Gradient Lengthening Effects: Evidence from Tagalog

Afton Coombs University of Southern California acoombs@usc.edu GLOW2015 4/18/15

1

Research Questions

2

!  Can qualitative and quantitative effects in language emerge from the same grammar, using the same formalism?

!  Specifically: can qualitative and quantitative effects in the prosodic domain be captured by the same formalism?

Dynamical Systems

3

!  Differential equation determines how a state changes based upon the previous state

!  System itself is constant !  Potential function = x2/2

Nonlinear Dynamical Systems

4

!  A force function: f(x) = x3 −x !  Corresponding potential: V(x) = x4/4 – x2/2 !  Differential equation determines how a state changes

based upon its current state. !  Capture a number of biological functions, such as hand

coordination and gait (Haken, Kelso, Bunz 1985)

!

Previous Proposals for Dynamical Systems as a Language Grammar

5

!  Incomplete word-final devoicing in German (Port and Crawford 1989, Gafos and Benus 2006) !  Word final voiced obstruents not completely neutralized to

voiceless obstruents !  Interaction between input faith and markedness of a final

voiced obstruent

!  Tashlhiyt Berber tonal alignment (Röttger et al 2013) !  Rightward tendency of H tone in a certain sentence tune –

both a final syllable preference and late alignment within the vowel

!  Crucially contrasts with other sentence tunes that align H tone earlier in both syllable and nuclear segment

Determining the attractor layout

6

!  Control parameter changes the system overall: V(x) = (a*x4)/4 – (b*x2)/2 + c*x

!  Parameters a and b determine the number and location of stables states

!  Allows double or single wells (simplest formalism)

a=1, b=3, c=0 (bistable) a=0, b=-1, c=0 (monostable)

Determining the Attractor Layout

7

!  a and b parameters determine the location and strength as well as number of stable states

a=1, b=3, c=0 a=1, b=7, c=0 (attractors = -2, 2) (attractors = -3,3)

Determining the Attractor Layout

8

!  Parameter c determines tilt V(x) = (a*x4)/4 – (b*x2)/2 + c*x

!  a=3, b=9, c=0 (no tilt)

!  a=3, b=9, c=3 (left tilt)

Connecting Model with Data

9

!  Simulating pattern with an added noise factor produces expected probability distribution

!  This can be compared to actual experimental distributions

Predicted Probability Distributions

10

Left tilt a=1, b=3, c=3

Right tilt a=1, b=3, c=-3

Righter tilt a=1, b=3, c=-6

Background: Lengthening Effects !  Domain-final lengthening well-attested cross-linguistic

phenomenon !  Word boundaries (Oller 1973, Klatt 1976) !  Phrase boundaries (Byrd and Saltzman 1998, Wightman 1992)

!  Stress also linked to length in some languages (Hayes 1995)

11

Where Lengthening Effects Do Not Apply

!  Avoiding footing/stress of a domain-final syllable (NonFinality Constraints: Prince and Smolensky 1993)

!  Avoiding an “overlong” syllable (Alaskan Yupik: Hayes 1995)

!  No rhythmic lengthening of a phrase-final syllable (Chickasaw: Munro and Ulrich 1984; Gordon and Munro 2007; Gordon 1999, 2004)

12

Tagalog: No Lengthening of a Final Syllable?

!  Generally allows stress on either the penult or ultima !  Traditionally analyzed as showing increased duration of a

stressed penult, but not of a stressed ultima (Schachter and Otanes 1982, Soberano 1980) !  ˈCVː.CV !  CV. ˈCV !  *CV. ˈCVː

13

Nonlinear Dynamical Systems and Vowel Duration

14

!  Two categories (long/ short) captured by two modes !  Continuous change within a mode allows for a principled

range of values Short Vowel Long Vowel

Standardized Duration Standardized Duration

Nonlinear Dynamical Systems and Vowel Duration

15

Short vowel a=1, b=3, c=3

Long vowel a=1, b=3, c=-3

Longer vowel a=1, b=3, c=-6

Standardized Duration

Standardized Duration Standardized Duration

Hypothesis !  Stressed ultimate vowels in Tagalog do show increased

length compared to unstressed ultimate vowels !  Ultimate vowels in Tagalog show a lesser stress increase

compared to penultimate vowels !  Penultimate vowels when stressed undergo a mode shift

