Diagnosing phonological categories - University of Manchester · 2011-10-25 · Diagnosing...

Diagnosing phonological categories

Patrycja Strycharczuk

University of Manchester

9th Sept., 2011

What defines phonology?

Phonology defined by categoricity

Phonology defined by contrast

Contrastive 6= categorical

Voicing 6= spreading

Outline





Two aspects of speech

Phonology Phonetics

symbolic representations; physical representations;allow idealizations of continuous in time and space;

temporal chunking (segmentation), internal temporal structurequalitative categorization (labels), allows overlap;timelessness; quantitative values on multiple independent dimensions;

rules manipulate rulesfeatures and feature values, interpret feature values in time and space,associations; can be gradient.

thus phonological rules canbe category changing,produce static changes over whole segments,be lexical/cyclic.

Keating (1996, 263), based on Keating (1988); Cohn (1990)

Categoricity/gradience distinction

I Models of the phonology-phonetics interface(Pierrehumbert, 1980; Keating, 1988, 1996, 1990; Cohn,1990, 1993);

I Idealised model (Pierrehumbert et al., 2000; Myers, 2000;Scobbie, 2005; Cohn, 2007);

I Empirical research (Barry, 1992; Cohn, 1993; Zsiga, 1995;Holst & Nolan, 1995; Nolan et al., 1996; Ellis & Hardcastle,2002);

I Evidence from acquisition (Maye et al., 2002, 2008;McMurray et al., 2009);

Predictions/consequences

I Phonological categories signalled by areas of stabilitywithin continuous phonetic space.

I Phonetic rules are not category-changing (category change= phonological rule).

I Redundancy (phonology duplicating phonetics).

Outline





Toronto school

I Avery et al. (2008); Hall (2006); Dresher (2009)

I Contrastivist Hypothesis:The phonological component of a language L operates onlyon those features which are necessary to distinguish thephonemes of L from one another. (Hall, 2006, 20)

Enhancement

I Stevens et al. (1986); Stevens & Keyser (1989); Keyser &Stevens (2006); Stevens & Keyser (2010)

I Stevens & Keyser (2010):I Underlying representations are entirely feature-based and

contain only distinctive features.I Differences between underlying and surface representations

are mainly due to strategies of enhancement and overlap,which introduce, delete or extend gestures, but do notoperate on features.

Outline





Ecuadorian Spanish /s/-voicing

I Word-final /s/ becomes voiced before a vowel or a sonorantconsonant in the next word./gas#akRe/ [ga.za.kRe] ‘acrid gas’/gas#noble/ [gaz.no.Ble] ‘noble gas’

I Regressive voice assimilation to voiced and voicelessobstruents/gas#kaRo/ [gas.ka.Ro] ‘expensive gas’/gas#blaNko/ [gaz.BlaN.ko] ‘white gas’

gas acre

a s a

gas noble

a s n

Individuals’ realisations of /s/-voicing

Speaker 1

Fricative duration (ms)

Voi

cing

(ms)

-20020406080100120

0 20 40 60 80 100 120

Speaker 2


Voi

cing

(ms)

-20020406080100120

0 20 40 60 80 100 120

Speaker 3


Voi

cing

(ms)

-20020406080100120

0 20 40 60 80 100 120

Speaker 4


Voi

cing

(ms)

-20020406080100120

0 20 40 60 80 100 120

Speaker 5

Fricative duration (ms)V

oici

ng (m

s)-20020406080100120

0 20 40 60 80 100 120

Speaker 7


Voi

cing

(ms)

-20020406080100120

0 20 40 60 80 100 120

Speaker 8


Voi

cing

(ms)

-20020406080100120

0 20 40 60 80 100 120

Data from 7 native speakers of Quito Spanish on /s/ voicing in theprevocalic (gas acre) and pre-sonorant (gas noble) environments. Theellipses represent 95% confidence intervals for /s/ voicing before voiced andvoiceless stops.

Two types of speakers

I ‘Central’ speakers: broad range of variation, use the centreof the phonetic space defined by voicing duration andfricative duration;

I ‘Peripheral’ speakers: categorical variation, tend to voicecategorically, avoid the centre of the phonetic space.

Modelling voicing duration and voicing ratio in QuitoSpanish

I Two mixed effects models:I of voicing duration in /s/-voicing environmentsI of voicing ratio in /s/-voicing environments

I Random effects:I SpeakerI Word

I Fixed effects:I manner of articulation (sonorant vs. vowel) of the segment

following the sibilant;I speech rate (normal vs. fast);I phonetic strategy of the speaker (central vs. peripheral).

Speech rate effects

I Speech rate manipulation as a test ofcategoricity/gradience (Sole, 1995, 2007).

I Varying degrees of voicing duration across speech rates →categorical.

I Varying degrees of voicing ratio across speech rates →gradient.

