Perceptual Dialectology and Perceptual Sociolinguistics

1
Perceptual Dialectology and Perceptual
Sociolinguistics

• SPRÅKFORUM
• Universitetet i Agder, April 25, 2008
• Dennis R. Preston
• Michigan State University
• University of Copenhagen
• preston_at_msu.edu

2
Linguistics
Perception Linguistics
Production Linguistics
Language Regard
Discrimination Comprehension
Conscious regard Folk Linguistics)
Subconscious regard (Language Attitudes)
Production, perception, and regard
3
Linguistics
Perception Linguistics
Production Linguistics
Language Regard
Discrimination Comprehension
Conscious regard Folk Linguistics)
Subconscious regard (Language Attitudes)
Production, perception, and regard (revised)
4
The end result of this revision on the right is
that a. a speaker has meant to communicate
X, and b. a hearer has understood that the
speaker meant X. What could be simpler! Its
not simple.
5
Underlying Production Linguistics is a domain of
cognitive principles that allows utterances to be
made. Underlying Discrimination/Comprehension is
a (different?) domain of cognitive principles
that allows understanding. Linguists are most
excited about finding those principles that
enfranchise speaking and understanding the
competence that underlies performance. Thats
not our concern today.
6
Scholars have not ignored regard. The right
hand side (tacit knowledge) has been most
exploited by social psychologists under the label
language attitudes the left hand side
(declarative knowledge) has been the domain of
folk linguistics. Im actually sure this is a
continuum, but well pretend that these two sides
are independent and ask why we should care about
the left hand side at all.
7
A THE ETHNOGRAPHY OF LANGUAGE REASON
Folk linguistic studies must be done if we want a
complete ethnography of language for any group.
If we do not know what nonlinguists believe about
language, we lack full information about perhaps
the most important element of their cultures.
8
B THE LINGUISTIC THEORY REASON
Folk linguistic studies should be done if we have
any interest in the insights of those who use
language daily. Why would we assume that any
linguists could not gain clues about language by
listening to the linguistic comments of the folk?
9
C THE LANGUAGE VARIATION AND CHANGE REASON
It would be surprising if folk belief did not
bear on many elements involved in variation and
change. Although much goes on below the conscious
level, not everything does, and folk clues about
winners and losers in language variation and
change should be interesting, perhaps even
explanatory.
10
D THE APPLIED LINGUISTICS REASON
How could one imagine doing applied linguistics
without knowing the linguistic beliefs of the
group with whom the work is to be done. To do so
is to invite disdain for or outright opposition
to ones attempts.
11

• So, maybe its a good idea to do folk
  linguistics, but two reasons are usually given
  not to do it
• Many things in language that the folk talk about
  arent really there at all, or things the folk
  say about language are completely wrong, and
• Things that are there are completely
  inaccessible to folk knowledge.

12
For example, here is William Labovs concern
about nasals Frequently, if you ask somebody
what he thinks of this style of speech
(nasalized), hell say its very nasal and if
you produce a speech of this sort (denasalized),
hell say thats very nasal too. In other
words, the denasalized speech characteristics of
some urban areas and extremely nasalized speech
are treated in the same way. (Labov, discussion
of Hoenigswald 196623-4) But Labov does not
differentiate between the folk and linguistic
taxonomies of nasal.
13
(No Transcript)
14

• Yes, the content of the computational system is
  inaccessible to nonexperts, but if one takes a
  more surface oriented notion of structure, many
  facts are (Silverstein 1981), but may be
  available in different ways (Preston 1996).
• But there is no special value in folk opinions
  being correct. Only one of the justifications
  outlined above (linguistic theory) would hinge on
  that necessity.

15
I would like to examine this accessibility issue
more carefully but cannot do so in detail.
Silverstein (1981) surveys 5 conditions for folk
linguistic availability 1. Unavoidable
referentiality 2. Continuous segmentability 3.
Relative presuppositionality 4. Decontextualized
deducibility 5. Metapragmatic transparency
16

• Even recognizing such detail does not
  characterize how folk attention may express
  itself. Preston (1996) provides the following
  classification
• 1) Availability Not all areas (whether of
  performance, ability, or reaction) have equal
  availability. They may be ranked as follows
• a) Unavailable the folk do not comment on some
  topics (e.g., specific phonological features of
  some accents).
• b) Available the folk will discuss some
  matters carefully
• described by a fieldworker (e.g., deviant
  sentences), but they do not normally do so.
• c) Suggestible although seldom initiated in
  ordinary
• conversation, the folk will comment on topics if
  they arise they do not require elaborate
  description from a fieldworker.
• d) Common topics of usual folk linguistic
  discussion.

