High Vowel Devoicing in Japanese

Phonology Forum 2014
High Vowel Devoicing in Japanese:
As an Indicator of Standardization of Dialect
PhD Lecture
日本語狭母音の無声化
ー共通語普及の指標としてー
Hi-Gyung Byun
Akita International University
August 20th / The University of Tokyo, Komaba
Contents
• Introduction
• Purpose of the study
• Methodology (materials, acoustic analysis)
• Results (Inoue Data, Takada Data)
• Discussion (standardization of dialect)
2
Introduction
• Vowel devoicing in Japanese
• Distribution map of vowel devoicing
• Issues regarding the current map
• Resolving the issues
3
INTRODUCTION
Vowel Devoicing in Japanese
• In Japanese, there is a regional difference in occurrence
of vowel devoicing. For example:
• Tokyo Japanese devoices a high vowel between voiceless
consonants with high frequency, while Osaka Japanese
devoices with very low frequency in the same condition
あきた
a k i̥ t a
devoiced vowel [ i̥ ] by Tokyo Japanese Speaker
あきた
akita
voiced vowel [ i ] by Osaka Japanese Speaker
• The map of vowel devoicing (Hirayama, 1960; Nihon Hoso
kyokai, 1966, 1998) shows the regional difference
4
INTRODUCTION
Distribution Map of Vowel Devoicing
The ONLY map of vowel devoicing
Prominent areas of vowel devoicing
Non-prominent areas of vowel devoicing
Hokuriku
Izumo
Chugoku
Shikoku
Kyushu
Chubu
Northern Tohoku
Prominent areas
•
•
•
•
•
•
Southern Tohoku
Kanto
Some parts of Chubu
Hokuriku
Izumo
Kyusyu
Southern Tohoku
Kanto
Kinki
NHK Japanese language
pronunciation and accent dictionary
(Nihon Hoso Kyokai, 1966, 1998)
Non-prominent areas
•
•
•
•
•
Northern Tohoku
Many parts of Chubu
Kinki
Chugoku (except Izumo)
Shikoku
5
INTRODUCTION
Issues Regarding the Current Map
• This map is not made by investigating under a consistent methodology
throughout the country (Sawaki, 2005)
• The border lines are not clear, and the map needs to be made up
based on the words with the same sound environments (Mase,1977)
Some more issues…
• No criteria for dividing into two areas, prominent vs. non-prominent
(When we say “prominent,” how often does vowel devoicing occur?)
• No explanation for differences within the same (non) prominent areas
(Is a high vowel devoiced in the same condition for Kanto and Kyushu areas? Usually, Kanto area
often devoices a high vowel in word-medial position when a devoiceable vowel is not accented;
while Kyusyu area often devoices a high vowel in word-final position, irrespective of the accent )
• If the map is based on Kindaiti’s (1954) study, the data do not show
the current status of vowel devoicing
(Standardization of dialects (common Japanization) has progressed for several decades and is still
progressing, so we can easily imagine there will be some changes regarding vowel devoicing)
6
INTRODUCTION
Some issues
Resolving
on the
thecurrent
Issues map
• It is almost impossible that an individual investigates the entire country
by oneself (Sato, 1985)
• Recently, large-scale speech databases have been open for research
purpose
• This study made use of two different, large-scale, speech database
covering most parts of Japan and various age ranges
• One was collected between 1986 and 1988 from 41 prefectures that
included 607 participants (high school students and their grandparents),
⇒Inoue Data (It is appropriate to examine regional and generational differences)
• The other was collected between 2006 and 2007 from 7 areas that
included 463 participants ranging in age from 14 to 90 ⇒Takada Data
(It is a follow-up investigation to Inoue Data, and best for investigating age differences)
7
PURPOSE
