Content uploaded by Yukinori Takubo
Author content
All content in this area was uploaded by Yukinori Takubo on Apr 08, 2019
Content may be subject to copyright.
Mutual Intelligibility as a measure
of linguistic distance and intergen
erational transmission
Yukinori Takubo
National Institute for Japanese Language and Linguistics
ytakubo@ninjal.ac.jp
kikigengo.jp
1
Approaches to Endangered Languages in
Japan and Northeast Asia: Description,
Documentation and Revitalization
Aug. 8th, 2018 NINAJL Lecture Hall
Contents
•Endangered Languages in Japan
•Mutual Intelligibility tests as a means to m
easure distance between two languages
and the degree of intergenerational
transmission
2
ENDANGERED LANGUAGES I
N JAPAN
3
How many languages are there
in Japan
•Until 2009 the ‘official’ answer was:
2
Namely
Ainu and Japanese
But according to UNESCO’s 2009 announce
ment 4
•in addition to Ainu and Japanese, the follo
wing 7 languages were included as distinc
t ‘languages’ spoken in Japan:
Hachijo, Amami, Kunigami (Northern Okina
wan), Okinawan, Miyako, Yaeyama and Yo
naguni
•UNESCO added that they all endangered.
5
1. Ainu
2.Hachijo
3. Amami
4. Kunigami
5. Okinawan
8Yonaguni
7Yaeyamay
6Miyako
Tsugaru
(Aomori)
Language endangerment
degree of intergenerational transmission
. safe language is spoken by all generations; intergenerational transmiss
ion is uninterrupted
2. vulnerable most children speak the language, but it may be restricted to c
ertain domains (e.g., home)
3. definitely endangered children no longer learn the language as mother t
ongue in the home
4. severely endangered language is spoken by grandparents and older ge
nerations; while the parent generation may understand it, they do not speak it to
children or among themselves
5. critically endangered the youngest speakers are grandparents and olde
r, and they speak the language partially and infrequently
6. extinct there are no speakers left
UNESCO Intangible Cultural Heritage - Endangered languages http://www.unes
co.org/culture/ich/index.php?pg=00139
7
Languages spoken in Japan
Standard Japanese: safe
Kunigami, Okinawan, Amami, Miyako, Hachi
jo: definitely endangered
Yaeyama(Ishigaki), Yonaguni: severely enda
ngered
Ainu: critically endangered
8
•All except for Ainu had been treated as ‘di
alects’ of Japanese until UNESCO .
•UNESCO classified those 8 varieties as ‘la
nguages’ distinct from standard Japanese
on the ground that they are mutually unint
elligible to each other.
9
What is the difference between
‘a language’ and ‘a dialect’
• A language is a dialect with an army and
navy. (cited by Max Weinreich as a remark made by
a member of the audience in one of his lectures Wikipedi
a)
=>The distinction is arbitrary.
•Criterion in modern linguistics: Mutual intel
ligibility
=>The distinction is gradient.
10
•It is quite improbable that UNESCO had a
n objective criterion that distinguished ‘lan
guage’ and ‘dialect’.
•They did not create an objective test that c
an measure ‘mutual intelligibility.’
•So that’s what we did!
11
2. OUR ATTEMPT AT MUTUAL I
NTELLIGIBILITY TEST
12
Mutual intelligibility test
The test created by O’Grady and his team fo
r measuring the mutual intelligibility between
Jeju, Seoul and Busan dialects of Korean
(O’Grady 2014).
•The subjects listen to a short story and an
swer 10 questions.
•Table 1 Percentage of correct responses t
o the comprehension questions. N=10 for
each
Jeju native
speakers
Seoul South Cellado South
Gyengsangd
o
89.21 12.03 6.00 5.26
Differences of our project from O’Grady et al.’s
Languages to be tested
•- 5~6 languages of the Ryukyus
•- 2, 3 languages of Mainland Japan
•- Needs better control of the test sentences to be
used
Intergenerational transmission
•Application of MI index to intergenerational trans
mission
Mutual intelligibility index
Our goal is to create an objective measure f
or mutual intelligibility: MI index
MI index can be used
1) to measure distance between two langua
ges or two varieties of a language,
or more importantly,
2) to measure intergenerational transmissio
n 15
MI Index
•MI index measures the distance between any two languag
es
•It can be used to make a (arbitrary) distinction between lan
guage and dialect.
MI Index less than 30=> different languages
MI index more than 30= dialects
•It can be used to measure intergenerational transmission
MI index less than 60=>revitalization possible
MI index more than 60=>revitalization difficult
MI index is independent from genealogic
al relations
•If there are many loan words shared in giv
en two languages, the value of MI index m
ay be high even if they are genetically unr
elated.
•The index does not represent the genealo
gical distance between two languages. Fo
r that purpose, shared innovation between
the two languages must be used.
Language distance measured by MI inde
x
•Previous index of the distance was measured by
vocabulary, which may not be a reliable index of
mutual intelligibility.
• The MI index measured by our methods can sim
ulate mutual intelligibility.
•The procedures are simple enough to be conduct
ed for endangered languages, where the speaker
s are sometimes very old.
•The index has to be measured in the same way f
or all pairs of languages.
Intergenerational transmission
•The MI index may be used to measure the
rate of intergenerational transmission.
•The MI index can thus serve as an objectiv
e criterion for the degree of endangerment.