17
2) Accuracy Although it has no bearing on the
value of the data, folk descriptions of
language may be inaccurate or accurate. 3)
Detail A linguistic object may be characterized
with great specificity or none. a) Global
for example, the phonological detail of an accent
might be unavailable, but that does not limit
comment on the accent. b) Specific in some
cases, linguistic characterization is detailed
(e.g., accounts of speakers who are said to
drop their gs in -ing forms). 4) Control
In both account and performance, folk linguists
may or may not control the variety (or any
aspect of it) under consideration.
18
Given these conditions, one may ask why language
is ever overtly noticed. The Japanese
sociolinguist Takesi Sibata has a simple (and I
believe partly correct) explanation ?the
average language user is so involved with
communicating that he is usually not conscious of
the words he uses (1971375). And I would add
not conscious of the words others use either.
Even items that have the linguistic and pragmatic
character to be available, because of what we may
call the communicative mandate, go unnoticed.
19
What sorts of language acts allow us to overcome
the communicative mandate? Again Sibata has
identified at least part of the answer It
appears to be natural for forms which differ from
those which one usually uses to attract ones
attention (1971374). And I would add to this
usually uses or that one expects to be used
....
20
Step a Speaker
produces a feature X. Step b Hearer notices X,
since their own rule (or expectation) is
different. Step c Hearer recognizes that X is a
feature used by person type Y. Step d Hearer
accesses caricatures of Ys. Step d Hearer imbues
X with (some of) the characteristics retrieved
at step c. (Step e Hearer utters folk remark.)
21
Important fact Language itself does not carry
language regard features intrinsically. This
model suggests that regard features are formed by
an association between language features (at any
linguistic level) and nonlinguistic caricatures
held about groups of speakers.
22
If all this is true, regard responses of any sort
are only clues to what we are really after The
identity of person regard concepts and their
application to language material. Not everything
we have done so far in folk linguistics is wrong,
but I recommend taking a careful look at our
methods (and results) within this more
cognitively oriented model.
23
Perhaps the best-known techniques in folk
linguistics have been done within perceptual
dialectology. Two techniques hand-drawn maps
and ratings of areas will illustrate
qualitative and quantitative approaches to such
folk data, respectively, and I will relate the
findings to what we have already discussed.
24
Young female respondent from Crewe (Montgomery
2007)
25
Hand-drawn map of US dialect areas (Chicago,
1984, age 18, EA, male, coach)
26
There are many ways to talk about this map one
might investigate the areas and the labels. This
respondent labels Chicago as Normal talk for the
average person. Detroit is a place for Black
fro talk, and much of the South is the home of
Southern talk the worst English in
American. This respondent cares for much more
than regional distribution (a linguists a fact).
27
He has made sense of a silly request (where do
people speak differently?), by applying a
sense-making strategy (from b), one that
reformulates our task into one about the regional
implications of good and bad English. Subsequent
quantitative interpretation of numbers of maps
from southeastern Michigan led to a
generalization about where such respondents put
the speech areas of the US.
28
Generalized areas and density of response by
southeastern MI respondents. This map is
interesting as regards outlines, but more
interesting for the number of respondents who
isolate different areas. 94 identify a South,
61 a home area, and 54 a NYC area. Why?
29
If large numbers of Michiganders draw a South and
a NYC area, there must be a reason, one which
appears to be something other than linguistic
distinctiveness. The b here may be that
questions of linguistic difference are sent
through a correctness evaluator before they are
responded to. Armed with this information, it
was possible to go directly for attitudinal
responses that reflected the b fact uncovered in
this previous research.
30
Southeastern Michigan ratings of the 50 states,
Washington, D.C. and New York City on a scale of
1 to 10 for correctness.
31
These Michiganders are not reticent to assert
that Michigan has the very best English and that
places like Alabama and New York City have the
worst. Do Northerners just dislike Southern US
English?
32
Judges LV1 speakers LV2
speakers Attribute Status Solidarity Status Solid
