Of Moice and Men: The evolution of a male-led sound change

1
Of moice and men The evolution of a male-led
sound change
Photo by John Frank Keith
2
Of moice and men The evolution of a male-led
sound change

• Sociolinguistic studies show language change led
  by
• Women
• The interior social classes

Supported by the data from the study of
Linguistic Change and Variation in Philadelphia
LCV (Labov, 2001)
Conformity Paradox Women deviate less than men
from linguistic norms when the deviations are
overtly proscribed, but more than men when the
deviations are not proscribed (367)
The Curvilinear Principle Linguistic change from
below originates in a central social group,
located in the interior of the socioeconomic
hierarchy (188)
3
Of moice and men The evolution of a male-led
sound change

• Problem variable in the LCV data The raising of
  the nucleus of the diphthong /ay/ before
  voiceless consonants (ay0)
• Led by men
• Shows no social stratification

psych
nice man

• Is (ay0) a counter-example to typical language
  change?
• How does (ay0) progress through the speech
  community over time?
• What about the movement on the front/back
  dimension of (ay0)?
• If (ay0) does not behave like other vocalic
  changes in progress, are there certain
  gender-based evaluations of this variable? That
  is, do certain variants sound more
  masculine/feminine?

Questions from the patterning of (ay0) in the LCV
data
4
Of moice and men The evolution of a male-led
sound change
The current study Of moice and men The
evolution of a male-led sound change OMM

• OMM
• Re-study of Philadelphia 30 years after LCV
• Data collected from (2000-2003)
• Focus on (ay0) and secondary focus on (aw)
• Included self-identified gays and lesbians as
  part of the data set

Striving for high comparability with the original
study, OMM followed the methodology and data
analysis of the LCV as discussed in Labov, 2001
Microphone and recorder differences were not
taken into consideration, but will be looked into
in the future
5
Of moice and men The evolution of a male-led
sound change
Methodology

• Sample 65 native Philadelphians
• The data sociolinguistic interviews (at
  subjects house) including formal tasks of
  semantic differentials, minimal pairs tests,
  reading passage and a word list
• Social Coding Each speaker was coded for various
  social characteristics following the LCV (see
  Labov, 2001 for further details) - education,
  occupation and residence converted into
  socioeconomic class category (SEC)

• age
• sex
• education
• occupation
• residence value
• mobility

• house upkeep
• ethnicity
• foreign language background
• generation
• neighborhood of origin

6
Of moice and men The evolution of a male-led
sound change
Methodology

• Also coded for sexual orientation

• Sexual orientation for both F1 and F2 (ay0) is
  not a significant social factor predicting values
  as either a binary category (gay/lesbianhetero)
  or a combo 4-way split of sex and sexual
  orientation

Binary Category Binary Category Sex/Sexual Orientation Combo Sex/Sexual Orientation Combo
F1 F2 F1 F2
p lt . 0.9478 (F2) p lt 0.5843 p lt 0.6660 p lt 0.3294
7
Of moice and men The evolution of a male-led
sound change
Methodology
Acoustic vowel analysis

• LPC analysis in Praat
• Single-point, synchronous nuclear measurements of
  F1 and F2
• Additional auditory support for single-point
  selection
• Vowels of all Plotnik 25 vowel classes were
  measured - at least 5 tokens per class per
  speaker - complete vowel system for every speaker
  (200-500 tokens)
• Data cleaned for measurement errors
• Using Nearys Log mean normalization in Plotnik,
  each speakers cleaned system was normalized, and
  from these data, a mean F1/F2 for each vowel
  class (and phonetic subclasses) was calculated

8
Of moice and men The evolution of a male-led
sound change
Methodology
Some methodological concerns for investigating a
speech community in real time 3 decades later

• Subject recruitment representative neighborhoods
  have changed
• Updating the socioeconomic class index (SEI)

9
Of moice and men The evolution of a male-led
sound change
Methodology
Subject recruitment

• LCV neighborhoods selected to represent different
  social classes - Kensington (NE), South Philly
  (S), Overbrook (W) King of Prussia (NW)
• OMM neighborhoods sampled similar areas - biggest
  change was substituting Chestnut Hill/Mount Airy
  for King of Prussia
• Self-identified gays and lesbians recruited
  through personal contacts (sometimes
  relatives/friends of neighborhood subjects)

10
Of moice and men The evolution of a male-led
sound change
Methodology
Updating the socioeconomic class index (SEI)

• LCV used scale below to calculate socioeconomic
  score, which was used to calcluate socioeconomic
  class category (SEC)

11
Of moice and men The evolution of a male-led
sound change
Methodology
Updating the socioeconomic class index (SEI)

• The median residence values according to the
  census data have increased from 10,600 (1970) to
  59,700 (2000), so each level of the residence
  scale was multiplied by 5.632 to reflect this
  change
• According to the 1970 and 2000 censuses, the
  median education attainment level changed from
  10.9 years in 1970, to graduating high school or
  equivalent in 2000. This reflects an overall
  increase in the populations education, so 1
  point was added to each SEC to account for this.

