LINGUIST List 3.873

Sat 07 Nov 1992

Disc: Phoneticians (last posting)

Editor for this issue: <>

Directory

1. , 3.839 Final Devoicing (Previous topic was 'Phoneticians'

Regarding the issue of final devoicing (or the absence thereof)
in German, Polish, Russian, and Catalan, LINGUISTs will perhaps
remember that the discussion started with a query I posted some
time ago asking if there is any literature documenting the idea
that even the best phoneticians systematically miss and hence
mistranscribe important phonetic phenomena which can be shown
to exist by the use of instruments. I suggested that the fricative
pronunciation of intervocalic 'k' and 'g' in English is one possible
example, and I also noted that there has been a whole literature
in recent years which claims that final devoicing in German, Polish,
Russian, and Catalan is not a 100% neutralization rule, but that small
but measurable differences exist between underlyingly voiced
and underlyingly voiceless finals in these languages, even though,
as far as I knew, no phonetician had ever heard any difference.

 In the ensuing discussion, I asked for
any references that might indicate that I was wrong;
and a list of references was recently posted by
John Local. However, upon reading several of them, I
discovered that they actually agree with me, and that such
phoneticians as Sweet, Jespersen, and Vietor
specifically refer to final devoicing in German.

 Today, John Coleman posted a message saying that he
would like to add some new items "to the references which
John Local recently posted in refutation of Alexis's claim".
(I, of course, assert that the references so far SUPPORT my claim.)

 Almost all the new references are references to the
literature on the alleged instrumental evidence for the
existence of these small differences in German,
Catalan, Polish, and Russian (the literature which I talked
about at the beginning). This literature, as I have already
noted on LINGUIST, claims to have found instrumental evidence
for something which phoneticians had not previously reported
(and hence supports my position). (I should perhaps qualify
this: the Chen article on Russian does not say anything about
the earlier literature, and hence is neutral.)

 It is true, as Coleman notes, that one of these papers, "Port
and O'Dell (1985:455) state: "Furthermore, in a listening test,
German listeners were able to distinguish the voiced and voiceless
pairs with about 60% accuracy --- significantly better than
chance."

 However, this result does not show that any phonetician
had ever heard, and reported hearing, the distinction in
question. It only shows that speakers can distinguish things
with better than chance accuracy on a forced-choice task, which has
nothing to do with what phoneticians can hear. In fact, this
may actually confirm my point, since here is another
experimental technique which seems to be capable of demonstrating
a contrast that phoneticians cannot hear!

 John also refers to Malmberg, B. (1963) Phonetics. Dover
Publications, New York, where it says on p. 52n: "in the latter
languages, the voiced types are always fully voiced, while in
English, German, etc., they are often only half-voiced or even
voiceless without becoming fortes. So a solid distinction between
the series is retained."

 Now, I have not recently reread Malmberg, and I do not recall
the context of this passage, but I am willing to predict that, if
anyone examines the context, they will find that it does not refer to any
alleged lack of final neutralization, but to the fact that in non-
neutralizing environments, speakers of languages like these use
half-voiced or even voiceless lenes instead of voiced sounds. The
fact that he mentions English (which does NOT have final
neutralization) and German (which does) together would tend to
confirm this.
 John then goes on to state:

 I was taught in class that for some speakers and dialects at least
 Final Obstruent Devoicing in German is NOT absolute
 neutralization. . . .

But, as I have pointed out before, it is well-known that
there are German dialects without final devoicing--BUT THESE ARE NOT THE
DIALECTS UNDER DISCUSSION. (I might add that, in addition to
dialects, there may be certain other cases where there is no final
devoicing in some of these languages, e.g., the colloquial
vocatives of diminutives in -a in Russian, such as 'Volod,' [d, =
palatalized d], the vocative of 'Volodya'. But, again, these are
irrelevant, since here the voiced nature of the final consonant is
easily heard.)

 Having said all this, I repeat that many of the world's finest
phoneticians have given us detailed descriptions of German,
Russian, and Polish (Jones, Sweet, Jespersen, Vietor, Shcherba,
Rozwadowski, Benni, Jassem, those are just a handful of the
names that leap to mind). Neither they nor anybody in the
recent literature which claims the existence of small measurable differences
between underlying voiced and voiceless finals in these languages
has transcribed such a difference (again, with the irrelevant
exception of certain dialects and other well known special cases).

 Even the existence of these small measurable differences is
itself at issue, so it is entirely possible that THERE IS NOTHING
THERE TO HEAR in the first place. And if there is nothing to hear,
then it would be puzzling if "most ear-trained phoneticians in the
Ellis/Bell/Sweet/Jones tradition" had learned it.