Generational Change in Japanese Prosody: A Sociophonetic Analysis of Pitch Leveling

1
Generational Change in Japanese ProsodyA
Sociophonetic Analysis of Pitch Leveling

• Shoji Takano
• Hokusei Gakuen University
• Ichiro Ota
• Kagoshima University
• The 16th Sociolinguistics Symposium
• University of Limerick
• July 5-8, 2006

2
Background

• Takeshi Shibatas (1995) Informal Observation on
  Recently Emerging Changes in Japanese
• General preference for non-prominence among the
  younger generation because they want to convey
  images of novelty, freshness and urbanity in
  speech (p.181-2). For example,
• (1) Leveling of lexical accents
• e.g., kareshi boyfriend ? kareshi ( x
  indicates a pitch accent.)
• (2) Leveling of sentence intonation or pitch
  in general (p. 185)
• It may be fair to assume that sentence-level
  pitch leveling (2) precedes lexical accent
  leveling (1) (p. 185).
• However, it should be noted that the change in
  question is considered specific to Tokyo
  Japanese. We do not know whether the change has
  become widespread in local dialects as well (p.
  186).
• No follow-up study has been conducted
• no empirical evidence that verifies Shibatas
  observation is available today.

3
Takano Otas (2005) follow-up study

• Research Questions
• (1) empirically examined Shibatas hypothesis
  on sentence- level
• pitch leveling prevalent among the younger
  generation.
• (2) explored the possibility that the change in
  question is a nation-wide phenomenon.
• Data
• ? Japanese is a pitch accent language.
• ? 3 read sentences that consist only of accented
  (kifuku-shiki)
• words. (/ xx / indicates an accentual
  phrase.)
• (1) /Basu wa/ /nai kara/ /aruite/ /ikoo./
• (Lets walk since there is no bus service.)
• (2) /Doryoku/ /shitemo/ /imi ga/ /nai./
• (Its meaningless even if you make an effort.)
• (3) /yomeba/ /yomuhodo/ /yukai na/ /hanashi
  da./
• (The deeper you read, the more amusing you find
  the story.)

4

• Hidaka Subjects
• 20 Hidaka regional dialect speakers in 2 age
  groups all are natives in coastal areas of
  Hidaka (including the towns of Shizunai and
  Mitsuishi)
• Sapporo Subjects
• 20 Sapporo dialect speakers in 2 age groups
• Kagoshima Subjects
• 15 native informants in the southern Kyushu
  region

5

• Analytical Procedures
• ? How can we identify the leveling of pitch?
• ? In an utterance involving the leveling of
  pitch
• (1) Fundamental frequency (F0) of every
  accented phrase (AP) is weakened i.e., every
  pitch peak is relatively lower than otherwise.
• (2) The movement of pitch involves a steady
  declination due to the
• above-mentioned weakening of pitch accent.
• ? How can we measure the leveling of pitch?
• (see Figure 1)
• Step 1 Measuring the pitch range (PR) of each
  utterance as the
• benchmark ?
• Maximum F0 - Minimum F0 PR
• Step 2 Measuring the height of F0 peak for
  every AP ?
• Step 3 Measuring the relative decrease of F0
  peak (divided by PR)
• F0 peak of 1st AP? - F0 peak of 2nd AP, 3rd AP,
  4th AP, etc.
• F0 declination
• Step 4 Measuring the relative magnitude of
  pitch rise (divided by PR)
• AP-final F0 ? - F0 peak of the following AP
  F0 increase

6

• Figure 1 /Doryoku//shitemo//imi ga//nai/ by
  a 65-year-old Shizunai woman
• The utterance involving pitch leveling should
  show
• (1) The value of F0 declination is high
• Pitch declination curve is relatively steep and
  steady (without prominent pitch rise) toward the
  end of an utterance.
• (2) The value of F0 increase is low
• F0 rise of every AP is small-scale in its
  magnitude.

