While a substantial body of work has been done on identifying
linguistic factors in sound change—phonological structural
considerations (Martinet, 1955), the
speaker's articulatory effort (Lindblom et al.,
1995), or the listener's auditory misperceptions (Ohala, 1981)—the role of social factors in
initiation and actuation of sound change has been largely beyond the scope
of formal phonology or historical linguists until relatively recently. The
sociolinguistic work produced in the last few decades (starting with Labov, 1963; see Labov, 1994, 2001; Chambers, 2003) has convincingly demonstrated that
almost any phonological or phonetic sound change (or any linguistic
change) can be initiated and continuously influenced by a range of social
factors, such age, sex, and social class of speakers.
This article is a contribution to the continual investigation of the
role of social factors in sound change. It presents results of a
sociolinguistic study of a Russian dialect spoken in a small town
community of Pokcha in the Western Urals. As a result of the increasing
social pressure of urbanization, the dialect has been undergoing a rapid
shift towards Standard Russian. The focus of the study is on two
standardizing sound changes: (a) the unrounding of the unstressed vowel
/o/ ([o] → [a]) and (b) the split of a
retroflex post-alveolar fricative
.
As I will show, social factors of age, home background (raised in town or
out of town), and, partially, education play a crucial role in both
changes. Reference to these social factors, combined with an examination
of the phonetic characteristics of the sounds involved, also explains the
strikingly different dynamics of the two changes.
BACKGROUND
Pokcha: The social background
Pokcha is a small town in the Western Urals, Russia. It is a
predominantly rural community about 480 km north of the province's
capital city Perm' (formerly Molotov). Administratively, it belongs
to the Cherdyn' region (raion) of the Perm' province
(oblast'). The region, like other parts of the Western
Urals, used to be sparsely populated by the indigenous Uralic-speaking
Komi (Permyaks). Russians began to settle in the area along the Upper Kama
and Kolva rivers starting in the 15th and 16th centuries as a part of the
general eastward expansion (Chagin, 1988:9–16; Poliakova,
1998). Most new settlers were peasants from the Russian Northwest,
driven by over-population, poverty, and religious tensions. The region,
with Fort Pokcha and then Cherdyn' as administrative centers, became
a Russian strategic outpost on the way to Siberia and the Southern Urals
(Chagin, 1988:131–132; cf. Pallot, 2002:1060). With further waves of migration to
the west and south, the region, with few natural resources and diminished
military and commercial importance, fell into a relative economic
decline.
The socioeconomic stagnation of the region was broken in the 20th
century. The Bolshevik revolution in 1917 was followed by a land reform
that took property from the big local land owners and turned it over to
peasants. In the early 1930s, the “collectivization” campaign
was initiated on a countrywide scale, establishing collective farms and
cracking down on “excessive” private land-owning. The campaign
led to the elimination of the middle class and leveling of social
distinctions in the region (Ivanova, 1998).
Thus, the family of one of my speakers, Marija P., lost all property and
was evicted from Pokcha in 1932, not to return until eight years
later.
At the same time, the “cultural revolution” campaign aimed
to eliminate illiteracy, as well as “religious superstitions.”
New schools and libraries were set up in towns and in rural areas of the
region; churches were closed and turned into clubs or warehouses, or
simply destroyed. This happened to the
Blagoveščensky cathedral in Pokcha (built in 1785;
Chagin, 1988:137), which, according to my
informants, served as a collective farm warehouse in the 1930s and was
later partly demolished.
Because of its remoteness, the region became a convenient location for
a number of labor camps, Gulags, where political and civil
prisoners from all over the country were used as a free workforce at
logging sites (Pallot, 2002; Suslov & Gasheva, 1998; Varese,
2001). In the late 1930s to the early 1950s, the timber-rich
Cherdyn' region had the highest concentration of labor camps in the
Perm' province, with the latter having the highest number of such
facilities in the USSR (Varese, 2001:124–125).1
The
Cherdyn' district still hosts a range of correctional facilities,
including “special regime” and “general regime”
prison camps, and “penal villages” (Pallot, 2002:1078; cf. Varese, 2001:124–125).
The camps were directly
operated by
NKVD (the former
KGB), but local people were
often recruited to provide necessary services (Pallot,
2002:1066–1067). Thus, some residents of Pokcha
used to work as truck drivers or guards for the local
Usollag and
Nyroblag networks of camps (Vladimir V. and Ivan š. in my
survey sample). Former prisoners released from camps were often encouraged
or forced to settle in the region. The intermarriage of former prisoners
and locals has resulted in “the unusual population mix that still
characterizes settlements in the Upper Kama region today” (
Pallot, 2002:1068).
