6/1/2012

1

The transmission and diffusion of sound change across a hundred years

William Labov University of Pennsylvania

International Conference on Bilingualism and Comparative Linguistics CUHK 15 May 2012.

1

A definition of linguistic descent (transmission) as the major process in the family tree model

A language (or dialect) Y at a given time is said to be

descended from language (or dialect) X of an earlier time if

and only if X developed into Y by an unbroken sequence of

instances of native-language acquisition by children.

--Ringe, Warnow and Taylor 2002:63

2

Transmission the result of incrementation

• This is the normal type of internal language change, “change

from below,” which is responsible for increasing distances

between the branches of the family tree model over time.

• Such internal changes are generated by the process of

incrementation, in which successive cohorts and generations

of children advance the change beyond the level of their

caretakers and role models, and in the same direction over

many generations (Labov 1994: Ch. 14).

3

Some questions on the trajectories of sound change in progress

• Once a sound change in progress has been identified, can we trace it to its origin?

• Can we predict the future path of a sound change from its past trajectory?

• If not, what new explanations can be found for deviations from that path?

• What are the relative roles of phonetic features and lexical identities in the development of a sound change? [Wednesday]

Some answers to these questions will be derived by the application of new techniques of forced alignment and automatic vowel measurement to a large corpus of recordings of vernacular speech in Philadelphia.

4

LCV: The study of Language Change and Variation. Philadelphia, 1973-1979 Principles of Linguistic Change, Vol 2: Social Factors, 2001

ANAE: Labov, Ash & Boberg. Atlas of North American English, 1993-2001 (2006).

PNC: Philadelphia Neighborhood Corpus, 1972-2012.

Sources of data on the Philadelphia vowel system.

5

The study of Language Change and Variation [LCV] in Philadelphia, 1973-1979: the locations of six long-term neighborhood studies

6

6/1/2012

2

Apparent time changes in the vowel system of Philadelphia

ʌ

7

The Philadelphia Neighborhood Corpus [PNC] created from archives of LING560 “The Study of the Speech Community”

• Sociolinguistic interviews by students in “The Study of the Speech Community”

(W. Labov, G. Sankoff)

• Yearly, 1972-1992; Bi-yearly, 1994-2010:

– 59 neighborhood studies

– 1,087 recordings

• Time span:

– Dates of interviews: 38 years

– Dates of birth: 103 years

• Material transcribed to date:

– 49 neighborhoods

– 358 speakers

– over 150 hours of speech (average 29 minutes/speaker)

– 750,000 measurements of vowels > 50 msec duration

8

Locations of LING560 Studies, 1972-2010, transcribed and analyzed to form the Philadelphia Neighborhood Corpus

9

PNC subjects analyzed as of March 2012 by Age and Year of Interview [N=357]

0

10

20

30

40

50

60

70

80

90

100

1970 1980 1990 2000 2010 2020

Age

Year of interview 10

Hand measurement of the vowel system of John P., TS 516 a Philadelphia ANAE subject N= 321

11

The FAVE web site fave.ling.upenn.edu fave.ling.upenn.edu

12

6/1/2012

3

Results of automatic alignment and vowel extraction for Jean H., 60, Interviewed by J. Fruehwald in Group 2 of 2006 [N=7,101]

Is the gain in volume accompanied by a loss or gain in precision of measurement?

13

Number of vowels measured for two Philadelphia subjects: ANAE [John O’D, N=371] and PNC [Patrick M, Lock St.].

0

50

100

150

200

250

300

350

uw

riw ah

Kuw

uih

ro

yeh

rah

rTu

wey

Fo

hr oh

iyF

oy

ow

F*h

raw

oay

0æ

haw æ iy

C iey

C ^ e

Nu

mb

er

of

vow

els

me

asu

red

N(FAVE)

N(ANAE)

14

Standard error of the mean for F1 in analysis of two Philadelphia subjects: Jackson P., 35, N=321 [ANAE] and Patrick M., 65, N=6044] [PNC]

0

5

10

15

20

25

iw Kuw u ehr ahr ohr oh oy *hr aw o ay0 æh aw æ iyC i eyC ^ e

Seta

nar

d e

rro

r o

f th

e m

ean

fo

r F1

FAVE

ANAE

smean =s

N

15

Standard error of the mean for F1 in analysis of two Philadelphia subjects: Jackson P., 35, N=321 [ANAE] and Patrick M., 65, N=6044] [PNC]

