Untitled Document

CCL	The Comparative Communication Laboratory

Infant Research at the CCL

Infants in the process of acquiring language must overcome many significant difficulties. One such difficulty is learning to segment the words of the language. Unlike orthography, spoken language does not afford reliable cues for segmentation that are stable across languages. One mechanism that has been shown to be useful for young infants in solving this task is the tracking of the co-occurrence of different sounds. Many investigators have studied this type of statistical learning and have established that this mechanism is not domain specific for language and is actually available to other species as well. One of the main research programs at the CCL involves studying statistical learning mechanisms in order to determine what types of computations infants are capable of performing. In addition, we are very interested in how infants deal with multiple language input. Are infants capable of forming multiple representations for each language? This is a question we hope to answer using artificial language stimuli. In addition to studying statistical learning mechanisms as applied to word segmentation studies, we also study them in the context of phonetic learning. During the first year of life, infants transform from generalized discriminators of speech sounds into more adult-like discriminators. For example, at 6 months of age, Japanese infants can discriminate English /r/ and /l/, but at 12 months they struggle (as do native Japanese adults). In collaboration with Jessica Maye at Northwestern, we are studying a statistical learning account for this effect. In addition to these types of studies, we are also interested in studying the types of rules that infants are able to form. Essentially, we are studying whether infants can extract very simple rules from an input and generalize them to novel instantiations. Methodology The infant experiments at the CCL make use of a looking-time paradigm. That is, when infants visit the lab, they typically are familiarized to a string of sounds while they sit in their caretaker’s lap. After a short period has elapsed (typically under ten minutes), they hear test stimuli (which vary by experiment) and gauge their understanding by measuring how long infants will look to certain visual stimuli in the presence of an auditory test stimulus. The experiments are relatively simple and completely non-invasive. If you are interested in participating in an experiment, please contact the lab or email Dan Weiss (djw21@psu.edu). Relevant articles Maye J., Weiss, D.J. & Aslin R.N. (In review) Statistical phonetic learning in infants: Facilitation and feature generalization. Developmental Science Maye, J. & Weiss, D.J. (2003) Statistical cues facilitate infants’ discrimination of difficult phonetic contrasts. Proceedings of the 27th Annual Boston University Conference on Language Development Newport, E, Weiss, D.J., Aslin, R.N., and Wonnacott, L. (in prep) Statistical learning in speech by infants: Syllables or Segments? Weiss, D.J. & Maye, J. (in press) The role of contrast in the acquisition of phonetic systems. Proceedings of Contrast and Complexity in Phonology.

Infant Research at the CCL

image of infant

Infants in the process of acquiring language must overcome many significant difficulties. One such difficulty is learning to segment the words of the language. Unlike orthography, spoken language does not afford reliable cues for segmentation that are stable across languages. One mechanism that has been shown to be useful for young infants in solving this task is the tracking of the co-occurrence of different sounds. Many investigators have studied this type of statistical learning and have established that this mechanism is not domain specific for language and is actually available to other species as well. One of the main research programs at the CCL involves studying statistical learning mechanisms in order to determine what types of computations infants are capable of performing. In addition, we are very interested in how infants deal with multiple language input. Are infants capable of forming multiple representations for each language? This is a question we hope to answer using artificial language stimuli.

In addition to studying statistical learning mechanisms as applied to word segmentation studies, we also study them in the context of phonetic learning. During the first year of life, infants transform from generalized discriminators of speech sounds into more adult-like discriminators. For example, at 6 months of age, Japanese infants can discriminate English /r/ and /l/, but at 12 months they struggle (as do native Japanese adults). In collaboration with Jessica Maye at Northwestern, we are studying a statistical learning account for this effect.

In addition to these types of studies, we are also interested in studying the types of rules that infants are able to form. Essentially, we are studying whether infants can extract very simple rules from an input and generalize them to novel instantiations.

Methodology

The infant experiments at the CCL make use of a looking-time paradigm. That is, when infants visit the lab, they typically are familiarized to a string of sounds while they sit in their caretaker’s lap. After a short period has elapsed (typically under ten minutes), they hear test stimuli (which vary by experiment) and gauge their understanding by measuring how long infants will look to certain visual stimuli in the presence of an auditory test stimulus. The experiments are relatively simple and completely non-invasive. If you are interested in participating in an experiment, please contact the lab or email Dan Weiss (djw21@psu.edu).

Relevant articles

Maye J., Weiss, D.J. & Aslin R.N. (In review) Statistical phonetic learning in infants: Facilitation and feature generalization. Developmental Science

Maye, J. & Weiss, D.J. (2003) Statistical cues facilitate infants’ discrimination of
difficult phonetic contrasts. Proceedings of the 27th Annual Boston University Conference on Language Development

Newport, E, Weiss, D.J., Aslin, R.N., and Wonnacott, L. (in prep) Statistical learning in speech by infants: Syllables or Segments?

Weiss, D.J. & Maye, J. (in press) The role of contrast in the acquisition of phonetic systems. Proceedings of Contrast and Complexity in Phonology.