2014 Annual Meeting of the Linguistics Association of Great Britain

The Queen's College, University of Oxford, 1-5 September 2014

Workshop on Learning Biases in Natural and Artificial Language Acquisition

Friday 5th September 2014

This workshop, organised by Adam Albright and Andrew Nevins, will be held in conjunction with Adam Albright's Linguistics Association Lecture at the Annual Meeting of the Linguistics Association of Great Britain 2014.

Plenary speaker

• Adam Albright (MIT) - 'Generalizing phonological patterns with phonetic and featural biases'

Invited speakers

• Alex Cristia (CNRS) - 'Linguistically relevant information in the implicit abstraction of sound patterns from speech: Infant and adult data'
• Sara Finley (Pacific Lutheran University) - 'Shedding light on phonological representations through adult learning experiments'
• Elliott Moreton (University of North Carolina) - 'Phonological concept learning: Experiments and models'
• Ruben van de Vijver & Dinah Baer-Henney (Düsseldorf) - 'Learners’ little helper'

Background

What expectations or biases do learners bring to the task of learning phonological grammars? Work on language typology, diachronic change, and evaluation metrics for learning algorithms has identified a number of factors that might encourage learners to favor one hypothesis over another. These include preferences based on formal properties of the grammar, such as a bias for featurally simpler or more general processes, or a bias towards certain type of interactions. They also include substantive biases for certain types of processes, such as a preference for processes that target phonetically difficult structures, or a bias against processes that lead to perceptually salient alternations, or even limitations that make some processes completely unlearnable.

Until recently, the argument that learners favor some patterns over others has largely been based on indirect evidence: learning biases can provide an account of how grammatical preferences shape acquisition errors, language change, and typology. The past decade has seen a rapid rise of interest in studying learning directly in the lab, both among infants and adults. This work has studied the time course of acquisition of natural language (L1) patterns by children, as well as the rate or readiness with which infants and adults learn artificial grammars. The goal of this workshop is to bring together researchers employing a variety of techniques to study this kind of phonological learning “in the lab”. The workshop aims to foster a dialogue on questions such as: how can we relate performance in an artificial lab task to natural language acquisition? What kinds of biases have actually been supported by experimental results, to date? What kinds of biases do these techniques allow us to test, and what kinds of biases can only be observed within the context of a full-blown linguistic system with qualitatively and quantitatively more complex training, longer timescales of learning, and learning within richer semantic contexts? What is the contribution, if any, of participants’ L1 to the task of artificial grammar learning? We hope that the invited talks and the posters, selected from an open call for papers, will shed light on these and other questions through a range of theoretical and empirical contributions.

Call for workshop posters and papers for the main session

Abstracts are invited for posters to be presented as part of the Workshop on Learning Biases in Natural and Artificial Language Acquisition on Friday 5th September 2014. Abstracts for posters should follow the general guidelines for abstracts set out in the call for papers, but should be clearly marked (e.g. in the header of the abstract) as intended for the Workshop on Learning Biases Poster Session.

Abstracts can also be submitted for papers in the main parallel sessions of the LAGB 2014. We particularly invite papers that relate to the themes that will be addressed in the Workshop on Learning Biases in Natural and Artificial Language Acquisition. Papers in the parallel sessions will be scheduled for Wednesday 3rd or Thursday 4th September 2014.

The deadline for abstracts for posters in the Workshop on Learning Biases Poster Session (this workshop is open to calls for posters only) or for papers in the general session is Friday 11th April 2014. Notification of acceptance will be made in May 2014.

 Details of how to submit an abstract are available in the Call for papers, themed sessions and workshop posters.