Abstract 1 (Paola) Title: A functional approach to the acquisition of sound contrasts solves the inconvenience of separating underlying forms from perceptual strategies. Abstract:  There is a distinction between the mechanisms underlying perception and the ones underlying production.  I will argue for them to be not just processing devices (performance) but part of knowledge (competence).  Perceptual strategies are transferred, and so they belong to the grammar. Example: If you ask a SS of English what the difference between /i/ and /I/ is, they will say that it’s one of length; their judgment has been confirmed in identification experiments (perception and production). Spanish cue weighting: no difference in height equals “perceptual equivalence” “or single- category assimilation”, and so they use duration when differentiating between the segments.  Transfer depends on the grammar. If what people usually considered performance could be transferred from the L1, is it still possible to call it performance? If anything that you transfer has to be part of the grammar, then we can’t call such things performance transfers but knowledge transfers.  With this approach, we can account for category formation on the basis of different perceptual strategies, and so it could be claimed that there is a perception grammar that performs categorisation.  There is a grammar of performance (in perception) and so there is no distinction between competence and performance. Abstract 2 (Paul) Title: Extending the Gradual Learning Algorithm for style dependency Session: Formalism and Functionalism in Linguistics Abstract: The GLA (Boersma & Hayes 2001) is able to learn the constraint ranking for a grammar that shows free variation. At this talk, I will show how the Algorithm learns a grammar that accounts for variation conditioned by style or register. While the task of the learner in the original GLA was to adjust the ranking of each constraint, the task of the learner in the extended GLA is not only to learn the ranking of these constraints but also the style dependency of each constraint. Thus, if there are N constraints the learner will have adjust the values of 2N quantities. For the learner to be able to achieve this she will have to know the adult speaker’s surface form, the adult speaker’s style or register, and, if faithfulness constraints are concerned, the underlying form. I will show how the extending GLA works on the following examples: xx; pied-piping vs. preposition stranding in English. The conclusion is that examples of style dependant variation found in the literature can be handled well with the extended GLA.

Note: Paul will produce another abstract for GALA about how to solve subset and superset problems in language acquisition.

Abstract 3 (Paola &Paul) TITLE: The Gradual Learning Algorithm predicts the development of first and second language phonological contrasts

SESSION: formalism and functionalism in Linguistics

AUTHORS: Paul Boersma & Paola Escudero

AFFILIATIONS: University of Amsterdam & University of Reading

ADDRESS FOR NOTIFICATION: Paul Boersma Institute of Phonetic Sciences Herengracht 338 NL-1016 CG Amsterdam The Netherlands

PHONE: +31-24-3581839, +44-118-9755796

EMAIL: [email protected], [email protected]

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++ The Gradual Learning Algorithm predicts the development of first and second language phonological contrasts

Acoustically, the English vowels /I/ and /i/ differ in vowel height (first formant) and length (duration). One would expect, therefore, that listeners rely on both cues when distinguishing between these vowels. Escudero (to appear a, b), however, found that Standard Scottish English listeners rely almost exclusively on height, while Southern English speakers rely on both durational and spectral cues. Escudero hypothesised that this perceptual variation is caused by regional variation in production. In a production experiment, we found that in Scottish English, /i/ is much shorter than in Southern English, so that the actually produced durations of /I/ and /i/ overlap appreciably. Conversely, Southern English /I/ turned out to be much higher than Scottish English /I/, i.e. spectrally closer to /i/. To minimize the probability of miscomprehending the speaker's utterance, the Scottish listener must learn to rely mainly on height, the Southern English listener more on length.

Our hypothesis, then, is that optimal cue reliance depends on cue reliability.

Boersma's (1998) three-grammar model of phonological acquisition, which implements a separate perception grammar within an Optimality-Theoretic framework, predicts exactly this hypothesised relation when a general Gradual Learning Algorithm (Boersma & Hayes, to appear) is applied to the perception grammar (Boersma 1997). This grammar consists of a large number of constraints such as “an F1 of 350 Hz is not an /I/” and “a duration of 80 ms is not an /i/”, for evry value of F1 and duration, and for both vowels. Thus, if both the spectral and durational continua are divided into seven discrete regions, the grammar contains 28 constraints.

We let this algorithm simulate the behaviour of two virtual babies, little Elspeth and little Liz, who are brought up in Scotland and Southern England, respectively. We will show how both of these listeners start with the same perception grammar, in which the constraints are ranked at the same level. The learners then repeatedly "hear" words with /I/ and /i/, appropriately distributed with respect to height and length according to their language environment. Every time little Elspeth or little Liz miscomprehends a word, she reranks some constraints in her perception grammar. Gradually, Elspeth comes to rely almost exclusively on height, whereas Liz comes to rely on both duration and height, thus achieving adult-like minimisation of perceptual confusion.

We also let the algorithm simulate the behaviour of two adult Spanish speakers, Isabel and Carmen, who move to Scotland and Southern England, respectively. The results show that a listener with an established perception grammar will also rerank the constraints and can ultimately attain native-like perception.

Consequently, the formal three-grammar model of Functional Phonology, together with the Gradual Learning Algorithm, accounts for the acquisition of an effective perception of sound contrasts.

REFERENCES Boersma, Paul (1997): How we learn variation, optionality, and probability. Proceedings of the Institute of Phonetic Sciences, University of Amsterdam, 21: 43- 58. [= ch.15 of Boersma 1998]

Boersma, Paul (1998): Functional Phonology. Doctoral thesis, University of Amsterdam. The Hague: Holland Academic Graphics.

Boersma, Paul & Bruce Hayes (to appear): Empirical tests of the Gradual Learning Algorithm. Linguistic Inquiry, Winter 2001.

Escudero, Paola (to appear a): The perception of English vowel contrasts: acoustic cue reliance in the development of new contrasts. In Jonathan Leather & Allan James (eds.) Proceedings of the 4th International Symposium on the Acquisition of Second- Language Speech, New Sounds 2000, University of Amsterdam.

Escudero, Paola (to appear b): Input, L1, and universal strategies in the development of new contrasts: the L2 relative weighting of acoustic information. Paper presented at the 25th Annual Boston University Conference on Language Development, November 2000.