Re: Directionality of Sound changes

--- In phoNet@yahoogroups.com, "H.M. Hubey" <HubeyH@M...> wrote:

>
>
> Richard Wordingham wrote:
>
> > --- In phoNet@yahoogroups.com, "H.M. Hubey" <HubeyH@M...> wrote:
> > >
> > >
> > > Richard Wordingham wrote:
> > >
> > > > --- In phoNet@yahoogroups.com, "H.M. Hubey" <HubeyH@M...>

wrote:

> > > > >
> > > > >
> > > > > Richard Wordingham wrote:
> > > > >
> > > > > > --- In phoNet@yahoogroups.com, "H.M. Hubey" <HubeyH@M...>
> > wrote:
> > > > > > >(Personally I would also disallow k>s.)
> > > > > >
> > > > > > In one fell swoop, yes. However, k > c > tS > S > s is

not

> > > > > > impossible.
> > > > >
> > > > >
> > > > > Not believable. Most languages have ptksn. It is S that
> > develops
> > > > after s.
> > > >
> > > > Non sequitur.
> > >
> > >
> > > Here is what I mean. By no means am I being provacative.
> > >
> > > I am using a general law-like concept e.g. languages that have
> > voiced
> > > stops also have unvoiced stops.
> > >
> > > 1. Most languages have ptksn
> > > 2. Many languages also have a 2-way contrast of sibilants e.g.

s,

> > and sh
> >
> > I would call these 'statistical universals'. [p] can be stably
> > missing in the presence of [b], as you have pointed out. It is
> > better to say there are at least three points of articulation for
> > the stops.
> >
> > > 3. Few languages have 3-way contrast e.g. Semitic and Chuvash
> > >
> > > Now, if a 3-way contrast is learned by 2-way speakers, the 3
> > sibilants
> > > might collapse into a
> > > 2-way system. What I doubt is a language losing its 2-way

contrast

> > s, sh
> > > and developing
> > > a single sibilant.
> >
> > Historically attested examples are a bit thin on the ground. Can
> > some one check how many sibilants modern Indic languages have, if
> > one ignores Sanskrit loans.
> >
> > Lithuanian has 4 sibilants - [s], [S], [z], [Z].
>
>
> I should have said "unvoiced".
>
> > In the Slavonic
> > languages, the etymologically corresponding sounds are [s], [s],

[z]

> > and [z]. Thus the shibilants (whose correspondents in Centum
> > languages are velars) seem to have merged with the sibilants.

For the *attested* merger a slightly different pair of voiceless
sibilants (laminal and apical alveolar sibilants), you may care to
take a look at
http://www.buber.net/Basque/Euskara/Larry/note_19.html .

There are claims that the opposition between /s/ and /S/ is lost in
some forms of Irish English. This is a substrate effect; they are
redistributed in accordance with the allophonic laws of Irish. If
true, this should be as good as attested. Of course, substrate /
superstrate / adstrate effects probably form their own class of
sound changes.

> > Spanish used to contrast at least to sibilants [s] v. [S], and

may

> > have has voicing contrasts. Today, it only has [s].
>
> did s and S collapse to a single one? No.
>
>
> > [S] has become
> > [x]. The affricates are still around, but with some changes,

such

> > as [ts] > [T].

I believe the Spanish changes are attested, but I have not seen any
account of the evidence. I suppose the derivation of
English 'sherry' from Spanish _Jerez_ would be evidence for
inference rather than attestation.

> Here is where the "regularization"rules (postulates?) become

useful.

> They are supposed to
> be derivable from data. Is this attested or inferred?
>
> >
> > In other words, I am using another general idea (call
> > > it statistical or typological)
> > > that languages on the periphery have less phonemes e.g.

Hawaiian

> > (13)
> > > than those in the
> > > Main Theater of History (e.g. Mideast, crossroads of 3

continents).

> >
> > As another example, 72% of languages have at least 2 liquids.

That

> > has not stopped many Tai dialects from collapsing [l] and [r]
> > together, or losing one without replacement (via [r] > [h]).
>
>
> Are these inferred or attested? I do not believe that most of the

data

> that passes off
> as being 'true" is really true. They are all inferred.

For the Bangkok merger of /l/ and /r/, I would say attested. They
are distinguished in the native script, and correspond to the
Indic /l/ and /r/ in loanwords. Thai writing (and, to a large
extent, spelling,) is 700 years old. The merger is described as a
_Bangkok_ phenomenon, but I cannot bear witness to their being
consistently distinguished in rural dialects of Central Thailand.
The distinction does not appear to be artificial, for cognates
across the Tai-speaking area bear out the same contrast.

This contrast is also maintained at opposite ends of the Tai world -
in both Ahom (a language of Assam) and the Wu-ming dialect (a North
Tai dialect, spoken in the central part of Kwangsi).

I'm not sure whether the change /r/ > /h/ should be regarded as
inferred or attested. For attestation, I quote Li on the subect -
'Among the SW dialects, it is preseerved as r- in Ahom and siamese,
but Lü [= Tai Lue, Ethnologue code KHB] has a literary pronunciation
hr-, a voiceless r-, for the common h- in ordinary sppech; other SW
dialects show simply h-.' In the new Tai Lue script (
http://www.omniglot.com/writing/tailue.htm ) the 'low' "h" looks
suspiciously like the Thai "r". 'Low' in this context means for the
series of tones for words with formerly voiced initial consonants.
Words evidencing this change belong to this group. Similar spelling
evidence can be seen in Lao, where the low "h" descends from "r".
One can look this up in the orthography section for Siamese (code
THJ) at Ethnologue. Do you count all this as inference or
attestation?

There is also a change /hr/ > /h/ in Central Tai and most of SW Tai,
but I think this has to be inferred. Ahom is the only dialect
reported by Li to have escaped the change - it has /r/. Wu-ming
also has /r/. This change seems to have preceded Siamese writing.

Richard.