social correlates of linguistic variation in sheshatshiu
TRANSCRIPT
Social Correlates of Linguistic Variation
In Sheshatshiu Montagnais
SANDRA CLARKE
Memorial University of Newfoundland
For the past several years, the Montagnais community of Sheshat
shiu, located near Goose Bay in Labrador, has been the subject of
a fairly intensive sociolinguistic and ethnographic investigation. The
purpose of the sociolinguistic component of this study, the ethno
graphic aspects of which have been reported on elsewhere (Mailhot
1985), is to describe the extensive language variability which exists
within the settlement, and to attempt to establish its social corre
lates. Since the community of Sheshatshiu was permanently estab
lished only in the late 1950s, having served prior to that as a sum
mer gathering ground for the several nomadic Montagnais territorial
hunting groups who were to settle the community, an important fac
tor in clarifying intracommunity linguistic differences would appear
to be the various territorial group or band affiliations of its residents.
As Figure 1 demonstrates, these group affiliations are, from a fairly
broad geographical perspective, three in number:
1) a northern or Mushuau affiliation, with links to the Naskapi speakers who
were to settle in Davis Inlet and Fort Chimo
2) a more westerly or Uashau affiliation, with links to the Montagnais of Sept-
Isles and Schefferville1
1Thta grouping includes the classification termed "Sheshatshiu" in earlier studies, which was used to designate those subjects having a longer affiliation
with the community of Sheshatshiu.
65
66 SANDRA CLARKE
3) a southerly or Maskuanu affiliation, with links to the Montagnais of such Quebec Lower North Shore communities as St. Augustin, Mingan, Natash-
quan and La Romaine.
An important concern of the Sheshatshiu sociolinguistic variabil
ity project is to determine the extent to which there has emerged
linguistic homogeneity among younger speakers in a community char
acterized by the diverse regional backgrounds of its oldest generation.
In other words, to what degree has there developed, over the past 25
or so years, a distinct Sheshatshiu dialect of Montagnais, a dialect
which has resolved the problem of competing variants raised by the
marked divergences among its input or source dialects?
The answers to these questions, of course, should provide some
indication as to the extent to which territorial group affiliations have
continued to exert their force on the social structure of the commu
nity; these links appear to be particularly strong among the older
generations, whose social networks seem to be based almost exclu
sively on kinship ties. Some insight should also be obtained as to any
potential social hierarchy which, within the Sheshatshiu community,
characterizes speakers with different territorial group links.
It is of course to be expected that other social variables may play
a major role in clarifying linguistic differences that are presently ob
servable within the community. Age of subjects is an important con
sideration, since age correlates highly with such factors as amount of
education, exposure to English, and general acculturation. Thus the
older present-day community residents would have little if any formal
education, little knowledge of English, and a largely bush orientation.
This would contrast markedly with the community-centered lifestyle
of younger Sheshatshiu residents, all of w h o m have undergone some
amount of schooling in English.
Various sociolinguistic studies, which of course have dealt almost
exclusively with urban settings, have underlined the importance of
yet another social variable: sex of subjects. In such investigations,
the linguistic usage of males and females has proven to follow certain
fairly well-defined patterns, most notably that of a greater sensitivity
to prestige speech features on the part of females. It would be of
considerable interest to discover whether, in a social setting which is
very different from that which forms the focus of most sociolinguistic
investigation, the linguistic behaviour of men and women reflects this dominant pattern.
LINGUISTIC VARIATION 67
Certain preliminary results from the Sheshatshiu sociolinguis-
tics project have already been reported (e.g., Clarke and MacKenzie
1984). While these have exclusively involved formal speech style,
as elicited by means of a word list, they nevertheless suggest the
following hypotheses:
1) there is evidence in Sheshatshiu Montagnais to indicate development towards a more homogeneous dialect among younger speakers; nevertheless, this dialect does not clearly approximate any one of the competing territorial group dialects spoken by the oldest segments of the community.
2) subject sex is not as important a social variable as are age and territorial group affiliation in clarifying present-day intracommunity linguistic differences in Sheshatshiu Montagnais.
In investigating these hypotheses, this paper will focus on the
ways in which Sheshatshiu Montagnais appears to be dealing with
the problems raised by the not inconsiderable variation resulting from
the regional or territorial separation of its input dialects. Such varia
tion has been investigated, though minimally, in some of the sociolin
guistic literature. Trudgill (1984) reports on several investigations of
speech communities characterized by extensive dialect mixing as a
result of settlement by people of divergent regional backgrounds. He
notes that by the third generation, much of the linguistic variation
that results may be reduced. Yet the ensuing simplification is not
accomplished merely by means of dialect levelling; as an alternative
solution, competing variants, i.e., those that have different regional
origins, may be allocated new functional specifications. Given the
considerable territorial group differences displayed by the oldest in
habitants of the community, Sheshatshiu Montagnais offers an ideal
location to test the generalizability of Trudgill's observations.
Data for the present investigation are derived from the free speech
portion of the Sheshatshiu study. That is, the paper will present
a first analysis of the main body of linguistic data collected by
the project, namely relatively casual speech style as elicited in an
interviewer-interviewee type of situation. It must be stressed that,
owing to the immense amount of data collected, the results to be
presented are to be regarded as preliminary. The complete study
involves a sample of 87 subjects, representing both male and female
speakers of various age levels, from the three territorial groups pre
viously mentioned. This paper, however, will examine only a 19-
68 SANDRA CLARKE
subject sample; as represented in Table 1, this mini-sample consists
of three family groups each of which displays a different territorial
group affiliation, as well as a three-generational structure. The pre
liminary nature of the presentation will also be seen in the reliance
on frequency of usage ratios, rather than on full-fledged statistical
analysis, since such analysis was felt to be inappropriate for the small
sample size to be dealt with here.
Although the Sheshatshiu study has examined some 20 phonolog
ical variables, the present paper can deal only with a subset of these,
and so will concentrate primarily on consonant features; specifically,
on the variant pronunciations associated in present-day Sheshatshiu
speech with the Montagnais clusters /§c/, /s/ or /§/ plus stop con
sonant, and with the pre-aspirated sequences /ht/ and / h£/.
The Variable /§£/
The phonological variable represented as /§£/ derives from any
one of the Proto-Algonquian sequences */xk/, */tk/ and */§k/, which
in Cree/Montagnais fell together as /sk/ or /§k/ (for details, see
Cowan 1977); ensuingly, there occurred in Montagnais dialects a
palatalization to /§£/ before front vowels. [§c] is in fact the normal
reflex or variant to be found in present-day Mushuau or northern
areas of the Quebec-Labrador peninsula, while the [s] or [ss] vari
ant is associated principally with Uashau speakers now settled in the
communities of Sept-Isles and Schefferville.
Table 2 represents the ratio of usage of the [s(s)] or Uashau-like
variant for each of the subjects in the mini-sample.2 As can be seen
from the Table, the Uashau-like [s(s)] variant is clearly on the in
crease, the youngest age group or third generation displaying, no
matter what their territorial group background, almost universal us
age of this variant, to the exclusion of other realizations. Territorial
group dialect differences are obvious among the grandparents' gener
ation, however, since, as expected, Uashau subjects of this age group
display a higher degree of [s(s)] usage than do subjects representing
the other two groups. What is interesting is that the obvious trend
in favour of a uniform Uashau-like realization, as revealed by the
In each case, the decimal is to be interpreted as the percentage ratio of the actual usage of a given realization or variant to the total usage of the variable in all its possible realizations.
LINGUISTIC VARIATION 69
mini-sample, seems to be led by female subjects. As many urban
sociolinguistic studies have shown females to be in the vanguard of
change, particularly change in the direction of a prestige feature, this
variable already provides the suggestion that a similar tendency may
also be found in this tiny, rural Algonquian community.
Pre-aspirated Stop Clusters
The /§£/ variable has revealed a trend towards adoption by young
er speakers of a single variant from among the competing variant pos
sibilities offered to them by the characteristic dialects of the oldest
residents of the community. The second of the consonantal variables
to be examined offers a slightly different picture. This is the She
shatshiu treatment of Cree/Montagnais pre-aspirated clusters. In
the Quebec-Labrador peninsula, the Mushuau or northern dialects
typically show the greatest retention rate of the aspirated first ele
ment of such clusters; in Lower North Shore or Maskuanu dialects,
pre-aspiration is usually maintained only in word-final position; and
in Sept-Isles/Schefferville Uashau dialects, the aspirate element of
the cluster is typically deleted, the result being a simple stop conso
nant (see, for example, MacKenzie 1980:67-69). All three of these
patterns are found to varying degrees in Sheshatshiu Montagnais.
In this paper, discussion will be limited to the pre-aspirated se
quence /ht/; this is because /he/ presents a fairly complicated soci
olinguistic profile in which linguistic conditioning plays a consider
able role, whereas the remaining sequences — /hp/, /hk/, and /hkw/
— display few pre-aspirated tokens. Table 3 presents frequencies of
occurrence of the pre-aspirated pronunciation of the /ht/ sequence
among the 19-subject mini-sample. As in the case of /§£/, the con
siderable territorial group differences which exist among the grand
parents', and to a lesser extent the parents', generation are largely
resolved among the youngest group of speakers in the mini-sample.
The obvious development is toward loss of aspiration; once again,
this trend is led by the females, since with the exception of the old
est generation of Mushuau subjects, pre-aspiration is an exclusively
male feature among members of the mini-sample.
It is not to be assumed, however, that the trend towards reten
tion of a single of the variant possibilities offered by the divergent
territorial group backgrounds of the oldest members of the commu-
70 SANDRA CLARKE
nity represents the sole resolution of the competing variant problem
in present-day Sheshatshiu Montagnais. To illustrate this, let us
examine yet another possible realization of originally pre-aspirated
sequences, namely, a fricative variant. Fricative realizations are es
pecially common in the dialects of those with Mushuau or northern
backgrounds, and are also found, though to a lesser degree, in the
speech of those with Maskuanu or Lower North Shore affiliations (cf.
MacKenzie 1980:69-69).
Table 4 represents a fricative pronunciation of the /he/ sequence,
which, to judge from the usage of the 19-subject sample, is the pre-
aspirated sequence most subject to the fricativization rule in present-
day Sheshatshiu Montagnais, particularly when it occurs in word-
final position. As Table 4 demonstrates, the highest rate of frica
tive usage for this variable is associated with subjects of Mushuau
background representing the two older age groups of the generational
mini-sample. Such usage is, however, by no means exclusively indica
tive of a Mushuau affiliation, since it is found, although to a lesser
degree, among parents and grandparents of the other two territorial
groups. What is interesting is that, while fricativization appears to
be on the decline among the younger generation of Mushuau speak
ers as they move towards a degree of usage more characteristic of
other territorial groups in the community, the fricative realization
does not appear to be declining to any appreciable degree within the
other two groups. In actual fact, the fricative variant appears some
what more closely associated with male rather than female speech;
indeed, among the Uashau family of the mini-sample, it is clearly a
male marker.
If eventual analysis of the full 87-subject sample reveals this trend
to be significant, it will confirm an alternate resolution of the com
peting variant dilemma: certain of the possible linguistic realizations
offered by the dialect mixing situation to be found in the Sheshatshiu
speech community may come to take on a social significance which
they did not previously possess. In other words, the usage of the
mini-sample suggests that a feature which among older age groups
serves as a territorial group marker may be currently undergoing
reinterpretation as a marker of male speech.
LINGUISTIC VARIATION 71
/s/ or /§/ plus Consonant
To this point our examination of consonant variables would sug
gest that at least two solutions to the problem of competing variants
are to be found in present-day Sheshatshiu Montagnais: on the one
hand the trend towards maximal simplification or levelling though
the elimination of all but one variant; on the other, a tendency to
wards retention of two or more variants, with ensuing specialization
of social function for each. Whether or not these trends prove to
be significant, they are by no means the only ones to be observed
among the Sheshatshiu mini-sample. Examination of a third conso
nantal variable — /s/ or /§/ plus stop — reveals neither of these
two trends to be in evidence. Here, territorial group affiliation seems
to retain its significance for subjects of all generations, to a greater
extent than it did in the case of the variables previously examined.
In other words, we are dealing here with a more conservative linguis
tic variable, the competing variants of which do not appear to have
undergone any appreciable amount of levelling or reinterpretation
among younger generations of speakers.
In most dialects of Montagnais, the distinction between pre-
consonantal /s/ and /§/ has been levelled. In the northern or Mushuau
area of the Quebec-Labrador peninsula, the usual outcome of the
neutralization process is [s], while it is [§] in the Maskuanu dialects
spoken on the Quebec Lower North Shore. As in the case of the
variables previously examined, both of these variants are to be found
among the oldest residents of the community of Sheshatshiu, depend
ing on territorial group affiliation.
Tables 5 and 6 summarize usage of the [s] variant, among the
mini-sample, with respect to the consonant clusters /st/ ~ /§t/ and
/sk/ ~ /Sk/, respectively. Since no clear male/female usage pattern
is discernible within the 19-subject mini-sample, these Tables simply
represent mean usage over both sexes for each of the three territorial
groups under investigation. Table 5 clearly indicates that, over all
three generations, Mushuau subjects are the greatest users of the [s]
variant with respect to the /st/ ~ /§t/ cluster, while the means of
Table 6 reveal that Maskuanu subjects use an [s] pronunciation for
the /sk/ ~ /§k/ cluster much less frequently than do the other two
territorial groups. In other words, the variable under investigation
appears to be the first for which no apparent resolution has been
72 SANDRA CLARKE
found for the competing variant problem: thus favoured realizations
among all three generations clearly continue to indicate subjects'
territorial group affiliation.
The brief survey of the distributional patterns in our 19-subject
mini-sample that are associated with three different consonant vari
ables reveals that not all linguistic variables may be expected to be af
fected, at the same rate, by the levelling tendencies which are clearly
operant with respect to competing variants in present-day Sheshat
shiu Montagnais. One final pattern should be examined, however,
since it differs from all three of those previously investigated. For in
this case, the solution to the competing variant problem appears, on
the basis of the usage of our 19-subject mini-sample, to be the alloca
tion of specialized linguistic functions to variant pronunciations that
previously characterized membership in particular territorial groups.
Word-initial /a/ and /*/
The variables which illustrate such a development are two word-
initial short vowels: /a/ and /i/. A survey of the regional dialects of
the Quebec-Labrador peninsula reveals markedly different variants
in the case of these two variables. In the Uashau dialect spoken at
Sept-Isles and Schefferville, the tendency in the case of both of these
vowels is towards procope or deletion, so that words like /aku:p/
'coat' and /iskwew/ 'woman' would both tend to be pronounced as
monosyllabic through deletion of the initial vowel. In the Maskuanu
or Lower North Shore dialects, however, both of these vowels when
used word-initially tend to undergo a lonsing rule, resulting in [a]-
like and [i]-like variants. Among older Sheshatshiu speakers of differ
ent territorial group backgrounds, then, both tendencies are present.
Table 7 appears to confirm that the trend towards deletion is most
prevalent among the older Uashau-affiliated subjects of the commu
nity; thus the deletion rate of the Uashau grandparents and parents
surpasses that of their non-Uashau counterparts. Similarly, Table 8
confirms that initial short-vowel tensing is considerably more charac
teristic of the Maskuanu-afiliated subjects, among the parents' and
particularly the grandparents' generation.
Yet Tables 7 and 8 also indicate that the younger generations of
the community appear to be developing a highly interesting solution
to the competing variant problem. Table 7 reveals that, in the case of
LINGUISTIC VARIATION 73
initial /a/, the youngest generation has opted in favour of the deleted
variant, regardless of territorial group affiliation. Likewise, Table
8 indicates a considerable increase, among non-Maskuanu-affiliated
younger speakers, of the tensed [i] variant. In other words, the com
peting variant conflict appears in this case to be undergoing resolu
tion through the development of a differentiation in linguistic func
tion for the two principal territorial group variants. Were the pro-
cope or deletion rule to come to apply in the case of both word-initial
vowels, a word-initial vocalic contrast would be lost. The solution
has been a tendency to restrict the originally Uashau deletion rule
to words in initial short /a/, and to apply the originally Maskuanu
tensing rule to words in initial short /if. The two rules, then, have de
veloped a specialized linguistic function which was clearly not present in the input dialects.
Conclusion
The above examination of fewer than half a dozen phonologi
cal variables in the casual speech style of 19 Sheshatshiu residents
suggests that the problem of competing territorial group variants is
being dealt with in a number of interesting ways. These range from
the elimination of all but one variant to retention of two or more; the
latter trend may be accompanied by innovative tendencies, through
the association of particular linguistic variants with specialized social
or even linguistic functions.
What, however, do these observations have to offer concerning
the two hypotheses advanced earlier in this paper? With respect
to the second — namely the relative importance of the subject vari
able of sex by comparison to the social variables of age and territorial
group affiliation — it is clear from analysis of the casual speech of the
19-subject mini-sample that subject sex may play at least as impor
tant a role in the Sheshatshiu speech community as it does in larger
urban areas. As to the first hypothesis, to the effect that linguistic
change in Sheshatshiu Montagnais is not clearly progressing in the di
rection of any single one of the input dialects, there is some evidence
from the data examined in this paper that the more homogeneous
dialect developing among younger speakers is being more influenced
by a Uashau-like speech model than by any other of the regional
source dialects. This is particularly evident in the increased use of
74 SANDRA CLARKE
an [s(s)] variant for the /3c/ variable, and in the loss of aspiration
in pre-aspirated clusters. What is especially interesting, however, is
that in these two instances the Uashau-like variant also happens to
represent the least phonetically complex of the competing variants,
in the sense that it is the variant most subject to regular rules of
phonetic conditioning: an assimilation rule in the case of the /§£/
variable, and a consonant cluster simplification rule in the case of
pre-aspirated clusters.
For the moment, it is impossible to conclude which of the two
possible explanations just outlined is the correct one; that is, it is
unclear whether the Uashau speech variety represents a type of pres
tige model within the community of Sheshatshiu, or whether the
favoured variants have in these two cases emerged merely as the re
sult of regular rules of phonetic conditioning. Even though in both
cases female subjects are in the lead with respect to the adoption of
the phonetically conditioned variant, this observation is of little help
in selecting one or the other of the two possible explanations. As a
result of various urban sociolinguistic investigations, it is by now a
commonplace observation that females are typically more advanced
with respect to the adoption of prestige linguistic features. Never
theless, it has also been noted (e.g., Milroy 1985) that females may
in certain urban contexts be in the vanguard with respect to linguis
tic change that results in phonetic simplification. It is to be hoped
that analysis of the full 87-subject sample will more clearly indicate
which of these two competing explanations is the more likely one in
the case of the speech community of Sheshatshiu.3
REFERENCES
Clarke, Sandra, and Marguerite MacKenzie
1984 Language Change in Sheshatshiu Montagnais. Pp. 225-240 in Pa
pers of the Fifteenth Algonquian Conference. William Cowan, ed. Ottawa: Carleton University.
3The research reported on in this paper was made possible by a team research
grant from the Institute for Social and Economic Research of Memorial University.
The principal researchers are, in addition to the author, Marguerite MacKenzie, Jose Mailhot and Adrian Tanner.
LINGUISTIC VARIATION 75
Cowan, William
1977 *xk/ek proto-algonquien dans le Montagnais du 17e siecle. Pp. 143-150 in Actes du Huitieme congres des Algonqumistes. William Cowan, ed. Ottawa: Carleton University.
MacKenzie, Marguerite
1980 Towards a Dialectology of Cree-Montagnais-Naskapi. Ph.D. thesis, University of Toronto.
Mailhot, Jose
1985 La mobilite territoriale chez les Montagnais-Naskapis du Labrador. Recherches amerindiennes au Quebec 15:3-11.
Milroy, James
1985 The Methodology of Urban Language Studies: The Example of Belfast. Paper presented at the First Symposium on Hiberno-English, Trinity College, Dublin.
