[Fictionary] CAXON ballot (due 2/14) -- Eric wins!

Pierre Abbat phma at leaf.dragonflybsd.org
Fri Mar 3 05:20:30 UTC 2023


On Thursday, March 2, 2023 11:11:35 PM EST eLLioTT morEton wrote:
> Indeed, human languages have redundancy that reduces the chances of
> comprehension errors.  You can't just combine the sounds of a language
> any old way.  Each language has its own phonotactic restrictions on
> how they can be arranged, so that knowledge of a sound's neighbors
> constrains what the sound can be.  People actually use this kind of
> knowledge to help clear up perceptual ambiguities.  For example,
> English words can start with /dr/ and /sl/, but not with *dl/ or
> /sr/. and if you take a sound X that's acoustically intermediate
> between /r/ and /l/, and put it in the environment dXa, English
> speakers report hearing /l/, whereas if you put the same sound in the
> environment sXa, they report hearing /r/ [Reference 1 below].  When
> listening to words that start with /rd/ or similar un-English
> clusters, English speakers report hearing an extra syllable and have
> trouble distinguishing (e.g.) /rda/ from /r at da/ (@=schwa), whereas
> speakers of Russian (which has word-initial falling-sonority clusters)
> do not [2].

Did that experiment use /r/ (a trill) or /ɹ/ (the usual rhotic in the English 
accents I'm familiar with)? I don't know a Russian word beginning with /rda/, 
but I do know a Russian word /rta/ (рта, mouth's).

I once read of an experiment in which "banket" was presented to one ear and 
"lanket" to the other. They invariably heard "blanket", not "lbanket".

> Vowel harmony of the sort you mentioned (phonetically-based
> restrictions on what vowels can co-occur in the same word) can be used
> by listeners to detect word boundaries [3]; e.g., if /e/ and /o/ can't
> co-occur, then when you hear /kepola/, you know that it has to be /ke
> pola/, not /kepo la/.  But as far as I can tell from half an hour's
> rummaging on Google Scholar, no one has found, or even looked for,
> vowel harmony being used resolve perceptual ambiguity.  I wonder why
> not?

In Finnish, "aou" occur together, and "äöy" occur together, but not with each 
other — except in compounds like "syyskuu" (one of the months) and imitative 
words like "kroohpyyh". In Turkish, likewise, "aıou" don't normally occur 
together in a word with "eiöü". But there are a few native words which have 
such combinations, like "anne" (mother; the Azeri cognate is "ana", which 
obeys vowel harmony). Also compound words, as in Finnish, can have mismatched 
vowels, and borrowed words aren't always forced to comply. So "tirbuşon" 
(corkscrew) is borrowed from a French compound, where "tire" has a front vowel 
and "bouchon" back vowels.

Pierre
-- 
sei do'anai mi'a djuno puze'e noroi nalselganse srera





More information about the Fictionary mailing list