🚀 go-pugleaf

In article <1993Jan23.222721.4306@Csli.Stanford.EDU>, poser@Csli.Stanford.EDU (Bill Poser) writes:

[...]

> contrary to Kenneth Miner's memory, Maddieson's inventories
> are, for the most part, second hand like everybody elses 

I should have guessed that a survey of more than 400 languages (if my
memory <:-> serves) would have to have relied on second-hand data.

  and so raise the
> same problems of interpretation, inconsistent phonemicization, etc.
> It has its uses, but like any other such book, I would treat it primarily
> as a source of pointers to the primary literature and check out anything
> crucial myself. Note also that that it attempts to present a statistically
> useful sample, not an exhaustive list of languages, so relatively rare
> sounds and inventories may well be missing.

I have long felt very strongly that the data problem in
theory-building is vastly underestimated.  No theoretician *can* check
everything, and a lot of theory papers, as least in phonology,
providing they are using the original sources at only one or two steps
removed, are explicit about where the data/earlier analysis comes
from.  But in extended discussions of a particular theoretical issue
where data from a given language is crucial or at least relevant, the
thread may be for all practical purposes severed.

Typically there is an early analysis by someone or other, done in a
framework no longer current.  Someone does a re-analysis in a more
current framework, which then gives rise to a whole slew of theory
papers, all of them resting on the original analysis, which their
authors may not even have read.  The work on theoretical implications
of nasality in Guarani has gone like this, for example; also current
work on metrical analyses of Winnebago accent.  Discussions with other
linguists suggests the pattern is quite general.

I don't think this renders theory-building useless, because I think
progress is being made in spite of it, and that if a hypothesis turns
out to be any good it doesn't matter how it got started.  But I do
tell my students: save yourself some time and check the data/initial
analysis *first*...
> 
> Bill Poser

-Ken
-- 
miner@kuhub.cc.ukans.edu | Nobody can explain everything to everybody.
opinions are my own      | G. K. Chesterton