Re: Sound Change Susceptibility

Isidora wrote:

<<I have *no* actual data to back this up, but I am going to venture a guess
that certain sounds are going to be more prone to change, because certain
sounds are inherently more difficult to produce.  The interdental
fricatives are objectively more difficult to pronounce than the alveolar
fricatives.  And that is something that I know know for a fact.  I learned
it from my phonology professor who had specifically studied this phenomenon
and found a correlation between the age at which children learn to
correctly pronounce these marked phonemes and the frequency with which the
same phonemes were found in the world's languages.  In addition to
frequency, there is also a hierarchy: /s/ will be learned before /f/, which
will be learned before /T/, if I am remembering correctly - that was over
ten years ago, after all.  /m/ is one of the first sounds that a baby can
pronounce (thus accounting for the frequency of [ma] as a component in the
word for 'mother' in a lot of unrelated languages.>>

Mmm...   I don't know about this.   The first thing that raised a red flag 
was when you said *phonology* professor (1).   This sounds like a version of the 
"lazy speaker" hypothesis, which is basically "anything that's difficult will 
fall out of a language".   It's been applied to phonology, morphology, 
syntax, semantics...   At any rate, my phonetics professor (the same one who 
contends that phonology doesn't exist--I wouldn't go that far, but I certainly am 
closer to that end of the spectrum) contends that most sound changes are 
acoustically based--i.e., it's not about the *speaker* it's about the *listener*.   
For example, the hierarchy you produced up there coincides directly with the 
hierarchy of acoustically salient fricatives: [s] is more salient than [f], which 
is more salient than [T] *to the ear*, and the reason has to do with the 
resonances within the mouth and vocal tract (specifically, there's a node at the 
alveolar ridge, and an antinode in the labial region.   It's at a node and an 
antinode where you'll get the most acoustically salient sounds; in between, 
it's far less so).   A good example is Italian.   Historically, Italian medial 
voiceless stop clusters became geminates, so:

*nokto > notto

And, since I don't know Italian, I can come up with an embarrassing lack of 
examples.   :(   There are more, though.   Anyway, two things are interesting 
about these: (1) It would (hopefully) stand to reason that geminates are by no 
means easier to pronounce than unlike stop clusters; and (2) in all cases (and 
this is pretty much a universal), the cluster became a geminate of the 
*second* sound rather than the first.   There's no real production reason why this 
should be.   If you were to base it on one of the actual consonants (say that 
[t], since it was further forward in the mouth, would always win out), then 
you'd expect a sequence of reversed clusters to produce the same effect--that is 
*notko > notto--but it doesn't: It'd be *notko > nokko.   The reason is that 
the second consonant in a cluster is more salient than the first--or, if you 
prefer, onsets are more acoustically salient than codas.   There are technical 
reasons why this should be, but a good way is to actually pronounce both and 
record them and play them aloud.   While it's possible to tell the difference, 
you have to imagine being able to consistently tell the difference from speaker 
to speaker on a daily basis.   Eventually, what drives the sound change is 
that, since the coda [k] is not acoustically salient, it no longer becomes a 
crucial part of conveying the meaning of the word.  The crucial parts are: (a) 
The [t] onset, and (b) a double-consonant length.   Thus, the /kt/ cluster is 
reinterpreted as a [tt], and the geminate is born.   It's not a conscience 
decision on the part of the speaker that [tt] is easier to pronounce than [kt]; 
rather, [kt] is not *acoustically* different enough from [tt] (in Italian) to be 
counted as a separate set of sounds.   There was probably a time when both 
sounds were in free variation, but, once the [k] was lost, there was, of course, 
no going back.

Also, consider that Arabic still has a full set of pharyngealized stops.   
These sounds are "difficult" to pronounce, but only to people who don't have 
them.   They're not a problem to Arabic speakers, and so they've endured 
throughout the years.

I think my main point is that children's aquisition of their phonology should 
*not* be consider a vehicle for sound change.   If it were, then all 
languages would probably have their phonologies reduced very quickly--the English /r/ 
would be one of the first things to go (having three simultaneous 
places/manners of articulation--one of the rarest sounds in th world's languages, though 
I've seen data of other languages with it).

-David

(1) This is by *no* means a knock on your phonology professor.   It's just a 
reference to the general linguistic war between phonologists and phoneticians. 
  A famous phonetician, John Ohala, at my old university, Berkeley, was 
famous for saying that there was no such thing as phonology.   Phonologists at our 
university were also famous for saying that phonetics wasn't relevant to the 
study of language, and that things like the IPA (whose journal John's been an 
editor for) were ridiculous and to be avoided at all costs--i.e., nothing that 
a "serious" linguist would ever utilize.

Replies