LCS 11: Cognitive Science Nature vs. nuture divide Poverty of the - PowerPoint PPT Presentation

Agenda Pomona College LCS 11: Cognitive Science ֠ Nature vs. nuture divide ֠ Poverty of the stimulus and reactions Language acquisition ֠ Critical period ֠ Replication projects Jesse A. Harris ֠ Class evaluations March 25, 2013 Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 1 Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 2 Big picture Abilities Three related questions in acquisition research Just what abilities and skills are required to learn a language? 1. What abilities or skills are required to learn a language? 2. When do these abilities emerge? What specific stages are 1. Perceive the appropriate linguistic sounds there in language development? 2. Distinguish words and other meaningful elements in 3. What other cognitive abilities, if any, do these abilities speech stream depend on? 3. Pair elements in stream with meanings 4. Structure those elements into more complex structures Raises the all important question 5. Sufficient interaction with language in use What constitutes a knowledge of language? What kind of knowledge does a native speaker have about his or her language? Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 3 Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 4

Segmentation problem Segmentation problem (1) Thisisanaverageexampleofaspeechstream. Segmentation problem There are rarely definitive pauses between words in a speech stream. How can the child learn how to recognize word boundaries? This an average example of a speech stream Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 5 Segmentation problem Segmentation problem 0.2003 0 -0.1477 0 2.194 Time (s)

Segmentation problem Segmentation problem (2) a. Two dults (Uttered) Phonological bootstrapping b. Two adults (Target) Children use statistical regularities to determine likely word boundaries. (3) a. I don’t want to go to your ami (Uttered) ◮ Infants can’t use top-down knowledge to infer words b. I don’t want to go to your Miami (Target) ◮ Must make the most of the input ◮ Sa ff ran et al (1996) proposed that infants use transitional (4) a. I am being have! (Uttered) probability measures to infer boundaries. b. I am behaving! (Target) Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 9 Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 10 Segmentation problem Nature vs. Nuture Transitional probabilities Nature Nuture The probability of one sound X given another Y: Child’s acquisition of Child’s acquisition of language is acheived language is acheived P ( Y | X ) = P ( XY ) P ( X ) ≈ freq ( XY ) primarily through innate primarily through exposure. freq ( X ) knowledge. ◮ Statistical regularities ◮ Universality in grammar ◮ Model learning with ◮ Languages regulate what sounds can go together within a ◮ Specialized learning limited memory word, but not so much at a word boundary. device ◮ Child directed speech Within word High transitional probabilities ◮ Inborn biases and Between word Low transitional probabilities defaults ◮ Similiar development (5) pretty# baby pre + ty is high within word; ty + ba is low. Plausibly elements of both! Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 11 Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 12

Nuture Nature The extreme case of behaviorism: Impossible! Simple set of reward-based general 1. Too many possible responses learning principles that child uses to given stimuli learn language. No previous 2. Ignores complexity and knowledge or expectations. creativity of language structures Question 3. Too little exposure to develop rich knowledge through Does this view seem reasonable? imitation. What might be some intuitive concerns about such a view? Noam Chomsky B.F. Skinner Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 13 Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 14 Nature vs. Nuture Nature Poverty of the stimulus argument The language that children acquire goes beyond the input that they receive. Language is unlearnable from exposure alone. Instead, children acquire language by In support using an innate knowledge of language, along with very powerful ◮ Limited negative language speci fi c learning biases. evidence ◮ Children seem Innate knowledge of language is unresponsive to abstract , and known as Universal correction Grammar (UG). ◮ Grammatical judgments for novel forms Noam Chomsky ◮ Relative uniformity in development Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 15

Nature vs. Nuture Poverty of the stimulus argument Child My teacher holded the rabbits and we patted The language that children acquire goes beyond the input that them. they receive. Language is unlearnable from exposure alone. Adult Did you say the teacher held the baby rabbits? Child Yes. In support Adult What did you say she did? ◮ Limited negative Child She holded the baby rabbits and we patted them. evidence Adult Did you say she held them tightly? ◮ Children seem Child No, she holded them loosely. unresponsive to Jean Berko correction http://www.onbeing.org/program/unfolding-language-unfolding-life/256 ◮ Grammatical judgments for novel forms ◮ Relative uniformity in development Wug test Wug test

Wug test Wug test ◮ Full Wug test materials: http://childes.psy.cmu.edu/topics/ ◮ Video of Berko Gleason describing her work: http://video.pbs.org/video/1608569948/ Nature vs. Nuture Critical period Poverty of the stimulus argument The language that children acquire goes beyond the input that Developmental limitations they receive. Language is unlearnable from exposure alone. Acquiring a language requires that the child be exposed to su ffi cient language input within a developmental time In support Counterclaims window. ◮ Underestimates negative ◮ Limited negative ◮ Estimates vary: 5 years – puberty evidence evidence ◮ Period of neural plasticity ◮ Children learn language ◮ Children seem piecemeal according to ◮ Bilingualism unresponsive to cognitive limitations correction ◮ Profeciency ◮ Unclear how a rich UG ◮ Grammatical judgments ◮ Failure to develop language would be innately for novel forms ◮ Critical period observed in other species encoded ◮ Relative uniformity in development Jesse A. Harris: LCS 11: Cognitive Science, Language acquisition 24

Critical period ◮ Severely abused and isolated ◮ Discovered at age 13 without having acquired speech ◮ Eventually able to acquire a limited vocabulary, but never a fully complex linguistic system. 1. Few grammatical words 2. No auxiliaries 3. Improper verbal and nominal in fl ection 4. No passives 5. Negation limited