Purpose of the Study
The purpose of this study is
• to comprehensively examine vowel devoicing emphasizing
on regional and age differences as well as phonetic
conditions, and
• to provide new insights into vowel devoicing in Japanese
Based on the result from devoicing rate by each age group in
each area, this study will suggest
• a new distribution map that reflects current status
• three stages of change of devoicing rate
8
PURPOSE
Purpose of the Study (continued)
Furthermore
• we will verify that synchronic generational differences in
devoicing rate is a sound change in progress
• and for some areas, we will identify the times that devoicing
rate is changed dramatically
We will also discuss
• social factors for change of devoicing rate in some areas, and
• the future devoicing rate by applying the current devoicing
rate of five areas to an S-curve model
9
Methodology
•
•
•
•
•
•
Inoue Data
Takada Data
Investigation points
Target words
Sound environments of target words
Acoustic analysis
10
METHODOLOGY
Inoue Data
• A part of large-scale speech corpus collected by support of KAKENHI
from 1986 to 1988 (Inoue 1989)
• All utterances were recorded on analog tapes by high school students
and sent to the principal investigator
• 22 words which include a high vowel between voiceless consonants
were selected from 607 participants in 41 prefectures and converted to
digital format, and then analyzed acoustically
• The number of participants from Tohoku to Kyusyu by generation:
a rea Aomori Aki ta
total
7
8
15
7
7
14
a rea
Shi ga
Na ra
old
3
3
6
2
2
4
old
young
young
total
Iwa te Ya ma g Mi ya gi Hukus h Gunma Tochi g Iba ra ki Chi ba Sai ta m Ka na ga Ni i gata Naga n Yaman Is hi ka Hukui
7
8
15
8
8
16
14
12
26
22
22
44
10
11
21
11
13
24
7
7
14
3
3
6
7
8
15
4
8
12
Os aka Wa ka y Hyogo Tottori Okaya Hi ros h Ya mag Ka gaw Tokus h Kochi
3
3
6
3
3
6
8
8
16
1
1
2
17
17
34
4
1
5
1
1
2
17
17
34
8
7
15
8
8
16
16
15
31
6
6
12
Ehi me Hukuo
11
11
22
13
14
27
3
3
6
4
4
8
Oi ta
Sa ga
4
4
8
8
8
16
Gi hu
Ai chi
Shi zuo
Mi e
4
4
8
2
3
5
6
8
14
2
2
4
14
14
28
Naga s Kuma m Mi ya z Kagos h tota l
4
4
8
5
5
10
Hokkaido and some remote islands were excluded. No data in Tokyo, Kyoto, Toyama, and Shimane
5
5
10
10
12
22
299
308
607
11
METHODOLOGY
Takada Data
• A part of large-scale speech corpus collected by support of Hakuho
Research Grant between 2006 and 2007 (Takada2009)
• The recordings were made face-to-face for each participant using a
computer to digitally record utterances
• 34 words which include a high vowel between voiceless consonants
were selected from 463 participants in 7 areas and analyzed acoustically
• The number of participants for each area by age group:
area
Akita
Tochigi
Ibaraki
Tokyo
23 wards
Tama district
Osaka
Hyogo
Awaji Island
Kumamoto
total
Hyogo
60's
18
12
11
6
3
1
2
2
9
64
50's
13
7
10
15
1
7
5
2
10
70
40's
11
5
6
4
2
6
4
1
5
44
30's
13
2
7
5
2
5
8
2
5
49
20's
13
4
18
18
10
8
19
1
16
107
10's
17
11
28
17
8
20
19
2
7
129
I participated in the recording of Akita, Ibaraki, Tokyo, Osaka, and Hyogo
total
85
41
80
65
26
47
57
10
52
463
12
METHODOLOGY
Investigation Points
• Inoue Data
○ Takada Data
13
METHODOLOGY
Target Words
• Utterances in isolation
• Phonetic condition is word-medial vowel devoicing, which a high vowel is
devoiced almost regularly in Tokyo Japanese (average devoicing rate is over
80%). The target words are the words that include a high vowel between
voiceless consonants and the high vowel has no accent kernel
• Word-final vowel devoicing, which a high vowel comes between a voiceless
