Benct Philip Jonsson <[log in to unmask]> writes:
> Question: how comes EG becomes __ and not _g_?
> The latter would be tremendously cool for obvious
> reasons!

Yeah.  (But I still have 'ert' and 'er' for the 2nd and 3rd person
singular of 'to be', which creates similar fun. :-))

Anyway -- 'ego:' would regularly come out as 'eu' (disyllabic,
i.e. [E:Y]), which is a bit strange.  The reason is probably missing
rules in the sound change history file since I did not encounter a
Germanic word with a similar structure.  Maybe it could have become
'jo:' by a hiatus preventing accent shift rule.

Anyway -- the 'g' would probably not survive, it is often shifted to
/h/ and later lost in most cases (sometimes triggering further sound
changes before vanishing, though).  Compare _vega_ and past tense
_vo:_.  Final -g in Icelandic is usually from final -k in Old Norse
(e.g. _ek_).  And -k would be hard to produce from 'ego:'.

(Ok, ok, some -g do survive, e.g. in verb forms (_flaug_ 'flew'), so I
might find a way...)

At this point, I modified history and decided that a pronoun does not
necessarily adhere strictly to general sound shift rules since its
usage frequency is different.  So I shortened the final 'o' of the
Latin source and 'e:' is the result, which I found quite cool,
too. :-)

> FWIW I think it is a mistake to say that _au_
> is *phonemically* /9y/, although there can be
> no doubt that it is *phonetically* [9y], since
> there is no /y/ phoneme in Icelandic.

But the whole diphthong is the phoneme -- it does not need to consist
of parts that are phonemes.

> native speakers make: when they want to spell this
> diphthong "as it is pronounced" they invariably
> write it __ -- i.e. the rounding of the glide
> [y_^] is perceived as an assimilation to the
> rounded [9], the diphthong being *phonemically* /9i/.

Sure.  Just like Germans do for /OY/, which is written _oi_.  In fact
_oi_ in the German pronunciation of _Khoisan_ is essentially the same
as _eu_ is _heute_.

> <rant value="Benct's Icelandic transcription beef">
> FWIW I notate __ as /9/ since *phonetically* it is
> clearly [9], and also it corresponds to _e_ which is
> best regarded as /E/.

This is a good idea, I think I will adopt this.  (BTW, I
perceive Icelandic long /E/ as [e:E)] for the speakers I