Universiteit Utrecht – Faculty of Humanities
Constraints on Remote Dependency Learning
Adults and infant as young as 15 months have been shown to possess the ability to detect relationships between remote (non-adjacent) words in unfamiliar input (Gómez, 2002; Gómez & Maye, 2005; Kerkhoff et al., 2013). This ability may aid young L1 learners in detecting dependencies in their own native language, often marking important syntactic relationships like agreement or aspectual paradigms: for instance, by 18 months English-learning infants detect the difference between The cook is often baking and *The cook can often baking, demonstrating sensitivity to the relationship between the auxiliary and the aspect-denoting suffix on the verb (Santelmann & Jusczyk, 1998). The emergence of a remote dependency learning mechanism around the same age as sensitivity to morpho-syntactic dependencies in the native language suggests the possibility that it is this very learning mechanism which supports the acquisition of morpho-syntactic relationships; however, research up to date has offered little more than proof of concept in favor of this hypothesis.
Remote dependency learning has been shown to be a highly constrained learning mechanism: dependencies are reliably detected only in the presence of segmentation cues (Peña et al. 2002) or phonological cues (Newport & Aslin, 2004; Onnis et al., 2005), only if presented in a sufficiently variable context (Gómez, 2002), and only within a relatively small working memory window (Grama et al., 2013). Using an artificial grammar learning paradigm with both adults and infants, in this project I set out to investigate whether the specific ways in which remote dependency learning is constrained allow it to be a potentially viable tool for language acquisition. In Part 1 of my presentation I look at the perceptual/prosodic properties of the kind of (functional) words and morphemes that engage in morpho-syntactic dependencies; I investigate how these properties affect the detection of dependencies, showing that (i) remote dependency learning is affected by perceptual cues, (ii)Gestalt principles of perception guide learning, such that dependencies are easier to learn when the dependent elements are perceptually similar to each other and distinct in their environment, (iii) the perceptual cues that mark functional words/morphemes in natural languages may be crucial in detecting morpho-syntactic dependencies. In Part 2, I consider the developmental timeline of the acquisition of morpho-syntactic dependencies: sensitivity to individual functional words/morphemes precedes sensitivity to morpho-syntactic dependencies. I investigate whether prior familiarity with individual elements of a dependency enhances the ability to detect dependencies between them. Preliminary results show that prior familiarization with individual elements in fact has a negative effect on the ability to detect dependencies, suggesting that words and dependencies between them should be learned at the same time; this hypothesis is compatible with an earlier emergence of the remote learning ability (Friederici et al., 2011).
References
Friederici, A. D., Mueller, J. L., & Oberecker, R. (2011). Precursors to natural grammar learning: Preliminary evidence from 4-month-old infants. Plos One, 6(3), e17920.
Gómez, R.L. (2002). Variability and the detection of invariable structure. Psychological Science, 13 (5), 431-436.
Gómez, R. L., & Maye, J. (2005). The developmental trajectory of nonadjacent dependency learning. Infancy, 7, 183-206.
Grama, I.C., Wijnen, F.N.K & Kerkhoff, A.O. (2013). Constraints on non-adjacent dependency-learning: distance matters. In Baiz, S., Goldman, N. & Hawkes, R., (Eds), BUCLD 37 Online Proceedings Supplement. http://www.bu.edu/bucld/supplementvol37/
Kerkhoff, A.O., de Bree, E. , de Klerk, M. & Wijnen, F. N.K. (2013). Non-adjacent dependency learning in infants at familial risk of dyslexia. Journal of Child Language, 40 (1), 11-28.
Newport, E. L. & Aslin, R. N. (2004). Learning at a distance 1. Statistical learning of non-adjacent dependencies. Cognitive Psychology, 48, 127-162.
Onnis, L., Monaghan, P., Richmond, K. & Chater, N. (2005) Phonology impacts segmentation in online speech processing. Journal of Memory and Language, 53, 225-237.
Peña, M., Bonatti, L. L., Nespor, M. & Mehler, J. (2002). Signal-driven computations in speech processing. Science, 298, 604-607.
Santelmann, L. M. & P.W. Jusczyk (1998). Sensitivity to discontinuous dependencies in language learners: evidence for the limitations in processing space. Cognition, 69, 105-134.
