LINGUIST List 10.1914

Sat Dec 11 1999

Disc: What Exactly Are Allophones?

Editor for this issue: Lydia Grebenyova <lydialinguistlist.org>

Directory

1. Celso Alvarez Caccamo, What Exactly are Allophones?
2. Francisco Dubert, What Exactly Are Allophones?

I, from my profound ignorance, agree with most of Jorge Guitart's recent
message about the psychological thingness of phonemes, but perhaps not 
with everything -- particularly with his use of "representation" vs. 
"realization".

Just some thoughts. The notion "allophone" is inherently relational, yes,
but so is "phoneme" -- what is not relational in linguistics? But a thing
that relates to another thing is still a thing. Strictly speaking, any
phoneme is an allo-phoneme -- no language exists with just one phoneme.
However, would anyone suggest to do away with "phoneme" and to talk only
about the relational process of "phoneming"? ;-). 

So, the point about allophones is not their relational dimension, but what
exactly they relate to, and how. And, in this sense, to me the meaning of
"allo" is sufficiently transparent -- it simply means "other, different". 
Allophones relate to phonemes as representations, yes. But a
representation of an abstract object is not a relationship of difference,
as Jorge Guitart points out. The relationship of difference is the one
established with other phones: an allophone is "another" phone, and, thus,
just as phonemes establish among themselves relations of "otherness"
(Jakobson dixit), so allophones establish relations of "otherness" with
other (allo)phones. 

Now, the second problem may rest on the level of abstraction assigned to
allophones, that is, can any phone count as an allophone? Since variation
in speech is infinite, if the answer is yes then one of the notions is
redundant. I believe that in the speaker's mind not any one phone counts
as "another" phone. Infinite variants in sound are, well, sounds. But only
a few of these variants seem to enter the speaker's competence as
recognizable, contrastable, distinct units. Just run this home-made test
with someone (US pronunciation assumed):

- Please read (or listen to): KILL, LICK -- Are the last sound in kiLL and 
the first one in Lick the same sound? Probable answer: NO

- TOP, POT -- Are the first sound in Top and the last one in poT the
same? -- Probably not, though they may be (=released / unreleased)

- HAND, BAND -- Are the second sounds in hAnd and bAnd the same? -- Yes

But

- HAND, BANK -- Are the "A"'s the same sound? -- A typical Buffalo
answer: NO ;-).

- SHEET, SURE -- Are the initial sounds in SHeet and Sure the same? -- Yes

(A colleague of mine suggests another test: Pronounce LICK with the "l"
in KILL, and viceversa. Then ask if any word sounds weird. Probable
answer: yes. Now pronounce SHEET with the [S] in SURE, and viceversa. Is
there such a thing as an "allophone" of [S]?).

However, this "sameness" of sound in the speaker's mind is not "the same" in
acoustic phonetics. Spectrograms would surely show that the [S] in SHeet
and Sure are evidently different. Therefore, since all "same" phones are
acoustically different from each other, if one extends the criterion of
difference to acoustic properties, two different tokens of, say, a
released, aspirated [t] would be allophonic to each other, which is weird,
to say the least.

So, the (allo-) difference is mental, not physical. Phones are
recognizable mental units that correspond to representations of sets of
physical facts; allophones are recognizable variants of phones -- in
langue, not in speech.

Sorry for my simplicity,

-celso

 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Celso Alvarez-Caccamo Tel. +34 981 167000 ext. 1888
 Linguistica Geral, Faculdade de Filologia FAX +34 981 167151
 Universidade da Corunha lxalvarzudc.es
 15071 A Corunha, Galiza (Espanha) http://www.udc.es/dep/lx/cac
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

To Dan Moonhawk Alford:

In Galician, the lexeme HOXE(adverb) *TODAY* has only /oSe/ as his exponent
(where /S/ is a voiceless postalveolar fricative), but the lexeme
FACER(verb) *DO* has, in the indicative present tense the following forms:

P1 sing /fag-o/
P2 sing /fa-s/
P3 sing /fa-j/
P1 pl /faT-e-mos/ (where /T/ is a voiceless dental fricative)
P2 pl /faT-e-des/
P3 pl /fa-N/ (where /N/ is a velar nasal)


and in the indicative past tense

P1 sing /fiS-e-N/
P2 sing /fiS-E-tSes/ (where /E/ is a mid low front vowel and /tS/ a
voiceless postalveolar affricate)
P3 sing /fiS-o/
P1 pl /fiS-E-mos/
P2 pl /fiS-E-stes/
P3 pl /fiS-E-roN/ (where /r/ is an alveolar tap).

So, the lexeme FACER(verb) has tree roots: /fag/, /faT/, /fa/ and /fiS/,
the P2 sing exponent is /s/ in the present forms and /tSes/ in the past
forms, etc.

I would like to know if in your opinion "allomorphes" are relations too.
Are /fag/, /faT/, /fa/ and /fiS/ relations to FACER?

Thank you and best wishes

Francisco Dubert Garc�a
Departamento de Filolox�a Galega
Universidade de Santiago de Compostela
Santiago de Compostela
Espa�a
e-mail: fgdubertusc.es