Phonology Seminar Diane Brentari Purdue University City University

Phonology Seminar Diane Brentari, Purdue University City University & DCAL, June 14, 2006

Preliminaries l What is the difference between phonetic, phonological, and morphological elements in SL? examples l l l Selected fingers Joints Non-selected fingers “bent-B” and “straight B” forms of. . . l l (ASL) know, (BSL) vehicle

1. How to tell the difference l Phonetics, phonology, and morphology are not mutually exclusive!! l Consider a pair of forms that are minimally different for the element in question l Ask if there is a difference in meaning: Y/N -- word level & in what context – Y/N -- morphological level & in what context –

2. How to tell the difference l If there is no morphological or lexical difference in meaning in any context, the property is phonetic. l Sometimes phonetic properties can be indicators of dialectal variation

3. How to tell the difference l If there is morphological or lexical differences between the two minimally different forms, delineate the contexts in these differences occur l Typically it is the task of the phonologist to determine the systematic nature of these differences. (Consider the spoken-L example from English “s”)

Goals of a phonological model for cognitive science 1. provide structures analogous to those of spoken languages to use experimentally for psycholinguistic comparisons between signed and spoken languages -- e. g. , the syllable, the segment, the feature, the word 2. provide a basis for explaining modality differences between signed and spoken languages 3. provide a basis for explaining the neural mapping of language in SLs 4. provide a testable model for language acquisition and language breakdown

Goals of a phonological model for linguistics 1. Create an inventory of structures in the phonology of a language or language family : features, syllables, morae, words, phrase types. Which ones does a given language use? 2. Provide a means of distinguishing between elements that are phonetic, morphological and phonological using phonological processes and specify the phonologically relevant properties and only those. (e. g. , What is the morpho-phonemic status of each property of phonological structure “F” in ASL? ) 3. Create the simplest structure and set of rules/constraints for describing a particular language or language family. a. 2 -handed signs and constraints on them b. classifiers and how they differ from core lexical items c. the manual alphabet and words derived from it vs. lexical items.

Today’s goals: 1. Which structures do we need? --the syllable and related issues 2. What Modality differences are in these structures --syllable, segment, word, foot 3. How these knowledge of these structures helps in designing good psycholinguistic experiments? --word segmentation

Structure of the Prosodic Model

Heavy vs. light syllables l Distinction between heavy and light syllables as movements with more than one component (ASL examples)

1. Evidence for the syllable The role of the heavy and light syllables in word order changes; heavy syllables gravitate to clause-final position. Example: BOOK, 1 GIVE 3 JOHN *BOOK 1 GIVE 3 (continuative) JOHN *BOOK 1 GIVE 3 (habitual) JOHN

2. Evidence for the syllable l Nominal reduplication occurs only in stems with light syllables: Evidence from ASL: SIT/CHAIR, CLOSE-WINDOW/WINDOW, GO-BY-PLANE/AIRPLANE, SUPPORT, DEBT, NAME, APPLICATION, ASSISTANT – *DREAM, *THROW, *CATCH, *LEARN –

One example of a modality effect: Morpheme-to-syllable ratio monomorphemic polymorphemic monosyllabic Chinese sign languages polysyllabic English, German, Hawaiian West Greenlandic Turkish, Navajo,

simultaneity: 6 morphemes in one syllable 2+bent-over + upright-beings + go-forward + side-by-side + with care

How to use the syllable in psycholinguistic experiments l Choose a phenomenon well-studied in spoken languages. Why? l Use a design with a precedent in spoken languages, but tailor it for sign data. This can be easy or difficult. l Utilize the linguistic structures on which we have the most consensus. Why? l In this case we will look at “word segmentation. ”

Word segmentation l What it is. . . ? l What it is not: Pauses between words – Lexical access – Exclusively based on the “segment” –

What influences word segmentation in SLs and Sp-Ls? l Language experience – l Modality effects – l Expected result would be that speakers would use their Sp-L strategies to segment sign Expected result would be that speakers use the same (or similar) strategies to segment sign as signers do General properties of UG – The properties exploited for word segmentation in SLs and Sp-Ls are the same.