Trudgill, Peter
1984 Contact and Mixture in Colonial English dialects. Paper presented at the NWave XIII Conference. University of Pennsylvania, Philadel
phia.
76 SANDRA CLARKE
FIGURE 1
GENERATION
1 (Grandparents)
2 (Parents = son + spouse)
3 (Children)
TERRITORIAL GROUP
MUSHUAU
M
1 (69)
1 (44)
2 (21, 15)
F
1 (55)
(stepmother)
(deceased)
2 (24, 18)
UASHAU
M
1 (72)
1 (38)
1 (17)
F
1 (65)
1 (42)
1 (19)
MASKUANU
M
1 (76)
1 (52)
2 (31, 16)
F
(deceased)
1 (49)
1 (26)
TABLE 1. 19-subject 3-family mini-sample
(M - Male, 7 - Female. Age of each subject is provided within brackets)
(Mushuau -- having affiliations with the Kaskapi speakers of Davis Inlet/Schefferville)
(Uasiau. — having affiliations with Sept-Isles/Schefferville speakers)
(Maskuanu — having affiliations with Quebec Lower North Shore speakers)
LINGUISTIC VARIATION 77
GENERATION
1 (Grandparents)
2 (Parents • son + spouse)
3 (Children)
TERRITORIAL GROUP
MUSHUAU
M
.63
1.00
.91
F
.89
-
1.00
UASHAU
M
.83
.93
.95
F
.89
1.00
1.00
MASKUANU
M
.68
.73
.86
F
.85
.93
TABLE 2. [s] or [ss]-like realizations of the /sc/ cluster (Number of tokens per cell ranges from 9 to 48)
GENERATION
1 (Grandparents)
2 (Parents = son + spouse)
3 (Children)
TERRITORIAL GROUP
MUSHUAU
M
.40
.00
.00
F
.27
-
.00
UASHAU
M
.17
.04
.00
F
.00
.00
.00
MASKUANU
M
.00
.19
.00
F
.00
.00
TABLE 3. Pre-aspirated [ht] realization of the sequence /ht/ (Number of tokens per cell ranges from 10 to 58)
78 SANDRA CLARKE
GENERATION
1 (Grandparents)
2 (Parents = son + spouse)
3 (Children)
TERRITORIAL GROUP
MUSHUAU
M
.32
.25
.10
F
.13
-
.10
UASHAU
M
.06
.05
.08
F
.00
.00
.00
MASKUANU
M
.00
.06
.06
F
.19
.06
TABLE 4. Fricative realizations of the pre-aspirated sequence /hc7
(Number of tokens per cell ranges from 16 to 53)
GENERATION
1 (Grandparents)
2 (Parents = son + spouse)
3 (Children)
Overall
TERRITORIAL GROUP
MUSHUAU UASHAU
.83
1.00
.86
.87
.55
.84
.60
.66
1
MASKUANU
.55
.66
.71
.66
TABLE 5. [st] realizations of the /{B}t/ variable (Number of tokens per cell ranges from 8 to 58)
LINGUISTIC VARIATION
GENERATION
1 (Grandparents)
2 (Parents • son + spouse)
3 (Children)
Overall
MUSHUAU
.48
.78
.56
.54
TERRITORIAL GROUP
UASHAU
.43
.52
.78
.54
MASKUANU
.19
.23
.39
.29
TABLE 6. [sk] realizations of the / V k / variable (Number of tokens per cell ranges from 9 to 59)
GENERATION
1 (Grandparents)
2 (Parents = son + spouse)
3 (Children)
TERRITORIAL GROUP
MUSHUAU
M
.25
.57
.85
F
.36
-
.91
UASHAU
M
.45
.78
.90
F
1.00
.83
1.00
MASKUANU
M
.36
.74
.89
F
.56
.83
TABLE 7. Deletion of word-initial short /a/
(Number of tokens per cell ranges from 4 to 22)