Sound changes versus sound changes

[Larry Trask]

--On Saturday, June 30, 2001 3:01 am +0000 X99Lynx at aol.com wrote:

> I saw somewhere a tongue-in-cheek piece where it was predicted that by
> 2020 the whole French language would be reduced to one single sound -
> "en" - but that French would continue to be written as it is now.  The
> truth behind the humor captures what some think is the extreme nature of
> the French phonological system.

There is nothing "extreme" about French phonology, which in fact is rather
unremarkable by world standards.  It is merely that French is rather more
different in pronunciation from Proto-Romance than are most Romance
languages.

> Despite how well the developments that
> produced French can be traced and the underlying morphology is IE, the
> cumulative phonological system seems worlds apart from for example
> German.  The objective contrast between hearing French and hearing German
> simply does not suggest a close relationship.  Or perhaps any
> relationship at all.

This is hardly surprising.  French and German are not very closely related.
Their last common ancestor was probably PIE, spoken at least 6000 years ago.

Anyway, what languages "sound like" is not a linguistic fact of any
interest.

Spanish and French are rather closely related, but they sound very little
like each other.  The Spanish of northern Spain sounds very much like
modern Greek -- a speaker who knows neither can easily confuse these two --
yet Spanish and Greek are only very distantly related.  And Spanish sounds
quite a lot like Basque, yet these two are not related at all.  Spanish is
exceedingly closely related to Portuguese, but the two languages do not
sound similar: to my ears, Portuguese sounds more like Russian than it
sounds like Spanish.  And Galician is so closely related to Portuguese that
some linguists prefer to regard the two as dialects of a single language --
yet Galician sounds far more like Spanish than it does like Portuguese.

The staccato Spanish of northern Spain doesn't sound all that much like the
drawled Spanish of rural Mexico -- or, for that matter, like the
rhythmically different Spanish of southern Spain.  To a non-speaker of
English, I imagine that the English of Brooklyn and the English of Glasgow
-- not to mention the English of Mississippi -- sound like wholly different
languages.

The Greek of Cyprus sounds far more like Turkish than it sounds like
mainland Greek.  The first time I heard Cypriot Greek spoken, I was
convinced I was listening to Turkish, and I was bewildered because I
couldn't understand any of it (I used to work in Turkey).

> If morphology, grammar and syntax are the least likely things to be
> borrowed (a la DLW's finite verbs)

Dubious.

> then perhaps they are the least likely to change,

Clearly not so.

> and therefore are the features that least reflect accurately
> rates or degrees of change.

Sorry; this makes no sense.  If the grammar of a language changes
dramatically, then the language has changed greatly, and we cannot pretend
otherwise merely because the pronunciation hasn't changed much.

> And perhaps the phonological gap is a truer
> measure, if there is any reliable measure at all.

There is no acceptable metric for determining degree of change.  But it is
certainly out of the question to assign phonological change arbitrarily
greater weight than any other kind of change.

[on Pama-Nyungan]

> But I want to point out that there are equally members of that family that
> appear not to have changed much at all.

I don't think so.  Most Pama-Nyungan languages have not undergone the
exceptionally dramatic phonological changes called 'initial-dropping', but
they have still undergone plenty of change.

It is true that Proto-Pama-Nyungan *<gudaga> 'dog' survives as <gudaga>
'dog' in modern Yidiny.  But then it is also true that PIE *<nepot->
'nephew' survives as <nepot> 'nephew' in modern Romanian.  However, we
should not conclude that Romanian is practically unchanged from PIE: it
isn't.

> In fact, the high convergence
> percentages theorized by Dixon were meant to account for the large amount
> of commonality between many of those languages.  As I understand it,
> those who disagree with him about the degree of convergence have argued
> instead for a high degree of conservatism in those Pama-Nyungan
> languages.  And that does not mean that those languages did not change or
> even change a lot.  It's rather that as much as they did change, they
> never change much.

Well, there are certain typological characteristics which have proved to be
stable in most Pama-Nyungan languages other than those undergoing
initial-dropping.  But P-N languages have still undergone a very
significant degree of change.  Neighboring P-N languages are typically not
mutually comprehensible at all.

> And, perhaps, even if some IE languages show a certain amount of
> "similarity" in sound systems, it in no ways means they did not change
> for 3000 years before attestation.  It may be that they changed often,
> but through all those changes they never changed that much.

Similarity in sound systems is not a good metric for degree of divergence.

Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK

larryt at cogs.susx.ac.uk

Tel: (01273)-678693 (from UK); +44-1273-678693 (from abroad)
Fax: (01273)-671320 (from UK); +44-1273-671320 (from abroad)