For Sp. Ls and SLs, we know. . . SLs Sp-L General Word segmentation segment Yes//less ? Yes//more Yes//night rate vs. nitrate syllable Yes//less ? Yes//more Yes//it involves rhythm foot ? ? Yes//crucial word Yes// a lot ? Yes//Harmony effects

Some signs are disyllablic: DESTROY (ASL)

The syllable has a function in SLs, but how much? l Not all languages use all units equally. l Some rely more heavily on the word (Finnish), or on the mora (Japanese), or on the foot (English) or on the syllable (French). l Not only that, but languages use different units for different purposes. l demo. JSL, demo. HZJ, demo. BSL

Experimental Design l Native signers (13); native English Speakers (13) l Nonsense movie clips (168), based on principles of word formation in ASL) l Forced choice task “Is this 1 sign or 2 signs? ”

Hypotheses l Signers will have sharper judgments than non-signers about where the break between 2 signs belongs; language experience will matter. l Speakers will not use the rules of Sp-Ls for segmenting an SL; there will be a modality effect l Segmentation in signed and spoken languages require different strategies; the strategies will not be a part of UG

Stimuli: 168 nonsense forms: 5 HS x 6 M x 2 POA counterbalanced to create cue conflict

MOVEMENT conditions 1. 1 movement: one movement in the form 2. * or +hs an illicit combination of orientation change+handshape change . * path+path: a combination of two illicit path movements (e. g. straight+arc) 4. * path+hs an illicit combination of path+handshape change 5. * path+ or an illicit combination of path+orientation change 6. 2 legal movements: a grammatical combination of two path movements

HANDSHAPE conditions 1. 1 handshape: one handshape in the form 2. * 2 HSs (m) 2 handshapes in the form; both marked 3. * 2 HSs (u): 2 handshapes in the form; both unmarked 4. * 2 HSs (1+1): 2 handshapes in the form; one marked and one unmarked 5. 1 HS (aperture change): 2 handshapes in the form; both have the same selected fingers

Sample stimulus: cell 5/1 *path+or. D x 1 HS x 1 POA

Sample stimulus: cell 6/2 2 legal Ms x *2 HSs(m) x *2 POAs

Sample stimulus: cell 4/5 *path+HSD x 1 HS(ap D) x 1 POA

RESULTS (ANOVAs) Main effects: -PARAMETER (HS, M, POA)—significant l F(3, 72)=55. 8, p <. 001) -STIMULUS VALUE (1 - vs 2 -values)—significant l F(1, 24)=42. 5, p <. 001) -GROUP (signing, non-signing)—not significant 2 -way interactions -STIMULUS x PARAMETER F(3, 72)=6. 78, p <. 001 (HS, M, POA) -GROUP x PARAMETER F(3, 72)=2. 9 , p <. 04 (HS only)

Mean percentages of 2 -sign judgments for 1 - and 2 - value stimuli

Mean percentages of 2 -sign judgments across all HS conditions

Mean percentages of 2 -sign judgments across all movement conditions 1 mov 2 mov HS PATH OR PATH 2 mov HS PATH 2 mov 2 mov OR PATH

Results Regarding language exposure, signers will have sharper judgments than non-signers about where the break between 2 signs belongs, so there is an increased sensitivity, but the grammar of SLs matters only for HS. (slight effect of language exposure) l Speakers do not use Sp-L strategies to segment the sign stream For visual languages the basic strategy is 1 value=1 word (Modality effects) l Segmentation in signed and spoken languages require different strategies. (Not UG) l

Implications for SL phonology HS is special in sign language phonology; --signers pay more attention to it than non-signers in word segmentation --more categorical --between gesture and sign there is a big difference in HS inventory • There is a stronger modality effect in POA and MOV. This means that signers and nonsigners use this property. •

Implications for Word segmentation in signed & spoken languages (1) Word segmentation strategies in signed and spoken languages are different. spoken language word segmentation relies heavily on rhythmic cues--trochaic feet (children, breakfast). This is more “syllable-based. ” sign languages use domain based cues, which are more word-based (1 value=1 word).

Implications for Word segmentation in signed & spoken languages (2) Signers approach the task differently Signers paid attention first to movement most, then handshape, then place of articulation. Non-signers paid attention first to movement most, then place of articulation, and ignored handshape.

Future directions Is modality conditioning itself part of UG? -Do hearing babies and deaf babies have the same modality specific skills at birth? -How much of our modality conditioning influences how languages are formed?

Acknowledgments • • Petra Eccarius Robin. Shay Stefan Goldschmidt Pradit Mittrapiyanuruk Sam Supalla Ronnie Wilbur Purdue University Faculty Scholars Fund Thank you!!