16 Standardized Duration Standardized Duration

Hypothesis !  Stressed ultimate vowels in Tagalog do show increased

length compared to unstressed ultimate vowels !  Ultimate vowels in Tagalog show a lesser stress increase

compared to penultimate syllables !  Ultimate vowels when stressed show shift in the same mode

17 Standardized Duration Standardized Duration

Hypothesis: Predicted Probability Distribution of Tagalog Vowels

18

!  Stressed ultimate vowels in Tagalog do show increased length compared to unstressed ultimate vowels

!  Ultimate vowels in Tagalog show a lesser stress increase compared to penultimate syllables

Standardized Vowel Duration

Experimental Method

19

!  Ten words of type CaCa !  Five with penultimate stress and five with ultimate stress !  C = stop in all words

!  Two carrier phrases !  One placing target word in medial context

!  Gusto kong sabihin mo ang ____ na matahimik. !  want I say you <link> “____” that quietly. !  I want you to say “____” quietly.

!  One placing target word in final context !  Gusto kong sabihin mo ang "tabi" hindi "____.” !  want I say you <link> “side” not “____.” !  I want you to say “tabi” not “____”

Examples of Target Items !  Phrase-medial ultimately stressed

!  Gusto kong sabihin mo ang “taka” na matahimik. !  I said say “taka” quietly. !  “taka” = TAka ‘bamboo sticks’

!  Phrase-medial penultimately stressed !  Gusto kong sabihin mo ang “taka” na matahimik. !  “taka” = taKA ‘surprise’

!  Phrase-final ultimately stressed !  Gusto kong sabihin mo ang "tabi" hindi ”taka.” !  I said say “tabi” not “taka.”

!  Phrase-medial ultimately stressed !  Gusto kong sabihin mo ang "tabi" hindi ”taka."

Method: Task !  5 blocks of targets and fillers in pseudorandomized order

!  20 CaCa targets in both carriers !  38 fillers in both carriers

!  6 native speakers of Tagalog participated !  Participants shown one slide presenting the word, its

meaning, and its pronunciation. !  Then a second slide displaying the target phrase they are

asked to say !  Participants spoke into a head-mounted microphone

Results

22

0

20

40

60

80

100

120

140

Unstressed Penult Stressed Penult

Mean Vowel Duration (ms) by Stress Condition (Penult)

*

*

* = < 0.05

Results

23

0

20

40

60

80

100

120

140

Unstressed Ultima Stressed Ultima

Mean Vowel Duration (ms) by Stress Condition (Ultima) *

*

* = < 0.05

Tagalog Data

24

Distribution of Standardized Values of Vowel Durations (ms)

Standardized Vowel Duration (ms)

Num

ber o

f Tok

ens

-2 0 2 4 6 8

05

1015

20

Setting the Control Parameters: c values

Stressed Unstressed

Penultimate syllable -3 3

Ultimate syllable -9 -3

Penultimate = 1.5 Ultimate = -4.5 Unstressed = 1.5 Stressed = -4.5 (a = 3, b = 0.5)

25

Model of Experimental Results

26

"Short mode c=3 Unstressed penult

Long mode c=-3# Stressed penult

"Long mode c=-3 Unstressed ultima

Long(er)mode c=-9# Stressed ultima

Standardized Vowel Duration

Standardized Vowel Duration

Standardized Vowel Duration

Standardized Vowel Duration

Probability Distribution of All Possible Vowel Durations in Tagalog

27

Standardized Vowel Duration

Examples of Other Predicted Languages

28

!  A categorical positional effect but no categorical stress effect

!  A categorical stress effect but no categorical positional effect

Examples of Other Predicted Languages

29

!  No stress effect – iambic languages (Cahuilla, Seiler 1965) !  Penultimate: -3 !  Unstressed: 0 !  Ultimate: 3 !  Stressed: 0

