LINGUIST List 2.150

Sunday, 21 Apr 1991

Disc: Phonology

Editor for this issue: <>


  1. Richard Hacken, Canasta/canaster brouHAha
  2. AVERY ANDREWS, halle's argument
  3. , Phonology (Replies to Wojcik, Hammond, and Hutchinson)

Message 1: Canasta/canaster brouHAha

Date: Thu, 18 Apr 91 13:05:04 MDT
From: Richard Hacken <>
Subject: Canasta/canaster brouHAha
The statement is made that exceptions such as "canasta" are often "perceived
as foreign." In fact, the Oxford English Dictionary places stress on the
penult for both caNASta and caNASter, which may not sound right to ears west
of the caNARy Islands, but is British preference (for the latter) since the
mid Nineteenth-Century. The "foreign" origin for both was from the Spanish word
"canastra" for bucket -- Uruguay being the homeland of canasta. Meanwhile
our friend "canister" was apparently of older origins and derived from the
ultimate Latin from which "canastra" was also derived. In any case, since
Canasta as a game didn't sweep the English world before 1948, logic dictates
its Uruguayan stress pattern would have been adopted along with the name, much
as the stress for the word comPUTer is carried into other tongues not only as
a lone stress, but also as a loan-stress.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: halle's argument

Date: Fri, 19 Apr 1991 15:26:32 GMT
Subject: halle's argument
Re Halle's argument, I'm not claiming that there isn't a single mental
structure between the (observationally) allophonic and morphophonemic
branches of the rule, but only that the argument *as originally given*
doesn't demonstrate it (for the minds of contemporary Russian speakers),
because of the possibility of an alternative, historical, explanation
(which, however, does require that we assume that people can initiate &
pick up sound changes that cut across the allophonic/morphophonemic boundary).
Where simplicity arguments are convincing is where there isn't some alternative
explanation to that offered by synchronic mental structure (at what is
essentially Christopher Peacocke's `level 1.5' (Language and Mind, 1986)).

I agree that evidence for a common mental representation does come from
cases where both (observational) branches of such a process undergo
a common subsequent fate (by the way, how many such cases are actually
known?). But I'd deny that Halle & Chomsky just failed to be explicit
about this relevance of kind of evidence - I see no evidence that they
saw any need for it at all.

I think the point is worth fussing over, because it seems to me that
oversimplified simplicity arguments have had a very destructive effect
in the recent history of linguistics, since lots of people have been clever
enough to perceive the flaws in invalid ones, but have been unable to
perceive the force of the ones that actually work. And the resulting confusion
tends to cast discredit on the entire field.

 Avery Andrews (
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 3: Phonology (Replies to Wojcik, Hammond, and Hutchinson)

Date: Sat, 20 Apr 91 00:12:00 EDT
From: <>
Subject: Phonology (Replies to Wojcik, Hammond, and Hutchinson)
Ad Wojcik: It is important to remember that Ulaszyn, the Leningrad
phonologists, and the American structuralists of the 1930's (but
not earlier!!) all agreed in placing the phonemic level exactly
where Halle found it--and found it wanting, that is, AFTER all
neutralizing rules and before all allophonic ones. Halle's argument,
of course, boils down to saying that automatic neutralization rules
and allophonic rules are, invariably, one and the same, and that this
is what is wrong with this level of phonological representation.
It might help to use an English example, viz. cats' (the possessive
plural of cat):

 Morphophonemic kaetzz
 Moscow-phonemic kaetz
 Leningrad-phonemic kaets

As noted earlier, what I call Moscow-phonemic is the level called
phonemic by Baudouin, by the Moscow phonologists, and by Stampe.
Halle's argument, as I think has often been pointed out, does not
say anything about the validity of THIS level of representation.
Ad Hutchinson: It is important to point out, as I do whenever I
get the chance, that Larry Hutchinson essentially discovered
lexical phonology many years before it became official. That is,
both he and the lexical phonologists seem to want to have their
cake and eat it too by allowing a phonemic level (which seems to
be close to the Leningrad-phonemic one) but allowing phonological
rules to apply in complicated ways, so that the same rule can
apply in some cases before this level and in other cases after it.
Yet, while formally this can indeed be assured, it seems to me that
Halle's point can be strengthened by saying that no one has yet
shown the factual differences between the cases at issue which
would justify such a distinction. The fact that speakers are
SOMETIMES more keenly aware of neutralizations than of allophonic
variation (which was Ulaszyn's SOLE reason for introducing this
level of analysis into phonology) cannot be taken as such evidence
for the reason I tried to make clear recently: there are too
many cases where speakers are UNaware of phonemic contrasts and
where they ARE aware of subphonemic differences to make it possible
to glibly identify the level (if it IS a unique level, which I doubt)
at which naive identifications of sounds are made with the Leningrad-
phonemic level or anything close to it.
Ad Hammond: There seems to be some confusion regarding my
position on exception marking. I said that canAsta, KentUcky,
and so on are not exceptions because they are neither perceived
as foreign nor as forming an unproductive class. I still say
this. Second point: when I contend that English stress is
lexical in nature, I am not obliged to admit that EVERY logical
possibility must be allowed. Greek only allows stress on the
last three syllables of a word, but within that span it is lexical.
If Mike is right in saying that English never allows preantepenultimate
stress, then perhaps English is like Greek. However, I am not
convinced that this is so, as a matter of fact. Words such
as ROckefeller, sAlamander, and so on (as opposed to AlexAnder and
its ilk) do have the primary stress on the preantepenultimate 
syllable. The question arises why there are no words with
preantepenultimate stress that have three completely unstressed
syllables thereafter. As far as I know, the only such words are
derivates such as Admiralty. If this is a genuine generalization,
then it is a constraint on the freedom of lexical stress, but not
proof that English stress is predictable. 
Finally, I think my point about the final /y/ postulated by SPE
for words like industry was not well expressed. I object to this
because this is completely arbitrary: you can do anything you want
once you are allowed to take impossible sound sequences and put
them in your URs. If there are alternations, that is one thing.
But in this case there are not. Furthermore, in the case where
there could be analogous alternations (between syllabic and nonsyllabic
r), they do not work out as one would want. It is NOT the case
that syllabic r appears before suffixes in morphemes which take
penultimate stress and nonsyllabic in those with antepenultimate
stress. If such WERE the case, we could argue that it is reasonable
to postulate a connection between the syllabicity or otherwise
of a final vocoid in English words and the stress pattern. Then,
by parity of reasoning, one COULD extend to the case of final
[i] (or [I], depending on dialect), by saying that where necessary
we write this as /y/. That is, if there were a relevant contrast
between syllabic and nonsyllabic [r], one MIGHT reasonably extend
it to the case of syllabic and nonsyllabic [i]/[I]/[y], but
there is no such contrast. Thus, there is no direct support
for a [y] in words like industry, and there is no PATTERN (as
in Sound Pattern) in English that the final /y/ here would fit into.
Additionally, if we admit that words like Orchestra exist and
are stressed as marked, then there is no reason to even want to
analyze Industry and its ilk differently. That is, whatever
rule gives orchestra its initial stress should also work for
industry. Remember, it is Mike who wants orchestra to count as
having the regular stress pattern and canasta as being the exception.

I, of course, do not believe in anything like URs, but my point is
that even if you do, you SHOULD not want to postulate a final /y/
in words like industry and you SHOULD almost certainly not want
to claim that English stress is predictable. 

[End Linguist List, Vol. 2, No. 150]
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue