surge of news on endangered languages...

[William J Poser]

Few if any linguists wish to be the models for people 
learning their heritage language, and in every case that I am familiar
with, where recordings of the language exist, these are used as
pronounciation models. I suspect, therefore, that this is a theoretical
concern. In practice, I wonder whether there are any cases in which
people have learned incorrect pronounciation due to modelling themselves
on a linguist?

This raises four other points that I think are worth raisng.
First, it brings out the desirability of field linguists getting
good training in practical phonetics so that they will be less
likely to make mistakes and getting other people to listen
to the language in the hope that they will observe things that
the first linguist does not. It is unfortunate that the resurgence
of interest by linguists in endangered languages has come at
a time when fewer and fewer linguistics students receive thorough
practical phonetics training.

Second, although it is certainly better for learners to model
their pronounciation on that of fluent native speakers, we have
to recognize that whenever the language is learned by adults
or even by teenagers, and even by younger people outside of an
immersion situation, the result will not perfectly reflect
the model. Once the chain of natural transmission is broken,
it is broken forever. Of course we can try to minimize the
effects of broken transmission, but it is very naive to believe
that people learning their heritage language as a second language,
even with a fluent native speaker as model, will learn it perfectly.

Third, this is a nice example of what bothers me about the idea
that one can adequately document a language in three or four years.
I know what people mean when they say that this can be done,
and when done by skillful people, the result will be useful,
certainly much better than not having it, but subtle, rare, complex
and unexpected phenomena sometimes take even the best people many
years to figure out. Even languages like English and French, which have
been studied by far more people over a longer time than any
endangered language, still present mysteries. To take a language
that has not been studied quite as much, but still no doubt far
more than any endangered language, the precise phonetic cues to
the distinction among the three series of stops and affricates
in Korean is still a matter of debate. We need to recognize
that what can be accomplished in a few person-years, however
useful, will almost certainly not be complete and fully accurate.

Fourth, let me raise a point that may offend some people.
It is sometimes the case that the linguist's pronounciation is
actually MORE accurate than that of the last native speakers.
This situation can arise when the linguist has access to information
from people now deceased, and where the last speakers reflect
the changes that often accompany language death. There are quite
a few reports of changes in the phonological system of the last
generation or two of a declining language, often, though not
always, as a result of assimilation to the sound system of the
language that is replacing it.

I'll give two examples from Carrier. Conservative Carrier has two
sets of strong fricatives and affricates (s, z, ts, dz, and glottalized
ts'). One set is apico-alveolar, as in English. The other set is
lamino-dental. In some communities this distinction has been
completely lost (in which case the lamino-dentals have become
apico-alveolar). In others the older generation preserve the
distinction while the younger speakers have lost it. For language
learners it is quite difficult to learn both to produce and to hear.
I can hear this distinction, if I am paying attention, and can produce
it, with some difficulty and inconsistency, because I have spent
a lot of time with with elders who have the distinction, including
one who would actually correct me if I got it wrong.

It will probably not be very long before there are no longer any
speakers who have this distinction. If I produce it when I speak,
or write about it, surely we should not say that I am wrong
because the last speakers do not make it. It is of course up
to the community to decide what they want to do about teaching it.
In practice, at present, what seems to happen is that students
are not asked to learn it, both because it is so hard for them and
because most of their teachers cannot produce it, but people still
try to write it. 

My second example is more extreme. In one community there are only
three speakers left whom I would call truly fluent, and a handful
of semi-speakers. One of the semi-speakers, the daughter of a
truly fluent speaker, has a severely degenerate sound system.
She has completely lost glottalization and does not distinguish
glottalized stops and affricates from their unaspirated counterparts
either in perception or in production. Roughly speaking, she speaks
Carrier using only sounds found in English, which is by far her
dominant language. Suppose that she were to end up as the last
speaker of the language. Surely we wouldn't want to say that
the documentation that I and other people have produced, which
distinguishes the glottalized series from the unaspirated series,
is wrong. Here again, it would of course be up to the community
what they want to use as a standard - they might choose to standardize
on the speech of the last speaker even though it has lost contrasts
present in her mother's speech, or they might decide to standardize
based on the older form of the language. That is a political and
cultural decision that only the community can make. But, should
the community decide to go with the speech of the last speaker,
it would be wrong to say that a description of the language
as having a distinct set of glottalized stops and affricates
was wrong. It would be a correct description of the language
as spoken by earlier, more fluent speakers.

Bill