0

5

10

15

20

25

iw Kuw u ehr ahr ohr oh oy *hr aw o ay0 æh aw æ iyC i eyC ^ e

FAVE

ANAE

smean =s

N

16

I. Unidirectional change in Philadelphia

17

Incrementation of a hypothetical sound change for females born in 1962, 1970, 1986 with progressively diminishing adult incrementation: cut-off point 17 years

b. 1986

b. 1970

b. 1962

r2=.998

logistic increment

inverse square increment

18

6/1/2012

4

Adolescent peaks in analysis by age for all white subjects [N=310]

Age

Dia

go

na

l

400

600

800

1000

1200

100 80 60 40 20 0

/eyC/

Female

Male

Age

F1

900

850

800

750

700

650

100 80 60 40 20 0

/ay0/

Female

Male

/eyC/ on front diagonal /ay0/ on F1

19

Measures of temporal development for /eyC/

Range Coefficient r2

Date of birth 106 yrs 3.30 5.8%

Age 90 yrs -2.02 3.4%

Year of interview 37 yrs 3.87 2.1%

20

Increasing height of /eyC/ in made, pain, etc. by Date of birth and by Sex by Higher Education

Date of Birth

Dia

go

na

l

600

800

1000

1200

1900 1920 1940 1960 1980

/eyC/

Female

Male

Date of Birth

Dia

go

na

l

600

800

1000

1200

1900 1920 1940 1960 1980

/eyC/

<=12

>12

21

Front upgliding vowels of Mary C., 63 [1972], Daley St. PH73-5-1

/iyC/

/eyC/

/ayv/

/oy/ FLEECE

FACE

PRICE

CHOICE

22

Raising along the front diagonal (F2 – 2 * F1) of /eyC/ in made, pain, etc. vs. stability of /eyF/ in may, mayor, male, etc. by Date of Birth for white adults [N=293]

Date of birth

Dia

go

na

l

0

200

400

600

800

1000

1200

1900 1920 1940 1960 1980

Vowel

eyC

eyF

23

For /ey/, De Camp [1933[ shows only [ɛɪ] for both checked and free. PEAS

[1961] uses a comparable notation [ɛi] for most vowels. . . Tucker [1944] states

that "In 'long a,' as in day, the first element ranges from [æ] to [a]", and

compares it to Cockney in this respect.

When did the raising of /eyC/ begin? After 1944.

24

6/1/2012

5

Raising of F1 of /ay0/ before voicess consonants in like, right, pipe, etc. vs. stability of /ayV/ in ride, mile my, etc. by Date of Birth for white adults [N-298] “

Date of birth

F1

1000

900

800

700

600

1900 1920 1940 1960 1980

Vowel

ay0

ayV

25

When did the raising of /ay0/ begin? after 1944.

For /ay/, PEAS (1961) does not differentiate /ay0/ from

/ayV/. . . nor does De Camp (1933). Tucker (1944) states

specifically that both /ay/ and /aw/ have identical nuclei

before voiced and voiceless finals, though they are quite

different in his own speech and "in most American dialects.”

26

II. The reversal of sound change in Philadelphia

27

Apparent time changes in the vowel system of Philadelphia

ʌ

28

[æʊt]

[ɛ:ot]

[e:ɔt]

Reversal of raising of /aw/ on the front diagonal by Date of Birth in Philadelphia Neighborhood Corpus [N=264]

Date of Birth

Dia

go

na

l

-400

-200

0

200

400

600

/aw/

1900 1920 1940 1960 198029

Regression coefficients for Date of Birth and Gender before and after 1958

Born before 1958 Born after 1958

Variable coeff. prob coeffi. prob

Date of birth 3.34 <= 0.0001 -4.96 0.036 Female gender 193.68 <= 0.0001 109.33 0.023

30

6/1/2012

6

Date of Birth

Dia

go

na

l

-200

0

200

400

600

1900 1920 1940 1960 1980

/aw/

Female

Male

Date of Birth

Dia

go

na

l

-200

0

200

400

600

1900 1920 1940 1960 1980

/aw/

<=12

>12

Increasing height by F1 of /aw/ in south, down, now, etc. by Date of birth by Sex by Education

31 Date of birth

F2

1200

1400

1600

1800

2000

1900 1920 1940 1960 1980

Vowel

aw

owF

owC

Reversal of the fronting of /aw/, /owF/ [free] and /owC/ [checked] in the Philadelphia Neighborhood Corpus [N=264]

span =1.0 span = 0.7

Date of birth

F2

1200

1400

1600

1800

2000

1900 1920 1940 1960 1980

Vowel

aw

owF

owC

32

The relation of Philadelphia sound changes to the dialect geography of North America