Speech rate effects

3035

4045

Speech rate

Voi

cing

dur

atio

n (m

s)

Fast Normal

central

-

-

-

-

cat

peripheral

-

-

-

-

0.5

0.6

0.7

0.8

0.9

Speech rateV

oici

ng ra

tio

Fast Normal

central

-

-

-

-

cat

peripheral-

-

-

-

Interaction between speaker type and speech rate on voicing duration (left)

and voicing ratio (right).

/s/-voicing and overlap

I [z] is not a phoneme in Quito SpanishI voicing contrast lost in Spanish sibilants around 16c.

(Robinson 1979, and references therein)

I If phonological output cannot contain features which arenot distinctive, /s/-voicing must be phonetic (gesturaloverlap).

I Consequences:I Quito inter-speaker variation not predicted.I Theory does not distinguish between central and peripheral

types of speakersI Categorical behaviour as accidental (contra evidence from

language change and acquisition).

/s/-voicing and the Contrastivist Hypothesis

I Contrastive [voice] could spread to sibilants from alaryngeally contrastive segment.

I But there is no voicing contrast in sonorants.

I Hall (2006): [SV] ([Sonorant Voice]) can be used in afeature hierarchy to distinguish sonorants from voicedobstruents.

I supported by disparate laryngeal behaviour of sonorantsubclasses

I /v/ in Russian and Czech does not trigger RVA, butundergoes it.

I /rfi/ in Czech does not trigger RVA, but undergoes bothprogressive and regressive devoicing.















Some problems with [Sonorant Voicing]

I Contrastive specification determined based on phonologicalactivity.

I Crucially assumes that pre-sonorant voicing consists inspreading.

Outline





Pre-sonorant voicing in West-Flemish

I Word-final fricatives undergo voicing before avowel/sonorant consonant in the next word./zEs ja:r/ [zEz.ja:r] ‘six years’

I Word-final stops surface as voiceless before avowel/sonorant consonant in the next word./Axt ja:r/ [Axt ja:r] ‘eight years’

I Stops and fricatives are laryngeally contrastive (in onsets),sonorants are not.

I Both, stops and fricatives, undergo voicing before voicedstops.

Voicing tail

I Instances of gradient voicing in West-Flemish invariablyinvolve voicing in the initial part of the segment.

ɑu t w

Spreading from sonorants?

I Word-final fricatives tend undergo categorical, but optionalvoicing word-finally before a sonorant/vowel.

I Word-final stops before a sonorant/vowel surface either asvoiced, or with limited gradient voicing.

I No increased phonetic voicing in West-Flemish pre-sonorantstops (p=0.73 compared to voicing before voiceless stops)→ sonorant voicing not phonologically active.

I Passive voicing always in the initial part of the closure.

Fricative voicing as perceptual reanalysis

Model based on Ohala (1981)

No re-interpretation in stops

Influences on perception

I Coda fricatives showI a reduced oral gesture;I a lower oral pressure build-up;I lower velocity of air through the oral constriction;I a less intense frication (Sole, 2010).

I Shorter and less intense frication are cues to voicing infricatives (Stevens et al., 1992; Ladefoged & Maddieson,1996).

I Word-final pre-sonorant fricatives may be perceived asvoiced by listeners.

I Stops less likely to undergo the change.

Predictions

The diachronic perceptual reinterpretation hypothesis:

I captures the stop/fricative asymmetry;

I involves categorical voicing which is not spreading;

I is difficult to reconcile with the Contrastivist Hypothesis.

Summary

I Top-down approaches to the phonetics-phonology divisiondo not converge with a bottom-up distinction based oncategoricity and gradience.

I Empirical problems:I Enhancement and overlap cannot capture the distinction

between categorical and gradient allophony.I The Contrastivist Hypothesis cannot accommodate feature

insertion which is not spreading.

I Misguided use of Ockham’s Razor.

I Phonetic evidence is crucial to establishing what counts asa phonological category.

Acknowledgments

I Thanks to Ricardo Bermudez-Otero, Yuni Kim, and KoenSebregts for comments and discussion.

I The Quito Spanish data have been collected in acollaborative project with Marijn van ’t Veer, MartineBruil and Kathrin Linke (Leiden University).

I The West-Flemish data have been collected in acollaborative project with Ellen Simon (Ghent University).

I The research is supported by the Arts and HumanitiesResearch Council grant no. AH/H029141/1.

References

Avery, P., B. E. Dresher & K. D. Rice (eds.) (2008). Contrast in Phonology:Theory, Perception, Acquisition. Berlin: Mouton de Gruyter.

Barry, M. C. (1992). Palatalisation, assimilation and gestural weakening inconnected speech. Speech Communication 11, pp. 393–400.

Cohn, A. C. (1990). Phonetic and phonological rules of nasalization. Ph.D.thesis, UCLA.

Cohn, A. C. (1993). Nasalisation in English: phonology or phonetics.Phonology 10, pp. 43–82.

Cohn, A. C. (2007). Phonetics in phonology and phonology in phonetics.Working Papers of the Cornell Phonetics Laboratory 16, pp. 1–31.

Dresher, B. E. (2009). The Contrastive Hierarchy in Phonology. Cambridge:Cambridge University Press.

Ellis, L. & W. J. Hardcastle (2002). Categorical and gradient properties ofassimilation in alveolar to velar sequences: evidence from EPG and EMAdata. Journal of Phonetics 30, pp. 373–396.

Hall, D. C. (2006). The Role and Representation of Contrast inPhonological Theory. Ph.D. thesis, University of Toronto.

Holst, T. & F. Nolan (1995). The influence of syntactic structure on [s] to[S] assimilation. Connell, B. & A. Arvaniti (eds.), Phonology and phoneticevidence: papers in Laboratory Phonology IV, Cambridge: CambridgeUniversity Press, pp. 315–333.

Keating, P. A. (1988). Underspecification in phonetics. Phonology 5, pp.275–292.

Keating, P. A. (1990). Phonetic representations in a generative grammar.Journal of Phonetics 18, pp. 321–334.

Keating, P. A. (1996). The phonology-phonetics interface. Kleinhenz, U.(ed.), Interfaces in Phonology, studia grammatica 41, Berlin: AkademieVerlag, pp. 262–278.

Keyser, S. J. & K. N. Stevens (2006). Enhancement and overlap in thespeech chain. Language 82, pp. 33–63.

Ladefoged, P. & I. Maddieson (1996). The Sounds of the World’sLanguages. Cambridge, MA: Blackwell.

Maye, J., J. F. Werker & L. Gerken (2002). Infant sensitivity todistributional information can affect phonetic discrimination. Cognition82, pp. B101–B111.

Maye, J., W. D. J & R. N. Aslin (2008). Statistical phonetic learning ininfants: facilitation and feature generalization. Developmental Science11, pp. 122–134.

McMurray, B., R. N. Aslin & J. C. Toscano (2009). Statistical learning ofphonetic categories: isnights from a computational approach.Developmental Science 12, pp. 369–378.

Myers, S. (2000). Boundary disputes: The distinction between phonetic andphonological sound patterns. Burton-Roberts, N., P. Carr & G. J.Docherty (eds.), Phonological Knowledge, Oxford: Oxford UniversityPress, pp. 245–272.

Nolan, F., T. Holst & B. Kuhnert (1996). Modelling [s] to [S]accommodation in English. Journal of Phonetics 24, pp. 113–137.

Ohala, J. J. (1981). The listener as a source of sound change. Masek, C.,R. A. Hendrick & M. M. F (eds.), Papers from the Parasession onLanguage and Behavior, Chicago: Chicago Ling. Soc., pp. 178–203.

Pierrehumbert, J., M. Beckman & D. Ladd (2000). Conceptual foundationsof phonology as a laboratory science. Burton-Roberts, P. Carr &G. Docherty (eds.), Phonological Knowledge, Oxford: Oxford UniversityPress, pp. 273–303.

Pierrehumbert, J. B. (1980). The phonology and phonetics of Englishintonation. Ph.D. thesis, MIT.

Robinson, K. L. (1979). On the voicing of intervocalic s in the Ecuadorianhighlands. Romance Philology 33, pp. 132–143.

Scobbie, J. (2005). The phonetics-phonology overlap. QMUC SpeechScience Research Centre Working Paper 1.

Sole, M.-J. (1995). New ways of analyzing sound change. speech rate effects.Belgian Journal of Linguistics 9, pp. 21–44.

Sole, M.-J. (2007). Controlled and mechanical properties in speech: areview of the literature. Sole, M.-J., P. S. Beddor & M. Ohala (eds.),Experimental approaches to phonology, New York: Oxford UniversityPress, pp. 302–321.

Sole, M.-J. (2010). Effects of syllable position on sound change: Anaerodynamic study of final fricative weakening. Journal of Phonetics 38,pp. 289–305.

Stevens, K., S. Keyser & H. Kawasaki (1986). Toward a phonetic andphonological theory of redundant features. Perkell, J. S. & D. H. Klatt(eds.), Invariance and variability in speech processes, Hillsdale: LawrenceErlbaum, pp. 426–449.

Stevens, K. N. & S. J. Keyser (1989). Primary features and theirenhancement in consonants. Language 65, pp. 81–106.

Stevens, K. N. & S. J. Keyser (2010). Quantal theory, enhancement andoverlap. Journal of Phonetics 38, pp. 10–19.

Stevens, K. N., S. E. Blumstein, L. Glicksman, M. Burton & K. Kurowski(1992). Acoustic and perceptual characteristics of voicing in fricativesand fricative clusters. Journal of the Acoustical Society of America 91,pp. 2979–3000.

Zsiga, E. (1995). An acoustic and electropalatographic study of lexical andpostlexical palatalization in American English. Connell, B. & A. Arvaniti(eds.), Phonology and phonetic evidence: Papers in LaboratoryPhonology IV, Cambridge: Cambridge University Press, pp. 282–302.

Date post:	22-May-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Diagnosing phonological categories - University of Manchester · 2011-10-25 · Diagnosing...

Documents