Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

The lexical sets

181 views
Skip to first unread message

Andrew Usher

unread,
Jun 28, 2014, 6:12:44 AM6/28/14
to
I decided to make a revision of John Wells's lexical sets, because I find
them so useful but find that they have a few defects. I aim for the original
goal of describing both GA and GB (= RP), but other accents may be attempted
as well.

Because the NORTH/FORCE (= horse/hoarse) merger is complete in both reference
accents, I have dropped NORTH altogether. The merged vowel in GA is certainly
(by the evidence of my ears) closer to the unmerged FORCE than to NORTH; the
merger is ongoing - in the same direction - in ENE and in Ireland; and in
accents that do still distinguish THOUGHT + R can stand in for NORTH as they
are the same in such accents.

PALM can't be used as a keyword in GA anymore because of the requirement that
all the index words have an agreed pronunciation. I have used SPA, I believe
suggested by Amy Stoller; though I'd prefer FATHER it breaks the rhythm of
all stressed vowels having monosyllabic keywords. Indeed for many Americans,
who pronounce the L in 'palm', it now falls in the SALT set that I have added
- this set is also due to Wells I believe - which is really needed for both
GA and GB as it's shifting from THOUGHT to LOT in GB, while in GA the vowel
can be unique (between THOUGHT and STRUT).

I have also added the TOUR set for representation of the modern GA output
in foreign 'ur' words - including 'tour' but not usually 'tourism' or 'tourist'
- which is phonetically GOOSE + R. It still requires its own set due to
generally being merged with CURE in GB, while Americans may not rhyme the two.
The varisyllabicity of 'tour' is no different from that of words like 'power'
and does not need to be represented in broad transcriptions.

Further, 'Canadian raising' (which is really historic non-lowering) is shown
by having PRIDE beside PRICE. The distinction has become phonemic due to some
words - always including 'spider' - having the 'wrong' vowel is distinguishing
American accents. This distinction is also very important in Southern American
accents. The parallel distinction for the MOUTH diphthong is not yet added,
because it is more restricted geographically and does not seem to be yet
phonemic beyond the idiolect.

In the following table, the most characteristic quasi-phonetic representations
for GA and GB are given, using ASCII IPA. I have used Geoff Lindsay's update
of the GB vowel system, except for not merging the values of STRUT, NURSE,
and schwa. While the SPA/LOT (= cot/caught) merger is now widespread in GA and
must regrettably be considered standard, it is still not represented here. For
those with the merger, the merged vowel can range between the former values of
the two. The values of FACE, FLEECE, GOAT, and GOOSE are given as monophthongs
for GA and diphthongs for GB; it is not suggested that the reverse do not occur
but monophthongs in GB are now regional or archaic.

The actual phonetic value of that sound symbolised V is nearly IPA turned-a in
both accents but the transcription V (the pronunciation preserved in Scots) is
universal and I don't seek to change it. This value also occurs as an allophone
of schwa in morpheme-final position; before a pause it may be categorical. The
symbol A is often central in GA (not in GB); but IPA lacks a unique symbol for
this. This lack of an unambiguous symbol for the world's most common vowel is
the worst defect in the IPA.

Finally ~ and / indicate continuous and discrete variation, respectively, as I
see it. In all cases the first is the transcription I would prefer if there are
no compelling reasons to distinguish.

Name GA GB

FLEECE i Ij
KIT I I
FACE e Ej
DRESS E E
TRAP & a
BATH & A:
SPA A A:
LOT A O
THOUGHT O o:
GOAT o V"w
FOOT U U
GOOSE u Uw
STRUT V V
NURSE R V":
NEAR ir ~ Ir I: / i@
SQUARE er ~ Er E: / e@
START Ar A:
FORCE or o:
CURE Ur ~ R U: / u@
TOUR ur U: / u@
SALT O ~ V o: / O
PRICE aj Aj
PRIDE aj / Vj Aj
MOUTH aw aw
CHOICE oj oj

commA @ @
lettER R @
happY i i ~ Ij

Characteristic mergers:

GA TRAP=BATH SPA=LOT
Some accents: PRICE=PRIDE
Before R: DRESS=TRAP=SQUARE KIT=NEAR LOT=THOUGHT=FORCE STRUT=NURSE

GB BATH=SPA=START THOUGHT=FORCE CURE=TOUR PRICE=PRIDE
Unstressed: commA=lettER

Notes:

- The membership of each lexical set is NOT fixed. The pronouncing of one
word with the 'wrong' value does not imply a merger of the two sets. For
example, 'poor' may have CURE, TOUR, or FORCE; the latter is certainly the
most common nowadays but does not imply a CURE/FORCE merger. The index words
are chosen so as to be not subject to this variability in either GA or GB.

- 'Foreign A' words may be assigned to the TRAP or SPA sets with some
irregularity. They should not be assigned to BATH, even if they happen to
have TRAP in GA and SPA in GB, such as 'banana', 'soprano', 'panorama'
(second stressed vowel), or 'pajamas' (note that all those have the vowel
in an open syllable followed by a nasal, which is a pattern).

Andrew Usher

Andrew Usher

unread,
Jun 30, 2014, 8:26:28 AM6/30/14
to
For some reason I forgot the CLOTH set, with the characteristic mergers CLOTH=THOUGHT in GA; LOT=CLOTH in GB. I should have added:

Name GA GB

CLOTH O O

Note that the CLOTH set is given with the same IPA symbol in both accents
though it has different mergers. This is not a conservative feature but two
innovations in GB. Though it does not imply exactly identical phonetics, they
may be confusable: Jack Windsor Lewis noted that a word like 'sausage' may
be practically the same in both while being assignable to different phonemes.
In regards to that specific word (and some others) GA is more conservative in
that no word spelled with 'au' or 'ou' (save the exceptional 'aunt') ever gets
LOT.

Andrew Usher

Andrew Usher

unread,
Jan 4, 2015, 10:22:51 AM1/4/15
to
I would like to report that I have now started using Praat on my own voice and fully confirmed all the vowels I posted for General American. The conventional transcriptions are incorrect and misleading; the starting points of SQUARE, NEAR, FORCE resp. are identical to FACE, FLEECE, GOAT; that of CHOICE to that of GOAT; and those of PRICE/PRIDE and MOUTH to either PALM/LOT or STRUT (for the unraised and raised variants) to within the accuracy Praat gives me. The PALM/LOT vowel is actually central of course but since IPA doesn't give me an unambiguous symbol for that - it damn well should, as we shouldn't be shackled to the odd vowel system of 19c. French - the irregularity in my original post must be accepted.

The CURE vowel really does start closer to FOOT than GOOSE, as far as I can tell, but it's not very stable and so it would also be defensible to transcribe the tense vowel (consistent with SQUARE etc.) and say that the difference between CURE and TOUR is one is syllabicity alone. It's still needed because TOUR, unlike CURE, never merges with NURSE. So Coors (beer) can be said in one syllable, but is still never 'curs'.

Andrew Usher
0 new messages