consonant and a pause, is NOT the condition that a high vowel is devoiced
regularly (Byun 2007)
65 64 56
53
Devoicing rate (%) in word-final position
44 41 40 40
38 38 36 35 34
31 31 31 29 26 26
24 22 21 21 21 20 20
18 18 17 17 16 16 15 15 15 14 13
12 10
3 2
Hiroshima
Nara
Kagoshima
Osaka
Hyogo
Saitama
Yamanashi
Yamaguchi
Tottori
Tokushima
Ishikawa
Wakayama
Iwate
Kochi
Nagano
Shiga
Ehime
Akita
Nagasaki
Kagawa
Miyagi
Oita
Okayama
Hukuoka
Hukui
Saga
Hukushima
Gihu
Yamagata
Ibaraki
Shizuoka
Niigata
Aomori
Tochigi
Mie
Gunma
Kanagawa
Chiba
Aichi
Miyazaki
Kumamoto
Average of 10 words あき(秋),えき(駅),ひらく(開く),まつ(松),でぐち(出口),しゅじゅつ(手術),あいさつ(挨拶),るす(留守),りす(栗鼠),なす(茄子)
METHODOLOGY
Sound Environments of Target Words
Inoue Data
Takada Data
(22 words)
(34 words)
following consonant
following consonant
affricate
くし(串), くすり(薬)
affricate
ちち(父), ちち(乳)
つくる(作る)
つち(土)
ふたり(二人), ふた(蓋), ポスト
ふく(服), たすき(襷)
ふし(節)
plosive
きく(菊)
きさく(気さく)
くさ(草)
くち(口), くつ(靴)
ちから(力)
ちくわ(竹輪)
くし(串)
ごちそう(ご馳走)
ちちおや(父親), ちつづき(地続き) みちすじ(道筋)
つかれ(疲れ)
つくり(作り)
fricative
だいきち(大吉), きつえん(喫煙) れきし(歴史)
くとう(苦闘)
くき(茎)
fricative
した(した)
preceding consonant
くさ(草)
ちから(力), ちかい(近い), おちた(落ちた)
fricative
preceding consonant
れきし(歴史)
くち(口), くつ(靴)
affricate
あきた(秋田)
きしゃ(汽車)
plosive
きた(来た)
plosive
fricative
affricate
plosive
あつさ(暑さ)
つち(土), つつみ(包み)
うつし(写し)
しかた(仕方)
たのしさ(楽しさ)
しくみ(仕組み) しちじ(七時), しつれい(失礼)
オアシス
ふたり(二人)
ふさく(不作)
ふく(服)
ふち(淵), ふつう(普通)
ふし(節)
target following
vowel
vowel
i
i
u
u
i
i
u
u
i
i
u
u
non-high
high
non-high
high
non-high
high
non-high
high
non-high
high
non-high
high
: no words
The underline represents a target syllable (preceding consonant + target high vowel). The crossed out words ちつづき and ふち were
excluded from the analysis due to a problem in recording.
15
METHODOLOGY
Acoustic Analysis
fully voiced vowel
partially voiced vowel
devoiced vowel
2% or so on average for each area
•
•
•
fully voiced vowel: several periodic waves and several glottal pulses on the spectrogram
partially voiced vowel: quasi-periodic waves and weak glottal pulses on the spectrogram
devoiced vowel: no waves and no visible glottal pulses for a vowel on the spectrogram
devoicing rate (%) = the number of devoiced vowels / target vowels ×100
16
Main Results
• Description for each area and comparison
with the previous study
• New distribution map of vowel devoicing
from Inoue Data
• Three stages of change of devoicing rate
from Takada Data
• Synchronic generational differences of
devoicing rate from Takada Data
17
RESULT
Result
• The details of the devoicing rate for each
area are described by generation or age
groups in terms of five phonetic factors:
preceding consonant, following
consonant, target high vowel, following
vowel, and word accent ⇒⇒ ⇒⇒
Example of description of each area
The description part takes up two-thirds of the dissertation.
I will not mention the details here
Here, we focus on the results below:
• New distribution map
• Three stages of change of devoicing rate
• Synchronic differences of devoicing rate
as a diachronic change
18
RESULT
New Distribution Map
Current map
Inoue Data
old generation
(Birth year is 1930 and older)
Devoicing rate
is over 60%
especially prominent
or regularly devoiced
Prominent areas of vowel devoicing
Non-prominent areas of vowel devoicing
Estimated percent is used
for the area with no data
NHK Japanese language
pronunciation and accent dictionary
(Nihon Hoso Kyokai, 1966, 1998)
19
RESULT
New distribution map (Inoue Data)
Old generation
Young generation
(Birth year is 1930 and older)
(Birth year is between 1967 and 1973)
Clear generational and regional differences!