12
Of moice and men The evolution of a male-led
sound change
Methodology
Updating the socioeconomic class index (SEI)

• Translation of social class categories (SEC) from
  LCV to OMM

13
Of moice and men The evolution of a male-led
sound change
Statistical Analysis

• In order to examine all the independent variables
  at the same time, a stepwise multiple regression
  analysis was conducted using the following
  independent social variables

• age
• sex
• education
• occupation
• residence value
• mobility

• house upkeep
• ethnicity
• foreign language background
• generation
• neighborhood of origin

14
Of moice and men The evolution of a male-led
sound change
Apparent Time F1 (ay0) Results

• The stepwise regression analysis of (ay0)
  selected the following social variables as
  significant factors in predicting F1 (ay0) values

15
Of moice and men The evolution of a male-led
sound change
Apparent Time F1 (ay0) Results

• This model with age, occupation and generation
  can account for 46 of the variation (r2 0.46)
  of F1 (ay0) in the data, with age as a
  significant predictor at p lt .0001
• Data show change in apparent time

650
700
Predicted F1 (ay0)
750
800
14-29
30-39
40-49
50-59
60
age groups
16
Of moice and men The evolution of a male-led
sound change
Apparent Time F1 (ay0) Results

• Generation score of 3 significantly higher F1
  (ay0) values (non-raised variants) than the other
  scores
• Occupation score of 3 has significantly lower F1
  (ay0) values, while a score of 4 has
  significantly higher F1 (ay0) values (not
  curvilinear principle)

Predicted F1 (ay0)
occupation scores
Occupation scores based on regression estimates
(least squares means)
17
Of moice and men The evolution of a male-led
sound change
Apparent Time F1 (ay0) Results

• Apparent time shows no sex differentiation or
  social stratification

• Sex not significant social factor predicting F1
  (ay0) - distribution shown below (linear fit
  lines and p .90 bivariate normal elipses)

• SEC not significant social factor predicting F1
  (ay0) - distribution shown below

600
ay01
650
700
750
800
10
20
30
40
50
60
70
80
90
100
Regression lines for each social class of F1
(ay0) with age as a continuous variable
Age
18
Of moice and men The evolution of a male-led
sound change
Apparent Time F2 (ay0) Results

• F2 (ay0) does not show age as significant social
  factor predicting F2 values (no change in
  apparent time)
• SEC does show significant effects (plt.001), but
  when sorted by sex, only men show significant
  social stratification (plt.001) while women do not
  (pgt.10)

Predicted F1/F2 (ay0) values plotted by sex and
social class
19
Of moice and men The evolution of a male-led
sound change
OMM Real Time F1 (ay0) Results
Transformed LCV data into comparable age groups
with OMM

• F1 (ay0) in apparent time for both data sets

650
700
OMM
Predicted F1 (ay0)
750
LCV
800
850
under 30
30-39
40-49
50-59
60
age group
20
Of moice and men The evolution of a male-led
sound change
Real Time F1 (ay0) Results
Added 30 years to LCV ages and combined data sets

• Stepwise process selected age and sex as
  significant social factors (at the p lt 0.1 level)
  with 33.7 of variation explained by model (r2
  .337)
• Real time change shows larger decreases in F1
  (ay0) followed by plateaus of little change

650
700
Predicted F1 (ay0)
750
800
850
14-29
30-39
40-49
50-59
60-69
70-79
80-89
90
age group
Predicted F1 values of (ay0) for both LCV and OMM
data sets
21
Of moice and men The evolution of a male-led
sound change
Real Time F1 (ay0) Results

• Sorting the data by sex, varying moments of sex
  differentiation
• This picture is different from apparent time
  analysis in Labov, 2001 in that unified speech
  community in 80-89 age group

Predicted F1 (ay0)
age group
Predicted F1 (ay0) values for combined data sets
sorted by sex
22
Of moice and men The evolution of a male-led
sound change
Real Time F1 (ay0) Results

