upsurge

[Natasha L Warner]

Hi,

this discussion about which pronunciation information one can get from who
has been really interesting.  I'd like to share another example.

I work with the Mutsun (Costanoan, California) community.  The last fluent
speaker of Mutsun died in 1930, so it's a dormant language.  (We don't use
"extinct" or "dead.")  The community has been working on revitalizing the
language from the archival records of it since about 1996, and I've been
working with them since 1997, and some additional linguists have now
joined me.  I'm an outside linguist, not a community member, and much as
we would like to change this, there aren't any community members who have
had the chance to become trained linguists so far.  (Quirina Luna, the
first language leader in the community, is becoming an excellent
informally trained linguist.)

The documentation of the language is entirely written, except possibly for
a very few recordings of songs, which for several reasons aren't likely to
help us much on pronunciation.  There is extensive written documentation,
including several thousand pages of fieldnotes made by the linguist J.P.
Harrington, working with the last fluent speaker, Ascension Solorsano, in
1929-1930.  There are a couple of earlier sources, including one from
approximately 1815, when there were many fluent speakers.  J.P. Harrington
is known from his other work to have been an excellent phonetician, and
this skill is also evident in his Mutsun notes.  He gives very
detailed information on pronunciation, and it's clear that he spent a
great deal of his and Mrs. Solorsano's time and effort on clarifying
exactly which sounds were in which words.

One thing we have focused on throughout our work is knowing the sources,
each one:  knowing what can be trusted in which sources, and knowing what
kinds of transcription mistakes are typical of which sources.  If
Harrington tells us a word sounded a particular way, we trust him.  He
also gives us information about Mrs. Solorsano, the last speaker, that
leads us to believe she was exceptionally fluent for a last speaker.
There were many words she didn't know from the earlier sources, but she
seems to maintain phonological distinctions.  If the earlier sources use a
particular transcription that looks suspicious, we know whether it's a
likely mistake for that particular source or not:  we know which linguists
or amateur linguists neutralized which distinctions based on influence
from their native language (Spanish or English).  We also know which ones
couldn't find a way to write certain sounds, even if they could hear them.
(By "amateur linguists" here, I'm referring to people from outside the
community, not Mutsun speakers, who happened to start writing down native
languages, but weren't trained as linguists.  C. Hart Merriam gives us
excellent information on plant and animal species word meanings, because
he was trained as a naturalist, but his transcriptions are a disaster in
terms of sounds.)

One problem is that Harrington was almost _too_ good a phonetician:  he
often transcribes fine differences among sounds that weren't a distinctive
or consistent part of the pronunciation of the word.  All speakers of all
languages (including English) have variability:  one time you say the word
"bat" with your vocal cords already vibrating before your lips open on the
/b/, and the next time maybe you don't start your vocal cords vibrating
until just after your lips open.  The difference isn't distinctive in
English, so your listener doesn't even notice.  Harrington transcribed
every production he heard in as much detail as he could.  In Mutsun, he
uses about 6 different symbols for different kinds of /s/ or /S/-like
sounds, but the language only has two that are distinct, and he also tells
us they're usually much like English /s/ and /S/ (palatoalveolar
fricative, I mean).  It's _very, very_ clear throughout the data that only
those two are distinct in the language, and that those are the most
typical realizations.  (The remaining other symbols he uses also don't
seem to be predictable allophonic variation, they really do seem to be
random variability.)  I think linguists sometimes forget that natural
language does have a lot of random, free variation, as well as the
systematic kinds we're used to looking for.  One result of his careful
transcriptions is that one time, he may write /sii/ 'water' with one of
the "s" symbols, and another time write the same word with a different
symbol.  Sometimes community members trying to learn Mutsun from the
records try so hard to get everything "right," and this can lead them to
put a lot of effort into trying to learn six different sibilants, and
memorize which word to use which one in, when on the very next page that
information is likely to get contradicted, because it was free variation.