Stressed Unstressed

Penultimate syllable 3 3

Ultimate syllable -3 -3

Examples of Other Predicted Languages

30

!  No word effect (English, Harris and Umeda 1974; Japanese, Nakai 2014) !  Penultimate: 0 !  Unstressed: 3 !  Ultimate: 0 !  Stressed: -3

Stressed Unstressed

Penultimate syllable -3 3

Ultimate syllable -3 3

Conclusion !  Tagalog shows continuous effects of stress and duration

that parallel a categorical shift !  Such parallelism falls out naturally from this type of

grammar !  Potential for future analyses of prosodic systems (e.g.

Nava 2011)

31

References !  Byrd, D., and E. Saltzman. (1998). Intragestural dynamics of multiple phrase boundaries. Journal of Phonetics, 26:173-199.

!  Gafos, A. I., and S. Benus (2006). Dynamics of Phonological Cognition. Cognitive Science, 30:905-943

!  Gordon, M. (2004). A phonological and phonetic study of word-level stress in Chickasaw. International Journal of American Linguistics, 70:1-32.

!  Gordon, M. (1999). The intonational structure of Chickasaw. Proceedings of the 14th International Congress on Phonetic Sciences, 1993-1996.

!  Gordon, Matthew, and Pamela Munro (2007). A phonetic study of final lengthening in Chickasaw.

!  Harris, M.S. and N. Umeda. (1974). Effect of speaking mode on temporal factors in speech: vowel duration. Journal of the Acoustical Society of America, 56, 1016-1018.

!  Hayes, B. (1995). Metrical Stress Theory: Principles and Case Studies. University of Chicago Press.

!  Haken, Hermann, JA Scott Kelso, and Heinz Bunz. (1985). "A theoretical model of phase transitions in human hand movements." Biological cybernetics 51.5: 347-356.

!  Klatt, D. (1976). Linguistic uses of segmental duration in English: Acoustic and perceptual evidence. Journal of the Acoustical Society of America, 59:5:5:1208-1221.

!  Munro, P., and C. Ulrich (1984). Structure Preservation and Western Muksogean Rhythmic Lengthening. WCCFL 3: 101-202.

!  Nakai, S. (2014). An explanation for phonological word-final vowel shortening: Evidence from Tokyo Japanese. Laboratory Phonology 11/2013; 4(2):513 – 553.

!  Nava, E. and J. Tepperman. (2011). Modeling second language prosody acquisition: A dynamic systems theory approach. Rosetta Stone Lab.s.

!  Oller, D. K. (1973). The effect of position in utterance on speech segment duration in English. JASA, 54:5:1235-1247.

!  Port, Robert F. & Crawford, Penny (1989). Incomplete neutralization and pragmatics. German. Journal of Phonetics, 17, 257-282.

!  Prince, Alan & Paul Smolensky (1993). Optimality Theory: constraint interaction ingenerative grammar. Ms, Rutgers University & University of Colorado, Boulder.Published 2004, Malden, Mass. & Oxford: Blackwell.

!  Röttger, T., R. Ridouane, and M. Grice (2013). Phonetic alignment and phonological association in Tashlhiyt Berber. International Conference on Acoustics 19.

!  Schachter, P., and F. Otanes (1972). Tagalog Grammar. University of California Press.

!  Soberano, R. (1980). The dialects of Marinduque Tagalog. Pacific Linguistics B-69. Canberra: The Australian National University.

!  Turk, A., and S. Shattuck-Huffnagel (2007). Multiple targets of phrase-final lengthening in American English words. Journal of Phonetics, Vol. 35, No. 4.

!  Wightman, C., et al. Segmental durations in the vicinity of prosodic phrase boundaries. JASA, 91:3:1707-1717.

32