33

The height of /ey/ in made, pain, pay, etc. (ANAE Map 10.17)

Philadelphia

34

From ANAE Map 14.10: Canadian raising (green = F1(ayV) – F1(ay0) > 60

Philadelphia

35

The relative fronting of /ay/ and /aw/ in North America

South and Midland: F2(aw) > F2(ay) [rɑɪt næʊ] North: F2(ay) > F2(ayw) [raɪt nɑʊ]

36

6/1/2012

7

Relative fronting of /aw/ and /ay/ in North America ( ANAE Map 14.1)

Philadelphia

37

Fronting of /ow/ in North America (ANAE Map 12.3)

Philadelphia

38

III. Socially salient features of Philadelphia phonology: recent lowering of tense /æh/ and /oh/

39

Parallel tensing of /æh/ and /oh/ by Jean B., 60 [2006]

40

Paralell correction of /’æh/ and /oh/ by MacKenzie S., Masterman High School and University of Pennsylvannia

41

Lowering of /æh/ in mad, bad, man, etc. and /oh/ in off, talk, etc. by Date of Birth and Sex in Philadelphia Neighborhood Corpus [N=264[,

Date of Birth

Dia

go

na

l

200

400

600

800

1000

1200

1900 1920 1940 1960 1980

/æh/

Female

Male

Date of Birth

Dia

go

na

l

850

800

750

700

650

600

550

500

1900 1920 1940 1960 1980

/oh/

Female

Male

42

6/1/2012

8

Lowering of /æh/ in mad, bad, man, etc. and /oh/ in off, talk, etc. by Date of Birth and Higher Ed in Philadelphia Neighborhood Corpus [N=264]

/æh/ /oh/

Date of Birth

Dia

go

na

l

200

400

600

800

1000

1200

1400

1900 1920 1940 1960 1980

/æh/

<=12

>12

Date of Birth

F1

850

800

750

700

650

600

550

500

1900 1920 1940 1960 1980

/oh/

<=12

>12

43

IV. Social correction of Philadelphia phonology: from the split system to the nasal system

44

p t tʃ k

b d dʒ g

m n ŋ

f s θ ʃ

v z ð ʒ

z

mad, bad, glad only

Following segments with tensing of short-a in Philadelphia

p t tʃ k

b d dʒ g

m n ŋ

f s θ ʃ

v z ð ʒ

45

Some further conditions on the tensing of short-a in Philadelphia

a. Function word constraint: Function words with simple codas (an, and, I

can) are lax while corresponding content words are tense (tin can, hand, add),

while can’t with a complex coda, remains tense. This preserves the contrast

of tense can’t vs. lax can in environments where the /t/ is elided or

neutralized.

b. Irregular verbs. The irregular verbs ran, swam, began are lax.

c. Open syllable constraint: Short-a is lax in open syllables, yielding tense

ham, plan, pass, but lax hammer, planet, passive).

d Inflectional boundary closing: Syllables are closed by inflectional

boundaries, so that tense forms include planning as well as plan, staffer as

well as staff,

e. Learned words: Many learned or late-learned words with short-a in tense

environments are lax: alas, carafe.

46

Tense/lax split of short-a in Philadelphia: system of Jean A., 60, Lock St. [2006].

47

Nasal short-a system adopted by Alex B., 19, Masterman High School/ University. of Pennsylvania

48

6/1/2012

9

Short-a systems in the formation of the Inland North

Philadelphia

49

Some general observations on the causes of Philadelphia sound change

• There is no structural imperative behind the raising of /eyC/ and /ay0/ in Philadelphia

• The most consistent feature shared by Philadelphia sound changes is the alignment with the dialect geography of the North and North Midland and a shift away from features characteristic of the South and South Midland.

• Some time in the formative years of those born in the 1940s the dialect of Philadelphia was re-oriented from a Southern to a Northern variety of English.

• The triggering event most likely associated with this change is the population movement brought about by World War II.

50

Some benefits from the large-scale study of change in progress..

Extended data in real time allows us to track change from its inception and thus formulate, if not solve, the actuation problem. As the volume of data increases, significant differences may appear among members of the population, testifying to the differentiating force of the variable or the change in progress. But from the outset, the large-scale study of variation shows us how similar members of the speech community are, that the orderly heterogeneity of the community is a social fact in Durkheim’s sense.

51

www.ling.upenn.edu/~labov

52