This is where lots of communication between the linguists and the
community members gets really important.  Communication about what the
community's goals are:  To maintain the distinctions of the language in
one's own productive use of the language?  To create a simplified system
that will be easier to learn?  To pronounce one word faithfully to the way
an ancestor did one time, but not to try to speak on one's own?  And I
think trust among the community members and linguists is important: trust
that the linguists are paying attention to what the community's goals are,
and trust that the linguists have the skills to know what they know as a
secure fact about the language and what is unsure, and trust that the
linguists will be honest about what they are or aren't sure of about the
language.  In a dormant language situation, or an endangered language
situation with recent language change, there will always be things that
one can't be sure of about how the language was spoken before.  As Bill
points out, there are things about the phonetics of English we still don't
know, so it's likely that there will be some things one can't be sure of
for any language no matter how good the documentation.  This means that
linguists have to be able to say "I don't know how one should say that, it
might be this way or it might be that way, and here's my best guess, and
here's why I think that." It's also helpful if linguists can say "Yes, I'm
sure that this is how this sound is, even though that other place in the
notes says something different, and here's why I'm sure."  Richard, it
does sound like some more of these kinds of consultation might be a good
thing in your case.

In the case of Mutsun, the project was initiated by the community, and we
linguists work _with_ the community.  We really do consult each other when
decisions about which sounds to try to learn where, or how to write them,
have to be made.  The first practical orthography for the language is one
that a community language leader (Quirina Luna) and I developed together,
making the decisions (reasonably much) together.  The practical
orthography has been revised several times, most often at the instigation
of the community.  A few years ago we did a major overhaul of the
orthography, and that was not only motivated by the community, in fact,
they came up with all the choices of the new symbols, and then ran them by
us to see what we linguists thought of them.  We went with their choices,
as far as I can remember, and I believe they were better choices than the
ones I (a linguist) had come up with a few years earlier.  In this case,
the community and the linguists have always been in total agreement about
the importance of maintaining phonological distinctions the language had,
and we are lucky that the documentation, when pooled across the various
sources, is good enough to be pretty sure what the distinctions are.  But
the most important part of this is that linguists and community members do
consult with each other on these decisions.  (I agree with Bill, of
course, that a community is free to choose to teach a language in some way
that doesn't maintain distinctions if that's what they want.  The Mutsuns
have wanted to maintain the distinctions.)

We've had a different experience with suprasegmentals (stress and
intonation, in Mutsun) than with segments, though.  Even Harrington gives
us very unclear and inconsistent information about stress.  A later source
(Okrand) analyzed his data, and found a somewhat inconsistent pattern he
described as a rule for where to put stress on a Mutsun word, but Okrand
makes it clear that he isn't at all sure about it, and that the data isn't
consistent at all.  We realized that pattern was extremely hard to learn,
as it almost always puts stress in the opposite place from English, but we
tried to teach it for a while in the community.  Then we realized (after
consulting with linguists who specialize in stress) that the stress rule
Okrand worked out probably can't be right, because it's pretty unlikely to
occur in _any_ language of the world.  While that doesn't mean it's wrong,
combined with Okrand's uncertainty about it, and Harrington's extremely
inconsistent use of suprasegmental markings, _and_ the difficulty of
learning it, we decided to stop trying to teach it, and let people put
stress wherever they want, which will probably be based on English.
Linguists and community members made this decision together.  We decided
the slow-down to learning wasn't worth it in asking learners to try very
hard to memorize something difficult, for every single word, that was
probably not even right in the language.

We wrote about some of these issues recently, especially about imperfect
learning of a dormant language, in an article in the first issue of the
new Language Documentation and Conservation journal:

http://nflrc.hawaii.edu/ldc/June2007/

Finally, I want to reaffirm Bill's point about how people's knowledge of
one language's sound system influences how they learn the distinctions of
another language.  There has been a whole lot of research on this in
phonetics during the last 10-15 years (not on endangered language
learning, on bilingualism and language learning generally), so we now know
a lot more about how either a linguist or a child or adult who is fluent
in some language like English is influenced by the sound system of, for
example, English, when they try to learn any other language.  Having the
categories of one language makes it very hard to learn certain
distinctions in another language.  This is true whether you're a linguist
doing fieldwork or an adult or even a child learning a language in the
community, _if_ you already speak some other language.  If you're a kid
learning your heritage language as your first language, from fluent
speakers, _and_ you don't have knowledge of the categories of any other
language yet, then you don't have the problem, of course.  Having
knowledge of some other language's categories doesn't make it impossible
to learn the sound categories of a language, but it does make problems in
getting sound categories right likely.

Thanks, everyone, for the interesting discussion.  Sorry for writing such
a long reply.

Thanks,

Natasha Warner

*******************************************************************************
Natasha Warner
Associate Professor, Department of Linguistics
University of Arizona
PO Box 210028
Tucson, AZ 85721-0028