20
RESULT
Three Stages of Change of Devoicing Rate
abstract model
Initial stage
Middle stage
・High vowels between fricative and
plosive, fricative and affricate,
affricate and plosive are devoiced first
100%
Final stage
・High vowels in the three environments are devoiced nearly 100%
・High vowels of the other environments are devoiced gradually
100%
plosive
affricate
fricative
0%
plosive
affricate
fricative
e.g. Osaka Japanese speakers
in their 40s and older
・High vowels are devoiced almost
100 % except high vowels between
fricative/affricate and fricative
・The overall average devoicing rate
is over 80%
100%
plosive
affricate
fricative
0%
plosive
affricate
fricative
e.g. Kumamoto Japanese speakers
in their 40s and older
Akita Japanese speakers
in their 60s and older
plosive
affricate
fricative
0%
plosive
affricate
fricative
e.g. Tokyo Japanese speakers
in all age groups
21
devoicing rate %
devoicing rate %
• Devoicing rates in four areas are
increasing roughly with age
• With Akita and Kumamoto, younger
generation’s devoicing rate has
reached over 80 %, the final stage,
which is the same stage as Tokyo
• Hyogo and Osaka’s devoicing rates
are under 60% in all age groups
• As a result, a regional as well as a
generational difference remains
devoicing rate %
• Five representative areas
• Devoicing rate in Tokyo is over 80% in
all age groups
devoicing rate %
Takada Data
devoicing rate %
RESULT
100
80
TOKYO
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
AKITA
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
60
40
20
KUMAMOTO
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
HYOGO
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
OSAKA
60
40
20
• Now, can we call this synchronic
0
1900
generational difference a diachronic change?
1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
22
Discussion
•
•
•
•
Cohort analysis & apparent time
Age, period and cohort effects
Devoicing rate of same cohort
The line of demarcation of devoicing rate
•
•
•
•
Social factors of common Japanization
Factors for change of devoicing rate
S-curve model
Future devoicing rate
23
DISCUSSION
Cohort Analysis & Apparent Time
• A cohort is a group of people who show a common characteristic within a
certain period (e.g. a group of people who were born in 1970 and educated in the same area)
• In the studies of change in progress, the notion of “apparent time” has generally
been used for explaining a diachronic change from synchronic data
• “Apparent time” assumes that the grammar (including phonetics and phonology)
of individuals becomes relatively stable after the end of the language acquisition
period and remains the way of speaking during their lifetime. Thus, speech
differences among people of different ages reflect characteristics they acquired
in their formative years (Cedergren 1987, Chambers 2002)
• From this, we assume that same birth cohort shows great similarity in the use of
language at any time of investigation ⇒cohort effect
• There are two more effects: age effect, where people change as they get older,
and period effect, where society changes affect people regardless of their age
• Especially, the age effect (age-grading) must be differentiated from cohort effect
when we deal with a diachronic change
24
DISCUSSION
Age, Period and Cohort Effects (Glenn 1977)
age effect only
AGE
%
60s 50s 40s 30s 20s 10s
1960
40
50
60
70
80
90
1970
40
50
60
70
80
90
1980
40
50
60
70
80
90
1990
40
50
60
70
80
90
2000
40
50
60
70
80
90
2010
40
50
60
70
80
90
… 60
70
80 …
・
The value of AGE (column) is changing
…60 70 80…
・
The value of YEAR (row) does NOT change
・
The value of COHORT (skew) is changing
40… 60 70 80…
YEAR of investigation
YEAR of investigation
%
period effect only
cohort effect only
AGE
%
60s 50s 40s 30s 20s 10s
1960
40
40
40
40
40
40
1970
50
50
50
50
50
50
1980
60
60
60
60
60
60
:
60
1990
70
70
70
70
70
70
70
2000
80
80
80
80
80
80
2010
90
90
90
90
90
90
80
:
YEAR of investigation
fictitious devoicing rate
AGE
60s 50s 40s 30s 20s 10s
1960
20
30
40
50
60
70
1970
30
40
50
60
70
80
1980
40
50
60
70
80
90
1990
50
60
70
80
90
100
2000
60
70
80
90
100 100
2010
70
80
90
100 100 100
… 60 70
・
The value of AGE does NOT change
・
The value of YEAR is changing
…60 70 80…
・
The value of COHORT is changing
40… 60 70 80…
80 …
・
The values of AGE and YEAR are
changing
・
The value of between COHORTs is
changing… 60 70 80…
・The value of within COHORT does
NOT change …60 60 60…
Two of the three factors are related to each other
25
DISCUSSION
Age, Period and Cohort Effects
fictitious devoicing rate
AGE effect
AGE
%
60s 50s 40s 30s 20s 10s
COHORT effect
AGE
60s
50s
40s
30s
20s
10s
%
AGE
60s
50s 40s 30s 20s 10s
50
60
70
80
90
1960
40
40
40
40
40
40
1960
20
30
40
50
60
70
1970 40
50
60
70
80
90
1970
50
50
50
50
50
50
1970
30
40
50
60
70
80
1980 40
50
60
70
80
90
1980
60
60
60
60
60
60
1980
40
50
60
70
80
90
1990 40
50
60
70
80
90
1990
70
70
70
70
70
70
1990
50
60
70
80
90 100
YEAR
1960 40
YEAR
YEAR
%
PERIOD effect
One point in time in an apparent-time study, for example, 1970
• AGE effect is likely to be judged as a diachronic change because the values are
changing with age. On the other hand, we will not confuse PERIOD (year) effect
with a diachronic change because of no change with age
Two points in time, for example, 1970 and 1980
• There is no change between two times in AGE effect ⇒ age-grading
• There is a change between two times in COHORT effect ⇒ diachronic change
●
We can confirm whether a generational difference in synchronic data is
age-grading or a diachronic change by comparing data with two or more
different points in time
26
DISCUSSION
Devoicing Rate of Same Cohort
devoicing rate %
• Unfortunately, due to insufficient data,
it is impossible to compare the two data
sets by each cohort
○:Inoue Data
• It suggests that age difference in
devoicing rate of each area is not agegrading, but a sound change in progress
devoicing rate %
• No statistically significant differences
between 1970 cohorts of the two data
sets in four areas
devoicing rate %
• 1970 cohorts in the two data sets show
a striking similarity between average
devoicing rates
devoicing rate %
• We focus on 1970 (1966-1975) cohorts
100
80
+ :Takada Data
AKITA
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
60
40
20
KUMAMOTO
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
HYOGO
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
OSAKA
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
DISCUSSION
• Osaka is likely to have the line in 1970
• Hyogo might have the line in 1970
( Evidence from results of word accents )
• There is no clear line of demarcation in
Tokyo as noted in previous studies
• The difference among areas can be
explained in terms of standardization of
dialects (common Japanization)
devoicing rate %
100
80
TOKYO
60
40
No ○ marks in Tokyo
20
0
devoicing rate %
100
80
AKITA
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
devoicing rate %
• Akita and Kumamoto have the line of
demarcation approximately 1950 and 1970
respectively
+:Takada Data
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
100
80
60
40
20
KUMAMOTO
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
devoicing rate %
• Individual devoicing rate in five areas of
Takada Data
• Some areas’ devoicing rate increased
dramatically, and others gradually
100
80
HYOGO
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
devoicing rate %
The Line of Demarcation
○:Inoue Data
100
80
OSAKA
60
40
20
0
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
DISCUSSION
Social Factors of Common Japanization
• Inoue (1985) and Yoneda (1997) suggested some social factors of common
Japanization from the study on the standardization of modern Japanese in
Tsuruoka city, conducted by the National Language Research Institute in 1950,
1971, and 1991
• Three main factors are (1) education of standardization, (2) improvement of
transport, (3) acquisition of Tokyo Japanese through TV programs
• These social factors can also explain why the lines of demarcation of Akita
and Kumamoto exhibit around 1950 and 1970. Akita is to do with (1),
Kumamoto (2) (3)
• As for Osaka and Hyogo, these factors seem to have been little effect
29
DISCUSSION
Factors for Change of Devoicing Rate
•
•
•
•
•
Standard Japanese education of “spoken” language began in earnest in elementary school
after 1941, and Akita (largely Tohoku) was among the areas where educators tried