• SEC not selected as significant factor in the
  regression model, but sorting the data by SEC,
  age still a significant factor (p lt .10 level)
  for each class (change occurring in all classes)

550
600
UWC
650
LMC
700
F1 ay0
UMC
750
800
850
900
10
20
30
40
50
60
70
80
90
100
110
120
Age
Regression lines for each social class of F1
(ay0) for both studies
23
Of moice and men The evolution of a male-led
sound change
Real Time F2 (ay0) Results

• F2 (ay0) in combined data set - stepwise
  regression model selected age, occupation,
  residence and education as significant social
  factors (p lt.10)

Predicted F2 (ay0)
age group
Predicted F2 (ay0) values by age groups for
combined data
24
Of moice and men The evolution of a male-led
sound change
Real Time F2 (ay0) Results

• Real time analysis does not show clear social
  stratification of this change
• Sorting the data by SEC, age only significant
  factor in LWC (p lt 0 .0354) and UWC (p lt 0.0205)

Predicted F2 (ay0)
age group
Predicted F2 (ay0) values for both data sets by
age group and SEC
25
Of moice and men The evolution of a male-led
sound change
Real Time Summary

• The mechanism of (ay0) raising sound change
  change began by whole community, and then sex
  differentiation
• No clear social stratification of this variable
• Only real time analysis shows F2 backing over
  time
• F2 no sex differentiation, but social class
  stratification

26
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• 6 Speakers
• Jill - 24 year old woman Ben - 43 year old man
  (2 guises each)
• 1 other man and 1 other woman used as fillers
• Sentences
• 3 variables investigated (aw, ay0, and neutral) X
  2 sentences each

27
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• Creating the test
• Both Jill Ben practiced so they produce
  moderate Philadelphia tokens and more extreme
  Philadelphia tokens (extra Philly)
• Jill/Bens vowel system calculated through
  reading passage and word list
• Tokens for each guise selected from the many
  possibilities comparing the extra Philly tokens
  within each speakers regular vowel system
• Sentences spliced together from the selected
  tokens
• Sentences were duplicated (so each sentence
  played two times consecutively) and randomized
• Male then female speaker alternating
• Used filler speakers to make sure that no two
  identical sentence of the corresponding Jill/Ben
  guise occurred close together

28
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• Administering the SRT
• 36 sentence SRT administered as part of socioling
  interview (sometimes after, sometimes before)
• Evaluators were told to rate each speaker (3 men,
  3 women) on the following scale for each sentence.

29
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• 3 Analyses of SRT evaluations
• 1) looks at the data from all of the evaluators
  to see if patterns from the matched guise aspect
  are revealed from the entire speech community
• 2), following the analysis of the LCV SRT in
  Labov, 2001, examines the difference for each
  speaker/guise from the neutral sentence ratings
  to the ratings of each variable
• 3) uses a series of differences in each
  evaluators ratings to uncover any social
  variables which may affect the ratings

30
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• All evaluators matched guise evaluations
• 2-tailed, unequal variance t-tests conducted on
  mean evaluations for matched guises (boxed diffs
  are significant at p lt 0.01)

31
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• All evaluators intraspeaker evaluations

32
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• Social factors of evaluators
• Some significant factors, but not consistent
• Age or sex never significant
• Uniform speech community as evaluators

33
Of moice and men The evolution of a male-led
sound change
Subjective Reaction Test

• SRT Summary
• Uniform speech community
• Male and female speaker for (ay0) evaluated on
  different scales but male and female evaluators
  agree on this distinction and difference in
  sociolinguistic expectations of men and women

34
Of moice and men The evolution of a male-led
sound change
SUMMARY

• Real time support for apparent time analysis of
  LCV (ay0)
• (ay0) backing only shown in real time
• This variable shows language change progresses
  not linearly, but taking large steps forward, and
  then relative stability
• Sex differentiation not a given, but needs to be
  maintained at each step in the change

35
Of moice and men The evolution of a male-led
sound change
OMM To be continued
Whats next? (To be continued at NWAV35...)

• While (ay0) does not show sex differentiation or
  social stratification, the other new and vigorous
  changes do (eyC) and (aw)
• They also show a significant effect of sexual
  orientation
• What about other changes - incipient, completed?
  - in Philadelphian English
• Is Philadelphia becoming a Northern city and
  losing its Philly-ness? What does this all mean?
• Check out my website to download this
  presentation and find out more details about
  methodology www. jeffconn.net