> Haukur, I think Oskar was trying to explain to me once that pre-

aspiration (/t:/ as /ht/) is the modern way of doing long consonants -
is that right? My impression of long consonants like [t:] from
Japanese is that we're supposed to keep our tongue in the stopped
position for longer, not make two stops. But maybe that's not how it
works in Old Norse, because I have trouble seeing how one could hear
that the tongue was staying in the "t" position for longer, unless
the stop was always released in Old Norse ... ?

Okay, long consonants, or rather, long stop consonants (p, t, k), can
be a pain. With other consonants (e.g. m, n, l, r, s, and others of
the type called "syllabics", or "continuants"), it's easy to keep the
pronunciation longer. Abruptness is the nature of stop sounds,
however, so "maintaining" their length for an increased duration seem
contradictory (though not impossible, and not uncommon either).

Long consonants (either stop or continuant) are not such a big deal
as long as they're inside whole words, such as
in "þakka", "þetta", "kenna"; in those cases, the cluster just gets
divided up between the syllables - "þak-ka", "þet-ta", "ken-na". And
that's referring to Old Norse pronunciation.

When a geminate (long consonant) is in word-final position, it gets a
little trickier. In Modern Icelandic, only continuant geminates exist
(the stop geminates having been replaced by pre-aspirated stops).
Since MI has a phonetic condition that makes all (or most) consonant
clusters, including geminates, shorten the vowels that come before
them, whether one pronounces the geminate properly is not critical.
But when a continuant geminate stands word-final, its pronunciation
is usually dealt with in this way:

"hann er" -> [hAn nEr]
"húss eins" -> [hus sEins]

Consonant transfer is actually the "default condition", so single
finals get pronounced like this (AFAIK):

"hún er" -> [hu: nEr]
"hús eitt" -> [hu: sEiht]

In other words, the latter part of the geminate is transferred to the
following word. This cannot, however, happen if the following word
starts with a consonant, or if the word is the last one in an
utterance:

"hann fer" -> [hAn fE:r]
"hann." -> [hAn]

As to ON, I'm not sure (and I'd guess no-one is) what the geminate
phonotactics were like. I would personally assume the same as above,
since that is a natural thing to do. Thus we'd have:

"hann er" -> [hAn ner]

And with the stops too:

"satt er" -> [sAt ter]

In pre-consonantal or "utterance-final" environments, I'm not sure
how to deal with it, though. I think that

"satt." -> [sAtt]

is not a physical impossibility, but I don't think it could be called
a preferrable arrangement of sounds, effort-wise. One way of making
the distinction that I can think of, is to pronounce [sAt] with an un-
released [t] and [sAtt] with a released one, with otherwise no
distinction of length.

MI, in any case, solved the problem with its pre-aspiration feature
(though speakers of many other languages might not find that any
easier), which it shares with Faroese and Gaelic (though the latter
language is of a different ilk). Phonetically, [h] is just a
fricative, so I'd suppose pronouncing [hk] intervocally shouldn't be
much harder than [sk]. But pre-aspiration is certainly something much
easier if one learns it in childhood.

Óskar