Re: Consonant clusters
- In Swedish voice assimilation is always devoicing in both directions.
Granted Swedish lenis stops are actually voiced only between two vowels or
a vowel and a sonorant but in compounds like _fotboll bakdel utgång_
'soccer, back part/butt, exit' the assimilation is to the point that the
second stop is aspirated!
As for morphological salience counteracting cluster simplification it is
probably not altogether uncommon. However it is not universal. The Swedish
word _väst-kust-sk-t_ 'west-coast-ish-N.SG' is famous as a case of a
monster final cluster and there are two morpheme boundaries in the cluster,
but the word is actually normally pronounced ['v\Es:k8s:kt] although it now
occurs to me that all three morphemes are preserved at least in reduced
I guess syllable timing vs. stress timing might have a lot to do with
clusters being preserved in some languages but lost in others. It should be
noted however that Modern Greek has restructured all old stop + stop
clusters into fricative + stop, although stop + stop have been reintroduced
in loans from Ancient Greek.
As for Greek /ks/- becoming /z/- in English it may be because something
looking like _x_ was used as an abbreviation for final /Vuz/ in Old French.
When the diphthongs that /u/ was part of monophthongized _x_ was perceived
as a spelling for /z/, so I guess _x-_ /z/ is a case of mistaken graphemic
Den söndagen den 30:e december 2012 skrev Alex Fink<000024@...>:
> Let me try to draw this back on track.
> >Lots of natlangs obviously allow some wild and crazy consonant clusters
> >(Georgian, as the prototype example, but also Greek and of course
> >while others simplify them through deletion or assimilation or some other
> >process. I don't know, though, what triggers this. What makes
> >word-initial /pt-/ okay in Greek, but simplified in English?
> I doubt anyone will be able to give many answers to this that are
> satisfyingly non-circular. Initial [pt-] is legal in Greek because it was
> created (most recently by pj > pc > pt), and hasn't been gotten rid of
> since; it's illegal in English because it was gotten rid of (IE contained
> some), and hasn't been created since.
> In view of the question of loanword adaptation, which you seem to be
> getting at, this is maybe a bit facile, as for instance despite English's
> rejection of Greek [pt-], it doesn't reject the originally equally invalid
> [sf-]. I don't know what triggers that either: one can say things like
> that [sf-] was legitimised by the existing [s]+obstruent clusters, but
> that's post hoc and I wouldn't've known how to predict it, nor how to
> predict that [pt-] would resolve to [t-] and not [p@t-] or anything else.
> My own suspicion is that the key thing will be the nature of the phonetic
> cues which speakers use to recognise contrasts, which is something that
> (IME) doesn't get documented in detail very much.
> I've seen some theoretical work attempting to account for it, the great
> numerical majority of it pertaining to optimality theory. But I don't
> believe in optimality theory (at least certainly not in its full breadth),
> because both it seems subject to near-unfalsifiability in its broader
> extensions and it explicitly disclaims relation to actual mental processes
> (first we generate an infinite set of candidates? really?)
> In any case, probably in _written_ borrowings all bets are off. Despite
> an earlier thread on it I stìll have no clue why Greek initial [ks-] voices
> in English, when the by rights 100% parallel [ps-] doesn't.
> >And are there consonant clusters that are *really* unlikely in any
> >language? It seems that a voiced-unvoiced sequence like [bt] would be
> >really unstable and probably therefore become [bd], but is that universal?
> Actually it would rather more likely become [pt]; voice assimilation is
> usually anticipatory, and consonants releasing into a resonant usually
> assimilate less readily than those not. That'll be a pretty frequent
> >One thing that occurs to me is that if there is a lot of information
> >encoded in the consonant, it's more likely to be preserved. So perhaps in
> >Semitic languages where a lot of the semantics hangs on the consonants,
> >there'd be less assimilation of adjacent consonants. But is that true?
> The effect is true: Semitic observably has very few conditioned consonant
> changes for a family of its size and breadth. Information content, which
> here means root distinctness, is probably not a significant reason for
> that, though: there have been plenty of _unconditioned_ mergers of
> consonants, which wreak equal or greater damage on root distinctness. No,
> the reason I think Semitic points to is analogical pressure. Every Semitic
> consonant cluster, except those in purely affixal or closed-class material,
> is subject to an alternation with a form without the cluster. In view of
> the alternation, the original values of the consonants are likely to get
> restored if there is a conditioned change -- or really, are likely to be
> continuously kept restored against conditioned change. The pressure is
> especially great because the consonant sequence is all you have to go on
> for root recognisability in Semitic -- unlike in, say, SAE, where remnant
> ablaut etc. notwithstanding, roots usually contain a whole contiguous
> recognisable syllable which doesn't get altered much except a bit around
> the edges.
> Summary: one way clusters can be preserved is that morphological
> transparency of a juncture can undo, or ward off, conditioned sound change.
On 01/01/2013 16:54, Benct Philip Jonsson wrote:
> As for Greek /ks/- becoming /z/- in English it may be
> because something looking like _x_ was used as an
> abbreviation for final /Vuz/ in Old French.
No - nothing to do with it.
It is simply the modern English habit of dropping initial
plosive at the beginning of 'awkward' compounds. I guess
the habit began when native initial /kn/ changed to /n/ kin
early modern English.
Also to talk about initial Greek /ks/ becoming /z/ in
English is IMHO misleading. Initial Greek ξ which was
transcribed as _x_ that has become pronounced as /z/; and
it's nothing more mysterious than the change of early modern
English /gz/ --> current modern English /z/
>> On Sat, 29 Dec 2012 17:12:04 -0600, Patrick Dunn
>> In any case, probably in _written_ borrowings all bets
>> are off. Despite an earlier thread on it I stìll have
>> no clue why Greek initial [ks-] voices in English,
>> when the by rights 100% parallel [ps-] doesn't.
Strictly speaking, it doesn't. No Englishman (or English
women) ever heard ancient Greek pronounced as ancient Greeks
pronounced it. So to talk about a Greek sound shifting to
something else in English is not very accurate.
What we have here is simply a spelling pronunciation. Greek
ψ was transcribed in Latin by _ps_ and continues to be so
transcribed in English. Apart from the digraph _ph_,
English _p_ is always either [p] or [pʰ], so initial _ps_
got pronounced in early modern English as /ps/, and still
pronounced that way in French till the present day.
But ξ was transcribed not as _cs_ (and certainly not _ks_),
but as _x_. It was then pronounced just as English (or
French) pronounced Latin _x_, which, like the Latin _s_, got
voiced between two vowels, i.e. /gz/. Of course, initial
_x_ does not occur in Latin, but it was only a short step to
general that _x_ before a vowel is /gz/, hence initial x- =
/gz/. This was still the pronunciation being taught for
ancient Greek when I learnt the language in the 1950s, and
it is still the way that initial x- is pronounced in modern
Thus all we have in current modern English is a
simplification of early modern English pronunciation, i.e.
/ps/ --> /s/
/gz/ --> /z/
... with nothing to do with actual ancient Greek
pronunciation, but spelling pronunciations based on
contemporary pronunciations of Latin! ;)
There ant no place like Sussex,
Until ye goos above,
For Sussex will be Sussex,
And Sussex won't be druv!
[W. Victor Cook]