finding is not naturallyattributable to prior knowledge of the likely division of labor between verbsand constructions. A second experiment included one verb (out of six) thatwas witnessed in a single construction to describe both strong and weakeffects, essentially preempting the use of the other construction. In this case,participants were much more lexically conservative with this verb and otherverbs, while they nonetheless displayed an appreciation of the distinctsemantics of the constructions with new novel verbs. Results indicate that theneed to better express an intended message encourages generalization, whilestatistical preemption constrains generalization by providing evidence thatverbs are restricted in their distribution.

AB - A key question in language learning is what encourages and what constraintsgeneralization beyond what is witnessed in the input. Experiment 1 exposesparticipants to two novel word order constructions that differ in terms of theirsemantics: One construction exclusively describes actions that have a strongeffect; the other construction describes actions with a weaker but otherwisesimilar affect. One group of participants witnessed novel verbs onlyappearing in one construction or the other, while another group witnessed aminority of verbs alternating between constructions. Subsequent productionand judgment results demonstrate that participants in both conditionsextended and accepted verbs in whichever construction best described theintended message. Unlike related previous work, this finding is not naturallyattributable to prior knowledge of the likely division of labor between verbsand constructions. A second experiment included one verb (out of six) thatwas witnessed in a single construction to describe both strong and weakeffects, essentially preempting the use of the other construction. In this case,participants were much more lexically conservative with this verb and otherverbs, while they nonetheless displayed an appreciation of the distinctsemantics of the constructions with new novel verbs. Results indicate that theneed to better express an intended message encourages generalization, whilestatistical preemption constrains generalization by providing evidence thatverbs are restricted in their distribution.

KW - Language acquisition

KW - Artificial language learning

KW - Novel construction learning

KW - Statistical learning

KW - Argument structure constructions

KW - Generalization

M3 - Article

VL - 168

SP - 276

EP - 293

JO - Cognition

JF - Cognition

SN - 0010-0277

ER -