LINGUIST List 2.160

Wednesday, 24 Apr 1991

Disc: Morphophonology (Part 1)

Editor for this issue: <>


  1. "NORVAL SMITH, RE: Morphophonology
  2. , Morphophonology

Message 1: RE: Morphophonology

Date: Tue, 23 Apr 91 10:05 MET
Subject: RE: Morphophonology
Prepenultimate stress & the /y/ debate in English phonology.

It's always been a mystery to me why Americans and the English
(mis)pronounce Scottish names like Aberna/ethy, Abercrombie and the like.
The original stress pattern is /abrnEthi, abrnAthi, abrkrOmbi/ and so on
with primary stress on the penultimate and secondary stress on the
preantepenultimate syllable. An explanation, which I must confess I don't
very much like myself, could be that we have here the SPE /y/ in action
in US and south-of-the-border English:

 abrnethy, abrnathy, abrkromby

with stress effectively on the antepenult (if you include the schwa). The
problem with the /y/ in terms of modern hierarchical phonology is that the
difference between /y/ and /i/ is not made in feature terms at all, but in terms
of syllable structure, i.e. youcan't have a distinction between /i/ and /y/
unless you include syll. structure info. in the lexicon.
Norval Smith
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Morphophonology

Date: Tue, 23 Apr 91 10:23:10 EDT
From: <>
Subject: Morphophonology
To Kac: If Hutchinson in fact never intended to argue for the model
that he presents (one in which Russian voicing assimilation would
be a single rule, for example, but there would still be a phonemic
level), that would be remarkable given that years later essentially
the same idea WAS proposed seriously (under the name of lexical
phonology). Ah, what fools these mortals be...
To Wojcik: I quite agree that we are not far apart on the rather
involved issues of history of phonology, Slavic orthography, and
certainly not on the kind of phonology we consider reasonable.
I do strongly disagree with your interpretation of the i/y thing,
>From the top, (1) It is true that Leningrad phonologists appear
to confuse automatic and non-automatic phoneme-changing rules,
unlike Baudouin (for whom this was THE basic distinction) or
Ulaszyn, who appears to have made the three-way distinction between
non-automatic, automatic neutralization, and allophonic. In this
respect then Leningrad was just like Cambridge, Mass.

(2) Everything you say about Cyrillic Slavic spelling is correct,
it is indeed true that consonantal palatalization is expressed by
having two series of vowel signs, those that indicate that the
preceding consonant is palatalized and those that indicate that
it is not palatalized. However,

(3) If orthography is what explains the perceptual judgements and
such, then on your account the i : y pair should behave the same
as, say, the a : ja pair. Yet Russian speakers absolutely DENY
that the vowel sound in, say, the word spelled r-ja-d, phonetically
something like [r'aet], where ' indicates palatalization and ae
a front low vowel is at all different from that of a word like
ad [at], where the vowel is noticeably backer. Also, while
as you say (and as I said all along) the cases where orthographic
i is used in place of y, e.g., n-o-zh-i [nazhy], are purely
orthographic and children and such often write y in these cases,
that, of course, argues for MY position. That is, far from being
misled by the spelling, they ARE responding to the sound and
correctly identifying the cases where i occurs as different from
those where y occurs. Which, on standard theories of phonemic
perception, they should be unable to do!!!!

(4) I agree that SPE phonology failed utterly to explain alphabetic
writing systems, but, of course, my point is that phonemics, NP,
and such also do not quite work in this domain.

(5) I have already indicated some reasons why your analysis of i/y
in Russian as being due to the spelling is unreasonable. Hus, as
far as I can remember, explicitly called attention to the special
quality of the y sound, as well. Indeed, in order to use two
series of vowel signs to distinguish palatalized from plain consonants,
you do not need to believe that the two series correspond to different
sounds. Indeed, as noted, Russians not only do not do this, they find
it difficult to hear the REAL allophonic differences between the vowels
that occur next to palatalized consonants and those that occur next
to plain ones, EXCEPT in the case of i and y.

Also, it is NOT true TODAY that [i] and [y] are in complementary
distribution in Russian. The letter names i and y are a well-known
minimal pair. In Polish, there are more numerous contrasts. However,
in Baudouin's time, there apparently WAS compl. dist. in both languages.
In any event, what is crucial is that both then and now, speakers
hear the sounds as different, that even students of linguistics and
many trained linguists find the phonemicization of [i] and [y] as the
same phoneme intuitively no less bizarre than English speakers would
the identification of [h] and [ng], and that at the same time the
two continue to rhyme without any difficulty.

Finally, I would add that a lifetime of exposure to the spelling is
not required, I have no access to illiterates or very young children
who speak Polish or Russian at the moment, but I can assure you that
what I say works for kids around the age of 10 or 11. And in my own
case I found out about the whole thing when I was around 15, and
had exactly the reactions I have described here as typical of adult
To Hammond: Again, I think the discussion has helped tremendously
to clarify various issues, even if we continue to disagree on
certain points.

 (1) It is not true that English nouns with preantepenultimate
 stress must have a secondary stress later on somewhere,
 although this seems to be true of underived nouns (as I
 noted). There are derived nouns like Admiralty and bAronetage,
 after all. It may therefore be that the absence of underived
 ones is an accidental gap. This would need to be explored.

 (2) My point about sAlamander and such is that there must
 clearly be a lexical difference between these and items
 like alexAnder, because the primary stress is in a different

 (3) I still maintain that using a final /y/ in words like
 industry is (a) unnecessary because you could treat them
 the same as you would words like orchestra and (b) because
 there is no evidence for such /y/ being anything else than
 a diacritic (in effect, a covert exception marker). 

 (4) As to the general point, about postulating URs, I would
 agree (subject to the usual caveat that I do not believe
 in URs at all) that aspirated and unaspirated [t] are the
 same in URs, so that alternations are not the only basis
 on which URs are formed. However, this is a different
 situation, because we are talking about two phonetically
 similar segments appearing in complementary distribution.
 In the case of industry, I grant you phonetic similarity
 but there is NO distributional basis for the identification,
 PRECISELY because, as I keep saying, there is no evidence
 that a nonsyllabic in that position has anything to do
 with the stress pattern.

 (5) Polish stress is an interesting case, but your description
 is incomplete. Most speakers, including highly educated ones,
 tend to use penultimate stress in most if not all the words
 which theoretically are supposed to have antepenultimate stress
 (e.g. Afryka, Ameryka), AND I think most of these words are
 identifiable as being foreign in origin. Final stress (e.g.
 attache, atelier) is more stable, I think, but is even more
 clearly identified as foreign. Hence, prima facie, there is
 a case for exception marking here. (NB. The native words
 with antepenultimate stress, such as 1pl. and 2pl. past
 tense verb forms are even strongly subject to the tendency
 to use penultimate stress). I myself and my immediate
 family seem to use the antepenultimate stress pretty
 consistently, however, and there must be other such speakers,
 so it is not inconceivable that some people have exception
 marking and others lexical marking.

[End Linguist List, Vol. 2, No. 160]
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue