ne Sona Gal
Ties (Eh The cory ecommerce Ahr.
‘GungreJ Te soci eee of comer cole sig, Ura
tne A gn Mapa Sanna: (The Hangin i on
‘cai ono ak nea ce)
ove M4 ee or Mp Th Oervat Hangar isd. Baap
Later On he necontan o B ange. In Charles W. Krier (El)
“Goran Unt Morr Snes tagged Enns
Sta Above sl teyond posse) varase wisn). Baey & RWS
Th he ay of anne oon ni Wn BES Cen
Suma Go & Lamy, $n The scuin of eave spats by «apie.
Wench: ator. Weg, Mpc unos fr har of ge
Shs Laan Boor pecans
Language ia. Soeely
Qcaclemic Paso, 1978.
18
The Linguistic Market and
the Statistical Explanation of
Variability
David Sankoff / Suzanne Laberge
‘Though itis well known that the internal differentiation of spoken
language I elated to social class, the scintifc study ofthis relationship
poses a number of very difficult problems. Our experience with the
‘analysis of the Moniteal French corpus leads to the telization that
“irety correlating linguistically variable behavior with social class mem
ally oF dialecticaly, is nota well
re. TC ignores established facts such as that teachers,
'sctors, and receptionists tend to speak a more standard varity than other
people of similar social or economic postion. Adapting the notion of
lingostie market developed by Bourdieu and Boltanski (1975), we under-
took to construct an index which measures specially how speakers"
economic activity, taken in ie widest sense, requires orf necessarily
‘ssocited with, competence in the legimlzed language (or standard,
‘elite, educated, ete, language).
bea relatively easy notion to operationalize
esignment of index values, Spenkers could
be grouped according to occupation, and high valves could be asigned to
those working i educational, teary, plea, and adminsteative Fels,240 Dl Sekt Samm ater
and low values to those for whom the mastery ofthe lesitimized speech
‘arety isnot a erterion of selection: laborers, manual workers, and so
‘on. This approach, however, is unworkable for a varity of reasons
Fefated to the fact thal our task so classify individuals, not occupations
‘Thus, this method would be inadequate without a good deal of ad hoe
patching up’ for howsewives, students, the unemployed, and retired
purtons who make up about hai of our sample, and whose occupations
Ere not comparable with those mote directly involved inthe system of
‘production In addition, an objective” classification of speakers accord
{gto occupation would tend tobe static taking into account ther current
tution ony, No mechanistic weighting of occupations fo arrive at some
Kind ot average seore on an occupational history could really do justice to
the fat that speakers roles inthe linguistic market and their patterns of
Tinguistic Behavior) change and evolve throughout thei ifetime, Thirdly
{he notion ofan cbjective classification of occupations, in terms of the
Tingustic marke. i lusory, Constraction of such a classification would
egesaaily depend in many wayson more a less subjective judgments, on
the part of experts or laymen, with fespect to the nature, the derition,
find the measurement of the various entra used to evaluate a given
‘ccupation, Finally the constrction of soch a classification, prompted
by hopes of cross-language comparisons, would tend toimply that theres
ome astocation of & particular occupation wth » particular role inthe
Tinguiatie market independent of the sociohistorcal conjoncture, an a
umption we reject. The clase structure of different societies at diferent
imes is not the same, nor #8 the situation of a particular occupational
tategory. The methodological dictum that one should standardize such
{dex categories forthe sake of comparability trvalzes the socohistor-
tal analysis which fs crucial to the understanding of the guste market
in each ease
Tne approach we took toward solving this methodological problem was
rot to try to improve the objectivity of the classification procedure, But
{nversely o ty to use a8 fully and as egorously as possible the richness
ofthe intuitions or subjective reactions of individuals who participate in
the daily life of the community. Of course, using this type of data also
Implies a whole range of methodological problems. The two most impor
tant ar follows. Fist ti posible that a judge, that is, person whose
tessment is being used 48 data, fs not experienced or informed enough
to evaluate adequately the participation of a given individual in the in
fuistc market. Second, such assessments inevitably elect the do
{Geology ofthe society about equality of opportunity for economic suc-
est ndependent of social background, about relationships between in-
gence and linguistic behavior, about the nature of the legitimized
‘oe Line a
and Yea ma
language, and so on. These two types of problem can never be totally
Clminated from this type of study, ut our choice of judges was dsiened
fo take into account their effects. We selected eight professional and
fraduate student socilinguists who ad all observed and worked for
Ecveral years on sociolinguistic relationships within the francophone
SSinmunty of Montreal. This ensured that all Judges were as. well
informed as posible. Asfor the second problem, athe judges had shown
intheirown work acitial approach vis vis the sociolinguistic situation
in Quebec, trying to unmask the ideological prejudices prevalent about
testing of language and society. OF course itis impossible to avoid
Completely the influence ofthe dominant ideology, but a critical atitude
best penetrates the tse of beliefs and values which disguises the rla-
tions of power operating in the Hinguisie market.