aggressively to correct thick accents. It seems that the result of education appeared from
the 1950 birth cohort
Kumamoto is likely to be affected by transport and TV. During the years of steep economic
growth, transport had improved tremendously, increasing a wide range of interaction with
others
The Royal Wedding in 1959 and the Tokyo Olympic Games in 1964 carried a TV penetration
rate to over 90 % in 1965. Yoneda (1997) says Tokyo Japanese through television must have
given an impulse to the lower age groups, which are the most sensitive to language
Acquisition of the Tokyo accent of younger generation led to an increase in devoicing rate
Here are word accents produced by Kumamoto Japanese speakers and their devoicing rates:
100%
Types of word accent (total 100 %)
12%
86%
0%
60’s
29%
100%
29%
91%
67%
71%
50’s
40’s
87%
92%
LH
HL
HH
9%
30’s
13%
20’s
8%
10’s
The vowel
of the first
syllable is
the target
vowel
0%
-100%
Devoicing rate by word accent (each 100%)
100%
60%
58%
6%
15%
3%
60’s
50’s
40’s
88%
87%
83%
25%
28%
20%
30’s
20’s
10’s
LH
HL
30
DISCUSSION
S-curve Model
Percent adoption
(Devoicing rate)
100%
0%
S-shaped curve of language change
slow
(Aitchison 2001)
quick
If the generational difference in devoicing
rate is a sound change, the devoicing rate
would increase in line with an S-curve
quick
slow
Time
100
80
In many cases, but not all, a diffusion
process follows an S-shaped curve
60
TV
color TV
car
CD player
40
The penetration rate of durable goods based on Consumer0
Confidence Survey 2004 by Cabinet Office, Government
of Japan
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
20
DISCUSSION
Future Devoicing Rate
• A vertical dotted line represents the final stage
①TOKYO
100
Devoicing rate %
• The devoicing rate of the areas in
the process of change is approaching
the final stage on a wave of common
Japanization
• The younger age groups of Akita and
Kumamoto have reached the final stage,
and will maintain the status like Tokyo
• Osaka and Hyogo are in the process
of change, adopting devoiced vowels
(an upward trend)
80
③KUMAMOTO
60
40
②AKITA
④HYOGO
20
⑤OSAKA
30 years
about 80 years
0
60s 30s
50s 20s
40s 10s
30s 30s
20s 20s
10s 10s
60s 50s 40s
60s
50s
40s
60s 50s 40s 30s 20s 10s
④⑤
① 60s
60s 50s 40s 30s 20s 10s
② 60s
50s 40s 30s 20s 10s
50s 40s 30s 20s 10s
③ 60s 50s 40s 30s 20s 10s
• If Osaka keeps changing rapidly like Akita or
Kumamoto, younger speakers would reach the
final stage in 30 years, and older speakers in about 80 years
• Hyogo might take more time since it is more sluggish than Osaka
32
References
•
•
•
•
•
Aitchison, J. (2001) Language change: progress or decay? 3rd Edition. Fontana Press.
Cedergren, H. J. (1987) The spread of language change: verifying inferences of linguistic diffusion. In P. H.
Lowenberg (ed.), Language Spread and Language Policy: Issues, Implications and Case Studies. pp. 44-60.
Washington, DC: Georgetown University Roundtable on Language and Linguistics.
Chambers, J. K. (2002) Patterns of Variation including Change. In J. K. Chambers, P. Trudgill and N. Schilling-Estes
(eds.), The Handbook of Language Variation and change. pp. 349-372. Oxford: Basil Blackwell.
Glenn, N. D. (1977) Cohort analysis. SAGE Publications.
Yoneda, M. (1997) Survey of standardization in Tsuruoka, Japan: comparison of results from three surveys
conducted at 20-year intervals. Japanese linguistics 2, 24-39.
•
井上史雄(1985
1985)『新しい日本語-《新方言》の分布と変化-』明治書院 (Inoue, 1985)
•
井上史雄(1989)『日本語音声の地域差・世代差の音韻論的・音響学的分析』
昭和63(1988)年度科学研究費助成金 総合(A)研
究成果報告書 (Inoue, 1989)
•
NHK放送文化研究所編(1998)『NHK 日本語発音アクセント辞典 新版』日本放送出版協会 (Nihon Hoso Kyokai, 1998)
•
金田一春彦(1954)「音韻」
『日本方言学』吉川弘文館, 87-176 (Kindaichi, 1954)
•
佐藤和之(1985)「方言の音声と音韻」『新しい方言音声』(加藤正信編)至文堂, 29-46 (Sato, 1985)
•
•
沢木幹栄(2005)「周辺的要素の地域差」『日本方言研究会 第80回研究発表会 発表原稿集』89-92 (Sawaki, 2005)
高田三枝子(2009)『日本語語頭有声破裂音におけるVOTの地理的・世代的多様性』2005年度第1回博報「ことばと文化・教育」研
究助成研究成果報告書 (Takada, 2009)
•
日本放送協会編(1966)『
日本語発音アクセント辞典』日本放送出版協会 (Nihon Hoso Kyokai, 1966)
•
平山輝男編(1960)『全国アクセント辞典』東京堂 (Hirayama, 1960)
•
邊姫京(2007)「狭母音の無声化の全国的地域差と世代差」『日本語の研究』
3-1, 33-48 (Byun, 2007)
•
邊姫京(2012)『日本語狭母音の無声化―共通語普及の指標として―』東京大学大学院総合文化研究科, 学位論文
•
馬瀬良雄(1977)「東西両方言の対立」『岩波講座 日本語11 方言』岩波書店, 235-289 (Mase, 1977)
33
Thank you
Phonology Forum 2014