In 1941, almost all able-bodied men of the region were drafted and
sent to the World War II eastern front. Out of necessity, women became
involved in the economic life of the region at all levels. Prior to that,
it was hardly possible for a woman to occupy the position of collective
farm manager or director of a trade cooperative, as did for example, Vera
D. in my survey sample. The overall inefficiency of collective farming
prompted the Communist government to declare the countrywide amalgamation
of these farms and elimination of small villages in the 1950–1960s
(Poliakova, 1998:28). According to my
informants, almost all of over 20 villages of the Pokcha county turned
into “ghost towns,” with their population flooding into the
administrative center. Many old and middle-aged speakers in my sample were
raised in those smaller villages and subsequently moved to Pokcha. Whether
a person was raised in Pokcha or outside it became a social factor
stratifying the otherwise relatively socially homogeneous population of
the town.
The collapse of the Soviet Union and the fall of the Communist rule in
1992 brought political and economic changes. These changes, however, were
not enough to stabilize the crippled economy of the region. The abrupt
introduction of the “free market” pushed collective farming,
forestry, and many other state-owned businesses to the brink of collapse,
while the conditions for the new type of market were not yet established
(Pallot, 2002:1077–1079). This resulted in
a further decline in the standard of living, with mass unemployment and
chronic wage debts in the region, giving an impetus to migration. Now,
most younger mobile people from Pokcha are moving to larger urban centres
that have relatively more opportunities for employment and private
enterprise. For young residents of Pokcha, life in a city has become
associated with success and prosperity, contributing to the already strong
social pressure towards urbanization.
Pokcha: The linguistic background
The local dialect, a variety of the Northern Russian dialect, formed
in the 16th and 17th centuries, existed in a relatively stable state for
several centuries (Poliakova, 1998). The major
changes of the 20th century, particularly starting in the 1930s, could not
leave the Pokcha dialect intact. The speech of the region started to
undergo a gradual shift towards the standard usage. This shift was
accompanied by a significant degree of variation, which is the focus of
this study.
The following discussion of the variables is based on the
author's fieldwork. Although the rural dialects of Perm'
Province has been a focus of a number of linguistic studies (Shtern & Erofeeva, 1998; Skitova, 1961; Smoliakova,
1961), none of these works specifically investigated the dialect of
Pokcha, or nearby areas. In addition, most studies were primarily
concerned with lexicography rather than the phonology or phonetics of
these dialects (cf. Ignatkina, 1987:132). The
authors, however, mention the significant interspeaker variation and
differences between generations in speech patterns, leading to a
“collapse of dialect phonetic systems … under the destructive
impact of the literary language” (Shtern &
Erofeeva, 1998:70).
The vowel variable
Standard Russian and the Pokcha dialect have the same stressed vowel
inventory, consisting of five vowels /i e a o u/ (Bolla, 1981; Bondarko, 1981;
Timberlake, 2004; for Standard Russian).
(Phonetically, Standard Russian vowels are
.
In the Pokcha dialect, as in other rural dialects of Perm' Province
(Ignatkina, 1987:132; Shtern
& Erofeeva, 1998), the back mid vowel /o/ can be
realized as mid-high [o] rather than mid-low
,
as mentioned later.) One of the main differences between the two
varieties is found in the quality of some unstressed vowels. In Standard
Russian both /a/ and /o/ are neutralized in unstressed
environments to a reduced [a], which is phonetically unrounded
mid-low back
.
In the Pokcha dialect, the contrast between the unstressed allophones of
/a/ and /o/ is maintained: They are realized as slightly
reduced [a] (
) and [o], respectively. Examples in Table 1 illustrate the differences between the
standard and the dialect in the choice of variants for the
stressed/unstressed vowels /o/ and /a/.
Similarities and differences between Standard Russian and the Pokcha
dialect in the realization of stressed and unstressed vowels /o/
and /a/
Based on my acoustic analysis of the unstressed vowels in the Pokcha
dialect, the two variants of the unstressed /o/ (phonetically
) differ mainly in the values of the first and second formants
(F1 and F2; roughly corresponding to tongue height and tongue backness:
Ladefoged & Maddieson, 1996). The difference
in F1 is about 250 Hz and the difference in F2 is about 350 Hz for older
speakers (cf. Erofeeva, 1997 on the Perm'
variety of Standard Russian).
The pattern of the Pokcha dialect is typical of Northern Russian
dialects and is referred to in the literature as okan'e, or
‘using [o]’, as opposed to the standard
akan'e, ‘using [a]’ as the allophone
of the unstressed /o/ (Avanesov & Orlova,
1965). The pattern of okan'e is more conservative,
reflecting an earlier historical stage, while akan'e
represents an innovation.
The inventory and realizations of vowels in the Pokcha dialect (as
shown in Table 1) represent the older variety of
the dialect, or the old regional standard. It is still characteristic of
the speech of the oldest and predominantly rural speakers of the region.
As a result of increased social pressure, the phonological patterns of the
dialect have been changing to conform to those of the standard. The focus
of this study is not on the differences between these two varieties, but
rather on the sound change in the Pokcha dialect from the nonstandard
[o] to the standard [a] in all positions where the
unstressed /o/ is found. Overall, the sound change leads to a
phonological simplification that results in a positional neutralization of
a contrast. The variable is quite frequent in speech, given the high
occurrence of the unstressed /o/ in Russian (Bondarko, 1981:144–149).
Note that the vowel change appears to be phonetically gradual,
involving intermediate stages (more specifically,
;
cf. Shtern & Erofeeva, 1998). In the
current study I will not distinguish between the rounded variants
,
focusing on the change as “unrounding.” The status of the
mid-low rounded variant
,
however, is sociolinguistically important (as discussed later). It can be
regarded as a “fudge” (Chambers &
Trudgill, 1998:110), an intermediate variant that shares rounding
with [o] and height with
.
The consonant variable
Another difference between the dialects is in the presence or absence
of a phonemic contrast in voiceless post-alveolar fricatives. Standard
Russian has two such fricatives: a retroflex (apical) post-alveolar
and a laminal post-alveolar
,
with the latter being phonetically long
(Timberlake, 2004:54–55, 65–67).
The
of Standard Russian correspond to the short and long variants,
,
of the phoneme
in the Pokcha dialect, the correspondence typical of many Northern
Russian dialects (cf. Avanesov & Orlova, 1965:88–90). In these dialects, the original
(developed from the cluster
) has lost its laminal articulation and merged with the apical
.
The merger, however, can be considered incomplete, since the phonetic
length of the former
is still preserved (as
) in stressed prevocalic contexts.
2 This leaves a possibility of considering
in stressed prevocalic contexts as a separate phoneme
.
Such analysis is complicated by the fact that the dialect also has
as a realization of the heteromorphemic sequences
,
for example, s-šit'
‘to sew’, perfective (cf. Avanesov &
Orlova, 1965:89).
shows
the phonemic similarities and differences between the two dialects. As we
can see, in some lexical items, both Standard Russian and the Pokcha
dialect have
,
whereas in other lexical items, Standard Russian has
,
while the Pokcha dialect has
(or
).
Similarities and differences between Standard Russian and the Pokcha
dialect in the use of post-alveolar fricatives
The production of the two consonants involves conflicting articulatory
gestures: tongue tip raising for
and tongue tip lowering (and tongue front raising) for
,
acoustically manifested in lower spectral noise and lower second formant
(F2) vowel transitions for
compared to
(Bolla, 1981; confirmed by the author's
acoustic analysis).
The change examined in the current study involves a split of the
into
(which is in fact a reversal of the original, albeit incomplete, merger
of
into
). The focus here will be on the change in the quality of the
fricative, rather than on its duration. This sound change adds to the
complexity of the grammar and lexicon, because it introduces a new phoneme
to the inventory and restructures corresponding lexical representations.
It is relatively infrequent in speech, given the relative paucity of
in Russian, at least compared to the unstressed /o/ (Bondarko, 1981:144–149).
THE STUDY SET-UP
Sample and method
The data for the study was collected during my fieldtrip to the
Cherdyn' region. The speech sample for this study includes recorded
spontaneous speech from 32 informants, all residents of Pokcha, with a
total duration of about eleven hours. The subjects were asked to talk
about themselves—their youth, school, work, and life in town.
Factors
The factors examined in the study include speakers' age, sex,
education, and the place where he/she was raised. The sample comprises
speakers born in the period from 1910 to 1990 (see Appendix). Speakers
were divided with respect to their age into three age groups, roughly
corresponding to three generations (Table 3).
There are substantial time gaps between these groups: six years between
the oldest speaker in the first group and the youngest speaker in the
second group; and thirteen years between the oldest speaker of the second
group and the youngest speaker of the third group. All the groups include
from ten to twelve speakers. There are seventeen men and fifteen women in
the sample, fairly evenly distributed across the age groups.
The division of speakers into three age groups employed in the
study
In our examination of the two sound changes and estimation of their
relative timing, the age group variable will be very important. It is
assumed here that the speakers retain their speech patterns acquired in
their teens (the “third formative period”: Chambers, 2003:170–171); thus, these patterns give us a
rough impression of the degree of variation in the corresponding
decades.