arity A. Majority group LV1 LV1 LV1 LV1 B.
Majority group for Status/in-group for
solidarity LV1 LV1 LV1 LV2 C.
In-group LV1 LV1 LV2 LV2 D. Majority
group for status/minority group for
solidarity LV1 LV2 LV1 LV2 Ryan, Giles,
and Sebastian (19829) outline of language regard
types What type of society is the US? How can we
find out? If Michiganders are LV1, what are
southerners?
33
Southeastern Michigan ratings of the 50 states,
Washington, D.C. and New York City on a scale of
1 to 10 for pleasantness.
34
Auburn University ratings of the 50 states,
Washington, D.C. and New York City on a scale of
1 to 10 for pleasantness.
35
(Preston 1996b)
36
A simplified map of southern Michigan hand-drawn
areas of US dialect differences for trait
evaluation.
37
slow fast polite rude snobbish
down-to-earth educated uneducated normal
abnormal smart dumb formal casual bad
English good English friendly unfriendly
nasal not nasal speaks with without a
drawl speaks with without a twang
38
Michigan ratings of the North and the South for
12 attributes (scale 1 to 6)
39
Judges LV1 speakers LV2
speakers Attribute Status Solidarity Status Solid
arity A. Majority group LV1 LV1 LV1 LV1 B.
Majority group for Status/in-group for
solidarity LV1 LV1 LV1 LV2 C.
In-group LV1 LV1 LV2 LV2 D. Majority
group for status/minority group for
solidarity LV1 LV2 LV1 LV2 This study, in
which the responses are teased out with folk
categories, suggests, unlike the rating task,
that Northerners are insecure about the
pleasantness of their own variety, ascribing that
attribute to Southern speech.
40
1 2 3 4 5 6 7
S
N
ay a
41
(No Transcript)
42
Results for responses to male and female
speakers seven-step monophthongization of guide
43
15 voice samples of regional Norwegian for which
10 or more of the respondents have placed the
voice more than one county away from its actual
site (Gooskens 2005)
44
Percentage of correct dialect identifications
with intonational (sentence and word level)
characteristics removed (monotonized version,
Gooskens 2005)
45
Comparison of Noprwegian and Dutch percentage of
correct dialect identifications with intonational
(sentence and word level) characteristics removed
(monotonized version, Gooskens 2005)
46
Tokyo pitch accent and devoicing results
(Yonezawa Morris, in progress)
47
Kinki pitch accent and devoicing results
(Yonezawa Morris, in progress)
48
LANCHART Speaker Evaluation Experiment
49
Voices and varieties
C Conservative M Modern L Local b boy g
girl numbers order on the tape
50
(No Transcript)
51
Sibilant perception (Strand Johnson 1996)
Tokens were created that varied along the
continuum between sod and shod and placed in
front of a VC segment Listeners were asked to
choose whether they heard sod or shod. Some
were told the speaker was a man, others it was a
woman all heard the same tokens When
listeners believed the speaker was a woman,
they made the boundary between the
categories at a higher frequency than when
they thought it was a man.
52
i
heat
u
hoot
?
hit
?
could
e
hate
o
coat
?
?
head
caught
?
hut
æ
hat
?
hot
General American English
53
The Northern Cities Chain Shift
54
Hi. My name is Monica, and Ive grown up in
Lansing my entire life. My parents are from
Lansing, and my grandparents are from Lansing
too. I went to Waverly High School, which is
about ten minutes away, and I went to St. Gerard
for my primary education.
Monica /?/ is front up /?/ is back Lansing
grandparents /æ/ is high front ten
education /?/ is low back
55
The Northern Cities Chain Shift
56
Nancy Niedzielski played the word last for
southeastern Michiganders. They first heard the
vowel at the ACTUAL position.
Then she asked them to pair that pronunciation
with one of three others the same, one closer to
so-called canonical /æ/, and one even lower and
farther back.
57
(No Transcript)
58
(No Transcript)
59
Overall comprehension results for Peterson
Barney, 1952 (percent correct)
60
Overall comprehension results Preston, 2005
(percent correct)
61
Factive verbs the reason is (is) (that)
... ...the reason that women are more facile
with language and adapt more quickly to
situations is that their powerlessness makes them
more attuned to subtle ... feedback from the
power person. (Niedzielski and Preston
2003196) Presupposition The powerlessness of
women makes them etc...
62
THANK YOU VERY MUCH
63
Why is language regard important to
sociolinguistics in particular? The stability of
norms is important to basic concepts in the
field Evaluation of /r/ is typical of many
other empirical findings which confirm the view
of New York City as a single speech community,
united by a uniform evaluation of linguistic
features, yet diversified by stratification in
performance. (Labov 1972117, italics mine)