Editing Noam Chomsky (section)

===Universal grammar===
{{Main|Universal grammar}}
Since the 1960s, Chomsky has maintained that syntactic knowledge is partially inborn, implying that children need only learn certain language-specific features of their [[native language]]s. He bases his argument on observations about human [[language acquisition]] and describes a "[[poverty of the stimulus]]": an enormous gap between the linguistic stimuli to which children are exposed and the rich [[linguistic competence]] they attain. For example, although children are exposed to only a very small and finite subset of the allowable [[Sentence (linguistics)|syntactic variants]] within their first language, they somehow acquire the highly organized and systematic ability to understand and produce [[Digital infinity|an infinite number of sentences]], including ones that have never before been uttered, in that language.{{sfn|Dovey|2015}} To explain this, Chomsky proposed that the primary linguistic data must be supplemented by an [[innate linguistic capacity]]. Furthermore, while a human baby and a kitten are both capable of [[inductive reasoning]], if they are exposed to exactly the same linguistic data, the human will always acquire the ability to understand and produce language, while the kitten will never acquire either ability. Chomsky referred to this difference in capacity as the [[language acquisition device]], and suggested that linguists needed to determine both what that device is and what constraints it imposes on the range of possible human languages. The universal features that result from these constraints would constitute "universal grammar".{{sfn|Chomsky}}{{sfn|Thornbury|2006|p=234}}{{sfn|O'Grady|2015}} Multiple researchers have challenged universal grammar on the grounds of the evolutionary infeasibility of its genetic basis for language,{{sfnm| 1a1=Christiansen|1a2=Chater|1y=2010|1p=489| 2a1=Ruiter|2a2=Levinson|2y=2010|2p=518}} the lack of crosslinguistic surface universals,{{sfnm| 1a1=Evans|1a2=Levinson|1y=2009|1p=429| Tomasello|2009|2p=470}} and the unproven link between innate/universal structures and the structures of specific languages.{{sfn|Tomasello|2003|p=284}} [[Michael Tomasello]] has challenged Chomsky's theory of innate syntactic knowledge as based on theory and not behavioral observation.{{sfn|Tomasello|1995|p=131}} The empirical basis of poverty of the stimulus arguments has been challenged by [[Geoffrey Pullum]] and others, leading to back-and-forth debate in the [[language acquisition]] literature.<ref name="PullumScholz">{{cite journal|last1=Pullum|first1=Geoff|author-link1=Geoff Pullum|last2=Scholz|first2=Barbara|author-link2=Barbara Scholz|date=2002|title=Empirical assessment of stimulus poverty arguments|journal=The Linguistic Review|volume=18|issue=1–2|pages=9–50|doi=10.1515/tlir.19.1-2.9}}</ref><ref name="LegateYang">{{cite journal|last1=Legate|first1=Julie Anne|author-link1=Julie Anne Legate|last2=Yang|first2=Charles|author-link2=Charles Yang (linguist)|date=2002|title=Empirical re-assessment of stimulus poverty arguments|journal=The Linguistic Review|volume=18|issue=1–2|pages=151–162|doi=10.1515/tlir.19.1-2.9|url=https://www.ling.upenn.edu/~ycharles/papers/tlr-final.pdf}}</ref> Recent work has also suggested that some [[recurrent neural network]] architectures can learn hierarchical structure without an explicit constraint.<ref>{{cite journal|last1=McCoy|first1=R. Thomas|last2=Frank|first2=Robert|last3=Linzen|first3=Tal|year=2018 |title=Revisiting the poverty of the stimulus: hierarchical generalization without a hierarchical bias in recurrent neural networks|journal=Proceedings of the 40th Annual Conference of the Cognitive Science Society|pages=2093–2098|arxiv=1802.09091 |url=https://tallinzen.net/media/papers/mccoy_frank_linzen_2018_cogsci.pdf}}</ref>