“Certain sovilingusts(ncliding oe of the authors—S. L.) were not
included inthe panel because thee fart with the inguistic behavior
land interview content of the speakers inthe corpus was foo great Some
judges had previously worked onthe corpus, but were not familar enough
with the interviews 80 that this could affect ther evaluation.
“THE JUDGMENTS.
“The task assigned to each judge was relatively simple. He or she ws
presented with 120 slips of paper, on each of which was described the
ocioeconomic ile history of one of the speakers inthe corpus. This was
fceompanied by an instruction sheet containing essentially the material in
the fst and second sections ofthis chapter. The task, carried out inde-
pendently by each judge, consisted of assigning the individuals to ranked
[rou according to. single criterion, the relative importance of the
age inthe socioeconomic ie ofthe speaker, The number
the umber of speakers in any ofthe groups were not
pected. The judges were instructed to try not to apply objective or
‘explicit eritera fixed a prior (or example, education required to function
inva given poston, whether a job iavolves extensive contact withthe
publ, and soon), bu tobe guided by gencral knowledge and reaction to
the whole of the description to be evaluated
‘Cleary, we were notable to make all description slips strictly compar.
able. This is de in part tothe nature of economic activities; when it
Comes to occupations, for example, it did not seem worthwhile for the
purposes of eliciting judges responses, to elaborate on the specific act
fies of nursery schoolteacher, whereas this would be essential inthe
(mac ofa foreman or supervisor. Furthermore, we deliberately omitted orma Dod Set Suton abr
left vague explicit reference to age and edcatinal level. This may have
‘resulted in les accuracy and less comparable judgments, butt seemed
desirable methodological choice since we wanted to inchue age and
‘education as independent variables in subsequent statistical analyses. On
{he other hand, to take particularities of individual personal histories into
account, we often included in the descriptions details about parents ot
spouses to allow a better perception of the economic context in which
Speakers are situated, Thus the occupation of the head of the household
‘was given whenever possible in the ease of younger students since this
indicates the background from which they come. For housewives, details
bout ther husbands are indicative of thet socal interaction, with which
[roups and to what end
‘The slips were presented to each judge in a random order and were
‘denied only by 8 temporary code
CONSTRUCTION OF THE INDEX AND QUESTIONS
(OF CONSISTENCY
‘The classification task caried out by each judge resulted in a numberof.
ranked groups the rank of each group indicating the importance tot of
the legitimized language
‘The first step in manipulating these data was to convert exch judge's
‘rouping to a numerical seale. To render comparable the seals from
ferent judges, gven the variability inthe number of groups constructed
(anging between 4 and. 12), we simply assigned the value 0 10 each
‘member ofthe group judged most peripheral othe linguistic market. and
140 the group most central. The remaining groups were spaced at sql
imervals between 0 and
‘The aggregate index foreach speaker was thus defined asthe mean of
the values assigned to him or herby the eight judges. Figure 1 shows the