Artificial grammar learning by 1-year-olds leads to specific and abstract knowledge.

TLDR: Four experiments used the head-turn preference procedure to assess whether infants could extract and remember information from auditory strings produced by a miniature artificial grammar and found infants generalized to new structure by discriminating new grammatical strings from ungrammatical ones after less than 2 min exposure to the grammar.

About: This article is published in Cognition.The article was published on 1999-03-01 and is currently open access. It has received 603 citations till now. The article focuses on the topics: Artificial grammar learning & Formal grammar.

Figures

Table 1 Acquisition strings used in Experiments 1 and 2

Fig. 2. Grammar 2. Finite-state grammar used to generate grammatical strings in Experiment 3.

Table 2 Mean transitional probabilities for acquisition and test strings in Experiments 1 and 2. Ranges are shown in parentheses

Table 6 Counterbalancing design used in Experiments 3 and 4

Table 4 Test strings used in Experiment 2. Both string types begin and end with legal words, but ungrammatical test strings contain illegal transitions (marked by asterisks)

Table 3 Test strings used in Experiment 1

Frequently asked questions

Q1. What are the future works in "Pii: s0010-0277(99)00003-7" ?

Experiment 4 suggests that infants are abstracting some aspect of sequential structure, but future artificial grammar learning studies should be designed to test abstraction processes that might be involved in acquiring knowledge of grammatical categories. For instance, the finding of Werker and Tees ( 1984 ) that infants lose the ability to recognize nonnative contrasts by 10 months of age suggests that infants by this age are beginning to form more abstract representations of phonological categories. Findings reported by Shady ( 1996 ) and Shafer et al. ( 1998 ) suggest that infants begin abstracting syntactic structure soon afterward. 

Q2. What have the authors contributed in "Pii: s0010-0277(99)00003-7" ?

The authors discuss the implications of these findings for the study of language acquisition. 

Q3. Why were the transition probabilities for the test stimuli zero?

Because of the change in vocabulary between training and test, transitional probabilities for grammatical and ungrammatical test stimuli were zero. 

Q4. How many sets of strings were used as test stimuli?

The two remaining sets from the training grammar (10 strings total) and two sets from the other grammar were then used as test stimuli. 

Q5. What were the transition probabilities for ungrammatical tests strings?

Legal word pairs in G1 and G2 were non-overlapping, therefore transitional probabilities for ungrammatical tests strings were all zero. 

Q6. How do infants learn to recognize nonnative contrasts?

For instance, the finding of Werker and Tees (1984) that infants lose the ability to recognize nonnative contrasts by 10 months of age suggests that infants by this age are beginning to form more abstract representations of phonological categories. 

Q7. What is the effect of inverting a determiner and noun?

inverting a determiner and noun not only disrupts word order, but also affects correlations of stress patterns at prosodic boundaries. 

Q8. What did the researchers do to ensure that the observers did not bias the results?

The authors performed reliability coding of the videotapes made during each session to ensure that the observers did not bias the results. 

Q9. Why are test strings instantiated in new vocabulary?

Because test strings are instantiated in new vocabulary, learners can not distinguish the two grammars based on transitional probabilities between remembered word pairs. 

Q10. How many times would an infant be exposed to each acquisition string?

An infant accumulating 41 s exposure (the minimum time recorded in Experiment 1), would have been exposed to each acquisition string approximately 1.08 times. 

Q11. How did the experiment determine the infants' ability to learn grammatical word order?

Thus Experiment 2 investigated whether infants could learn grammatical word order when words occurred in variable positions in sequence and when first-order dependencies were lower than 1.0. 

Q12. What are the properties of finite-state grammars?

Finite-state grammars are limited in terms of their generative capacities (Chomsky, 1957), but nevertheless are complex systems with a number of interesting properties. 

Q13. How many of the participants were available for reliability coding?

Videotapes of 15 of the 16 participants tested were available for reliability coding in which a different observer from the one who made the original observations viewed the videotapes with the soundtracks turned off.