All the speakers in the study were born and raised in the north of
Perm' Province, mostly in Pokcha, or in smaller villages within 50
km. Speakers raised in Pokcha will be contrasted with those from the
outside; this variable will be referred to as the “home
background.” In terms of education, the speakers are divided into
two groups based on the number of years of schooling: education 1 (Grade
1–7) and education 2 (Grade 8–10 or more). The factors of
occupation and class were not treated as independent variables and thus
were not coded for. The sample was to a large degree homogeneous with
respect to these factors. Many speakers in the sample had (or used to
have) jobs as farmers, forestry workers, drivers, cleaning persons, and so
forth. A few subjects that at some stage in their life occupied
white-collar positions (e.g., a librarian or a director of a trade co-op)
are sometimes singled out on an individual basis as the results warrant.
The population in the region is, for the most part, rooted in the region.
Most of my subjects have never lived outside the region (or in some cases,
even never traveled outside it). Men in general, however, are more mobile
than women. Many men have spent a few years in the military outside the
region. The demographics of the speakers, indicating their year of birth,
sex, educational level, and home background, are given in the
Appendix.
Analysis
The vowel variable, the unstressed /o/ phoneme, occurred very
frequently in the speech of all 32 informants (on average, about 15 tokens
per minute of speech). Each instance of the unstressed /o/,
corresponding to the orthographic “o”, was classified as
belonging to one of the two variants, [o] or [a]. This
was done based on auditory judgments; selected tokens were verified by an
acoustic analysis of vowel formants (using Praat). The study involved a
total of 3,592 vowel tokens (i.e., about 110 tokens per speaker). The
consonant variable occurred much less frequently in the recorded sample
(on average, less than 1 token per minute). Three speakers did not have a
single item with it, leaving us with 29 other speakers in the sample.
Altogether the sample included 585 tokens (about 20 tokens per speaker)
for this variable. Each instance of the consonant variable, corresponding
to the orthographic “šč”
,
was classified as belonging to one of the two variants,
,
based on auditory judgments.
Based on the collected tokens, mean percentages of nonstandard vowel
and consonant variants were calculated for each speaker. These were
analyzed statistically by running a multivariate ANOVA with the following
between-speaker factors: age group (older, middle-aged, and younger), sex
(men and women), home (raised in Pokcha and raised outside), and education
(education 1 and 2).3
In the initial analysis
of the data, each token was coded for linguistic variables (e.g.,
preceding or following segments, stress placement, and frequency), in
addition to the social variables. The data were analyzed using the
sociolinguistic statistical program goldvarb, the Macintosh
version of varbrul (Rand & Sankoff,
1991). Because the results did not show significant effects of
linguistic variables (see Kochetov, 2000, for
the vowel variable), the current article focuses on the social variables,
which are reanalyzed using spss 13.0.
Bonferroni
post-hoc tests were used to determine differences between three levels of
the age group factor.
SUMMARY: TWO CHANGES
The analysis of the data for both variables,
, shows a gradual decrease in the use of the nonstandard
variants
at the expense of the standard variants
.
Both processes are, therefore, sound changes towards Standard Russian:
.
The processes correlate strongly with speakers' age group and, to a
lesser degree, with their home background (whether they were raised in
town or in a neighboring village). The level of education is also
important for the vowel variable, at least among middle-aged speakers.
Neither of the variables appears to correlate with sex. Despite these
similarities, the vowel and consonant changes are quite different in their
overall dynamics and degree of variation. These differences are further
discussed later.
As mentioned earlier, the use of speech variants by speakers of a
particular age group is likely to reflect the use of these variants during
the period of their teens (Chambers, 2003).
Thus, the speech of speakers in our sample who were born in the 1910s may
represent, or be close to, the speech in Pokcha of the 1920s. This would
allow us to estimate the dynamics of the vowel and consonant changes.
Graphs (a) and (b) in Figure 3 present mean
values of the nonstandard variants for each speaker, plotted against the
years when these speakers reached their teens. (Note that not all decades
are equally represented by the speakers in our sample.) In each graph, the
trajectory of the change is plotted as a cubic fit line with 95%
individual confidence intervals (using spss 13.0).
The trajectories of the vowel change (o → a) (a) and consonant
change
(b) as reflected in the percentage of nonstandard variants
in the speech of 32 speakers. The years indicate the approximate time
when the corresponding speakers developed their speech
patterns.