?
?
?
?
7

• Results
• In 2 (S-2, S-3) out of 3 sentences, the
  differences in both F0 declination and F0
  increase were found statistically significant
  between the age groups.
• For example
• Sentence 3 /yomeba//yomuhodo//yukai
  na//hanashi da/
• F0 declination ( divided by PR)
• /yomeba/ ? /yomuhodo/ /yomeba/ ? /yukai
  na/
• Younger Group -37.4 -36.7
• Older Group -10.3 -11
• p lt .05 p lt .01
• F0 increase ( divided by PR)
• /yomeba/ ? /yomuhodo/ /yomuhodo/ ?
  /yukai na/
• Younger Group 14.9 35
• Older Group 51.4 64.9
• p lt .01 p lt .05

8

• Group-specific Patterns of Pitch Contour in
  Sentence 3
• 1) Older Speakers (blue yellow) consistency
  in prominent pitch rise mild declination with
  pitch reset
• 2) Younger Speakers (red, skyblue, purple)
  relatively flat without prominent pitch rise
  steady pitch declination
• 3) The younger groups from three different
  dialect areas (Sapporo skyblue, Shizunai red,
  Kagoshima purple) share a relatively levelled
  movement of pitch.

9
The Present Study

• Research Questions
• (1) To reconfirm Takano Otas (2005) findings
  (i.e.,
• based on read sentences) by analyzing a more
• naturalistic style of speech elicited from news
  
• passage reading
• (2) To test native speakers perception of
  pitch
• movement as a social indexicality for the
  speaker
• age
• Data for Research Question (1)
• ? Reading of transcriptions of actual news
  passages
• read by anchorpersons on popular news
  programs
• broadcast in Japan
• ? Subjects 10 speakers in late 50s to early 70s
  from
• Sapporo (Sapporo-OLD), 10 speakers in
• early 20s from Sapporo (Sapporo-YOUNG), and 11
  
• speakers from Kagoshima (Kagoshima-YOUNG)

10

• Data for Research Question (2)
• ? Pseudo-Matched-guise Experiment
• with 30 Japanese college students
• as raters
• Procedures
• - The raters listen to 14 stimulus utterances
  (consisting of 2 different read sentences) three
  times but in different orders.
• - The raters are asked to guess
  approximately how old each of 14 speakers is
  (i.e., 10s, 20s,
• 30s, 40s, 50s or higher), based on their
  first impressions.
• Experimental Design
• - 14 stimulus utterances contain 6 target
  utterances (i.e., utterances for anlaysis) and 8
  non-target utterances (i.e., distractors).
• - 3 speakers who exhibited pitch movement
  typical to the younger group (i.e., target
  speakers) were chosen 8 speakers were also
  chosen as the distractors. All the 11 speakers
  are in their early 20s, the members of our
  Younger Groups (i.e., Hidaka and Sapporo
  speakers).
• - Each of the 3 target speakers produced 2
  utterances in different occasions one with their
  original pitch contour typical to the younger
  generation, and the other with a pitch contour
  synthesized by Praat, so that it could exhibit
  pitch movement typical to the older generation.
• ? The raters think that they are listening
  to 14 different speakers, but in actuality they
  are
• listening to 11 different speakers 8
  non-target distractors and each of the 3
• target speakers twice (i.e., the
  matched-guise), namely, the first time with an
  original
• pitch contour sounding younger and the
  second time with a synthesized pitch contour
  sounding

11

• Results for Research Question (1)
• Even in a passage reading task, there are found
  consistent differences in the degree of pitch
  declination between the two age groups (Sapporo
  Older vs. Sapporo Younger Kagoshima Younger).
• For example,
• Part of the news passage /kodomotachi
  no//mondai koodoo no//haikei o saguroo to/
• F0 declination ( divided by PR)
• ? ? ? ? ? ?
• /kodomotachi/ ? /koodoo/
  /kodomotachi/ ? /saguroo/
• Sapporo Younger -24.3
  -66.3
• Sapporo Older -15.7 -41.5
• p lt .05 p lt .02
• /kodomotachi/ ?
  /koodoo/ /kodomotachi/ ? /saguroo/
• Kagoshima Younger -20.7
  -72.3
• Sapporo Older -15.7 -41.5
• p lt .05 p lt .01

12
?
?
?

• Two younger groups patterns of pitch declination
  ( ? to ?, ? to ? in Sapporo-YOUNG and
  Kagoshima-YOUNG) are more radical than that of
  the older group (Sapporo-OLD).

13

• 2) The pitch contours of Sapporo-YOUNG and
  Kagoshima-YOUNG are quite similar.
• The differences in pitch declination are NOT
  statistically significant between the two groups
  from different dialect areas.
• F0 declination ( divided by PR)
• /kodomotachi/ ? /koodoo/
  /kodomotachi/ ? /saguroo/
• ? ? ? ? ? ?
• Sapporo Younger -24.3
  -66.3
• Kagoshima Younger -20.7 -72.3
• NS NS
• Notes
• These statistically significant age-linked
  differences have been found mainly in the degree
  of pitch declination. We have not obtained such
  consistent findings yet as to the degree of pitch
  rise. We leave this issue for further
  investigation.

14
Results for Research Question (2)Pitch contours
appear to function as a cue indexing the
speakers age.

• For example,
• ? the majority of raters
• judged Speaker B, a female
• speaker in her early 20s
• from Shizunai (Hidaka), as
• someone in her 30s or
• younger when speaking in
• her original, relatively flat,
• greatly declining pitch
• contour.
• ? The identical speaker,
• however, was judged mainly
• as someone in her 40s or older
• when speaking in a modified
• contour of pitch typical to the
• older generation.

15
? The remaining 2 target speakers were also
judged more or less in a similar fashion.
16
(No Transcript)
17
Summary Discussion

• (1) Pitch Leveling is Prevalent among the
• Younger Generation in Japan
• reconfirmed based on the speech elicited
  from two production tasks independent sentence
  reading and passage reading
• OLD - Less F0 declination (and more F0
  increase)
• - Realization of pitch accents is sure and
  steady.
• - F0 is vigorous and dynamic in its movement
  with
• relatively prominent pitch peaks and even
  with frequent
• pitch reset.
• YOUNG - More F0 declination (and less F0
  increase)
• - Realization of pitch accents is
  subject to catathesis
• ( a feature of the Tokyo dialect).
• - F0 is flat with relatively minor pitch
  peaks and
• accentual de-generation.

18

• (2) Pitch Leveling in the Speech of Younger
  Generation Could be a Nation-wide Phenomenon.
• a striking similarity in pitch movement
  shared by the speakers from three different
  regions Hidaka, Sapporo, and Kagoshima
• Is this one of the proccesses of
  standardization of regional dialects that
  has been in rapid progress widely
• in Japan?
• Is this a consequence of peoples exposure
  to media (e.g., TV, radio broadcast)?

19

• (3) Pitch Movement (especially, declination)
  Indexes the Speakers Age.
• native speakers are capable of perceiving
  differences in magnitude of
• pitch, and are also aware of social
  information (i.e., the speaker age) conveyed by
  particular types of pitch.

20
References

• Foulkes, Paul. (2005). The Social Life of
  Phonetics and Phonology. Plenary Address at
• the 5th UK Language Variation and Change
  Conference. University of Aberdeen.
• Foulkes, Paul Docherty, Gerard. (1999). Urban
  Voices Accent Studies in the British Isles.
  London Arnold.Kubozono, H. (1993). The
  Organization of Japanese Prosody. Tokyo Kurosio.
• Labov, W. (1972). Sociolinguistic Patterns.
  Philadelphia, PA University of Pennsylvania
  Press.
• Pierrehumbert, J., Beckman, M. (1988). Japanese
  Tone Structure. Cambridge, MA The MIT Press.
• Sibata, T. (1978). Shakai-gengogaku no kadai
  (Issues in sociolinguistics). Tokyo Sanseido.
• ________. (1995). Nihongo wa omoshiroi (Japanese
  is interesting). Tokyo Iwanami.
• Takano, S., Ota, I. (2005). A Sociolinguistic
  Study of Pitch Leveling in Japanese A
  Preliminary Analysis. Poster Presentation at UK
  Language Variation and Change 5, University of
  Aberdeen.
• Thomas, E. R. (2002). Instrumental phonetics. In
  J. Chambers, et al. (eds.), The Handbook of
  Language Variation and Change. New York
  Blackwell.
• ________. (2002). Sociophonetic applications of
  speech perception experiments. American Speech
  77(2) 115-47.