•Some younger speakers have a potential of
understanding the language without actuall
y speaking it. MI index may be able to det
ect those potential speakers.
The design of our mutual intellig
ibility test
The participants listen to two short stories (each le
ss than two minutes) and answer 20 questions for
each story.
The stories must be culturally neutral, so the conte
nts can not easily be guessed from the context.
•We used Pair Story and Pingu. The form
er is a silent movie about a pear thief used
widely by linguists for eliciting discourse a
nd the latter a clay animation spoken in un
intelligible penguin speech.
•We first created a pair of test stories for ea
ch language (or a variety of a language) to
be tested.
21
e.g. Do Yonaguni speakers understand Yaeyama?
e.g. Do Aragusuku-Miyako speakers understand Ikema-
Miyako?
22
Results of two pilot tests
Distance between Ikema (a variety of Miyak
o) and Okinoerabu (a variety of Kunigami)
A: Okinoerabu speakers listen to Okinoera
bu (Baseline)
B: Okinoerabu speakers listen to Ikema
C: Ikema speakers listen to Ikema (Baselin
e)
D: Ikema speakers listen to Okinoerabu
B.
•Okinoerabu->Okinoerabu
23
Results: the tests conducted in
Okinerabu
C: Kamihirakawa(Mother tongue)
57 years old or older N= average ag
e 73 18.2 20 91%
57-36 N Average age 42
17.3/20⁼86.5%
D: Ikema
57 years old or older N=
0.82/20 4.1%
24
Results 1: test conducted in Nis
hihara Miyako
•A. Ikema (mother tongue)
65 years or older (N=10 Average age 71
Mean 19 25 20=96.25%
2. 50 years or younger(N=10 Average age 47.
8
Mean 18.40 20=92%
•B. Okinoerabu (different variety)
1. 65 years or older (N=10) 3.15 20=15.75%
2. 50 years or younger N=10 3.6 20 =18%
25
26
ikema speakers (old) ikema speakers (young) Okinoerabu speakers (old) Okinoerabu speakers (young)
96.25
92
4.09 0
15.75 17.88
91.14
86.59
MI index
ikema okinoerabu
Two varieties of Ryukyuan
•Okinoerabu→Ikema No mutual intelligib
ility
•Ikema→Okinoerabu: Almost no mutual int
elligibility
27
•Ikema people performed better probably b
ecause, Okinoerabu being a Northern Ryu
kyuan, is closer to Okinawan. There are p
eople who were exposed to Okinawan, a
prestigious language of the Ryukyus.
•The median score instead of mean score f
or Ikema people listening to Okinoerabu is
9.3% for older people, 18.125% for young
er people.
28
Tokyoites listening to Tsugaru (A
omori) variety of Japanese
•Tokyo people (university students) listenin
g to Tsugaru.
Mean 3.58/20=17.9%
•Tokyo people listening to Okinoerabu
Mean 1/20=5%
29
Further use of the tests
•Our tests (stories and questions) can be a
very good material for teaching and learni
ng the languages in question.
•Interesting stories (less than two minutes)
and 20 questiongs asking about the variou
s grammatical aspects of the stories.
30
•Tsugaru
32
So how many languages are the
re in Japan?
•If we call those varieties of Japanese who
se MI index in relation to other varieties is l
ess than 30, ‘languages’, the number of dif
ferent languages would be 50 to 60, may
be more.
•And they are all endangered, except for St
andard Japanese and possibly Osaka Jap
anese.
33
References
•Kikigengo.jp
https://www.facebook.com/kikigengojp
•O’Grady, W. 2014. Jejueo: Korea s Other Language. ʼPaper presented at t
he 7th World Congress of Korean Studies, Nov. 7, 2014. Honolulu, Hawaii.
•Takubo, Y. and T. Motoki. 2013. The Digital Museum Project for the Langua
ges and Cultures of Ryukyu: The Case of Ikema Ryukyuan. World Oral Liter
ature Occasional Lecture Series. University of Cambridge, Cambridge, UK.
http://www.dspace.cam.ac.uk/handle/1810/226634 Link downloadable fro
m academia.edu
•Takubo, Yukinori (2017) The Digital Museum project for the documentation of endan
gered languages: The case of Ikema Ryukyuan. Festschrift for John Whitman (2017).
Pingu
１１ Father, mother and their small son are havin
g dinner.
I) Nishihara (Ikema):
Zza=tu mma=tu bikidun vva=nu yui=yu fai-yui.
O) Hyoo muni(Okinoerabu):
Aca=tu ama=tu jingwa=nu kwaa=nu jii kad-un.
Question: What are the people in the story doing?
35
１１ When Mother tried to give a piece of s
weet potato to her son, he shook it off
I) Mma= a ffa=nkai nn=nu fii=d=di asI-kyaa,
ffa=a tii=hii harau-tai
O) Ama=ga inga=nu kwaa=ni umu kamasa-
ndi sjan-kyaa unu kwaa=wa uri tii=si uiharo-t
an.
What was the mother doing?
36
１１ The sweet potato got stuck on Mother’
s clothes.
I) Nn=na mma=ga cIn=kai tabarii-nyaan-b
a, mma=nu cIn=na , syana-fu nai-tai.
O) Umunu amaga kibarani taccicantu amag
a kibaranu juguritan.
•Why has the mother’s clothes become so.
37