We can see from the graphs that standardization of the Pokcha accent,
as reflected in the two changes, started perhaps prior to the 1920s. In
the early 1920s, the use of the nonstandard variables is still close to
100% (with more variation for the consonant variable). The vowel change
(Figure 3a) approximates an S-curve. Almost
stagnant in the 1920s, the change progresses steadily from the 1930s to
the 1950s, approaching the 70% mark for the nonstandard variant. Recall
that this is the time of the major social upheavals in the region:
“collectivization” and “cultural revolution,” the
advent of the Gulag, World War II, and the exodus of village
populations into town. By the beginning of the 1970s, the use of the
nonstandard variant drops to below 50%, and then seems to stabilize at
around 35–40% in the 1980s and 1990s. The process is thus still a
change in progress, possibly turning into a stable variability pattern
(Chambers, 2003:121; Labov, 2001:74–75).
In contrast, the consonant change (Figure
3b) exhibits interspeaker variability already in the 1920s,
increasing in the 1930s and 1940s. The use of the nonstandard variant
drops to 0% in the 1950s, thus de facto completing the change, at least
for the speakers who grew up in Pokcha. In sum, the consonant change is
more rapid and variable than the vowel change.
Undoubtedly, some of the interspeaker variability can be attributed to
the relatively small number of tokens per person for the consonant
variable (on average, 20 tokens per person), compared to the vowel
variable (on average, 110 tokens per person). Yet, recall that some of the
variability can be well accounted for by speakers' age, home
background, and education. More importantly, the relatively small sample
size does not explain the dramatic “all or nothing”
differences between older and younger speakers in the use of the
nonstandard variant.
The striking differences between the vowel and consonant changes in
the pace can be possibly explained by different phonetic salience of the
two nonstandard variants and degrees of social acceptability associated
with them. In the speech of some-middle-aged and most younger speakers in
our sample, the nonstandard variant is often realized as the less rounded
“fudge” variant
,
intermediate between the [o] of the older speakers and the
standard [a] (not quantitatively examined in the current study).
The use of the “fudge” variant was therefore likely to be more
socially acceptable than the use of the traditional nonstandard
[o]. Interestingly, the “fudge” variant is still not
uncommon in the speech of the provincial capital, Perm' (Erofeeva, 1997; Ignatkina, 1987:135–136).
Unlike the vowel variable, the consonant variable did not develop a
fudged variant, because the two consonants involve conflicting
articulatory gestures (tongue tip raising for
and tongue tip lowering for
,
acoustically manifested in vowel transitions and frication noise). Given
this, the perceptually distinct
was increasingly associated with the nonstandard variety, possibly more
than the unstressed vowel variant [o] (or
); at the same time,
came to be identified with the prestigious town speech, more so than the
unstressed vowel variant [a] (more precisely
). While the nonstandard
still occurs in the speech of the provincial capital, Perm', it is
relatively uncommon and widely regarded as a marker of
“uneducated,” “village” register (Shtern & Erofeeva, 1998). Other factors, such as
distinct representation of the phonemes
in the orthography (
) have likely contributed to the rapid conclusion of the
consonant change. Thus, the brochure “Methodological recommendations
for future primary school teachers” (Gruzberg,
1984:5) explicitly instructs teachers to “pay special
attention to words with the letter
… to ensure that students pronounce the sound
correctly … and [to provide] students with as many
examples as possible” (my translation).
One interesting property of the sound change
is that it presents an apparent reversal of the earlier merger of
into
in the Pokcha dialect.
4 I am thankful to the
anonymous reviewer for raising this question.
It is agreed in the
historical and sociolinguistic literature that mergers are difficult, if
not impossible to reverse (see Labov,
1994:311–331 for a detailed discussion of merger
irreversibility). The difficulty of reversing mergers is related to the
general linguistic and cognitive difficulties of relearning a full range
of lexical items with the new contrast. If feasible, unmerging is expected
to be accompanied by frequent hypercorrection (
Labov,
1994:312). The apparent lack of such hypercorrection in the current
study (e.g., the use of
instead of
in words like šutka ‘joke’ and
smešnoj ‘funny’; see Table
2) suggests that a full reversal of the merger has taken place in
Pokcha, at least in the speech of most younger speakers.
5 The success of the merger reversal can be attributed to
the fact that the merger in the dialect has never been completed in the
first place, with traces of the contrast left in consonant duration
differences in certain contexts. Relatively lower frequency of
compared to
in Standard Russian,
6 A count of the
occurrences of the letters
in a sample of over 93,100 words (an electronic version of Suslov & Gasheva, 1998) revealed that
occurred 2.2 times less frequently than
.
as well as the transparent orthographical representations of the two
sounds, could have facilitated the merger reversal.
