Richard d'Alquen: Gothic ai and au (summary)

Wed Apr 5 19:58:20 UTC 2006

Published in 1974, this books offers an interesting solution to the 
problem of how to interpret the digraphs <ai> and <au> in the 
spelling of the Gothic Bible. It's generally accepted that they 
stood for [E] and [O] sespectively in some Gothic words, in almost 
all of which the digraphs are followed by <h> or <r> in a presumed 
stressed syllable. They also have this value in Gothic 
transcriptions of Greek biblical names, and are the most usual 
transcriptions for Greek epsilon and omicron.

However they also appear in certain words, standing for Proto-
Germanic [ai] and [au].  Debate has raged over whether these were 
diphthongs still in Wulfila's Gothic, and if so, why he would have 
adopted such a confusing spelling system.  Other points of 
uncertainty concern the etymology of [ai] and [au] in certain 
inflections, and the value of [ai] and [au] in words 
like 'saian', 'bauan', where they may represent Proto-Germanic [e:], 
on the one hand, and [o:] or [u:] on the other.

D'Alquen's theory is that the language found in the surviving Gothic 
manuscripts represents an Ostrogothic reworking of the original 
Visigothic bible translation.  From loanwords into Romance 
languages, and Gothic names recorded in Latin and Romance sources, 
he argues that the PG diphthongs [ai] and [au] were preserved in the 
Visigothic kingdom of Provence and Spain, but monophthongised to 
[E:] and [O:] in Ostrogothic Italy by the end of the 5th century, 
that is, by the time of our manuscripts.

He suggests that in Wulfila's language, the following short vowels 
were phonemic: /i/, /u/, /a/. In the original translation, Greek 
epsilon would have been transcribed with , omicron with . 
Traces of this practice can be found in a few words which had become 
naturalised in Gothic: aggilus, Makidonja, paintekuste > German 
Pfingsten, and a handful of others. (These words have traditionally 
been thought of a pre-Wulfilan loans.) The high vowels had allophic 
variants /E/ and /O/, before /h/ and /r/, which eventually became 
phonemic due to factors such as the assimilation of loss of /h/, 
seen e.g. in 'drausnos' for 'drauhsnos'.

Gothic for Greek epsilon is somewhat rarer than Gothic for 
Greek omicron. This d'Alquen explains with the following sequence 
of changes in Ostrogothic, completed by the time of the hypothetical 
bible rewrite:

1. [ai] > [E:]
2. [E] > /E/
3. all unstressed vowels shortened.  Since there is no phoneme /O/ 
yet, unstressd [au] falls together with [u].
4. [au] > [O:]

With this staggered theory of monophthongisation, he also explains 
the confusion of and <au> in the endings of u-stem nouns. In 
his opinion:

Visigothic (including Wulfilan Gothic)

N sunus
V sunu, sunau
A sunu
G sunaus
D sunau

Ostrogothic (the language of the manuscripts)

N sunus
V sunu
A sunu
G sunus
D sunu

By contrast, 'nasjis' [nasjis] and 'nasjais' [nasjEs] aren't 
confused, since as the time of shortening, there were already 
distinct phonemes /E/ and /i/. Similarly, he explains 'bauan' 
and 'stauida' as being from Pre-Ostrogothic [bu:an] and [stO:an], 
with another vowel shortening at the same time which affected long 
vowels before another vowel. Lacking a phonemic contrast between 
short /u/ and /O/, these would have fallen together in a single 
phoneme /u/, realised as [O] before another vowel. Vowel shortening 
would also account for the greater frequency of for <e> 
spellings in unstressed syllables. On the other hand, it is 
contradicted by the statement in the Vienna-Salzburg Codex that the 
<ai> in 'libaida' is like long <e> in the Roman alphabet (as used to 
write Old High German).

The lack of : <au> confusion in verbal endings is put down to 
conventional spellings due to the fact that there are no verbal 
forms distinguished solely by a diference between and <au>. A 
criticism of this argument might be that <ei> for <e> confusion is 
very common and not affected by any such restriction. This makes me 
think that the confusion seen in the u-stems might be due to 
analogy, and be caused originally by the anomaly of these two 
supposedly inherited forms of the vocative, rather than a phonetic 
development.

Although, by d'Alquen's own admission, the evidence for such a 
radically different spelling system in Wulfila's Gothic is slight, 
there is a lot about this theory which is is VERY CONVENIENT. It 
explains why <ai> and <au> are used for Greek epsilon and omicron, 
as opposed to simpler and more obvious symbols. It explains why 
Gothic occupies the same position as omicron in the Greek 
ordered Gothic alphabet in the Vienna-Salzburg Codex. It avoids the 
need for certain strained and exceptional etymologies to account for 
<ai> and <au> in unaccented syllables. The evidence from loans and 
names adds up to a fairly convincing argument for dialectal 
difference, although it might be debated whether the isogloss 
corresponded exactly to the ethnic difference Visigothic versus 
Ostrogothic, and when exactly the Ostrogothic monophthongisation 
occured, whether or not the whole sequence of sound changed proposed 
by d'Alquen was carried through in all Ostrogothic areas, etc. 
Useful tools would be a detailed database of the names with 
information about orgin and an investigation by someone 
knowledgeable about the practices of Latin scribes and changing 
writing traditions and local differences. East Germanic names in 
Greek sources aren't examined in detail. Another requirement for 
future research (in this area and others) would be a more accurate 
reconstruction of the Greek source text (Vorlage) for the Gothic 
Bible translation.

Not dealt with in this theory are the confusions of unstressed 
vowels in the Calendar and the Deeds, which might suggest that all 
vowel distinctions were breaking down in unstressed syllables.

In conclusion, I like the central premise, but I'm still thinking 
about the vowel-shortening aspect, and whether this complex sequence 
of sound changes is the ideal explanation.  I think the evidence--
which I haven't discussed here--for a continuing distinction in 
vowel length in stressed syllables (as opposed to differences only 
of height or tenseness) is very strong.  I'll maybe write something 
about that some time.

Does anyone have any thoughts on all this?  Can anyone spot flaws in 
the reasoning, or think of neater alternative explanations?

You are a member of the Gothic-L list. To unsubscribe, send a blank email to <gothic-l-unsubscribe at egroups.com>. 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/gothic-l/

<*> To unsubscribe from this group, send an email to:
    gothic-l-unsubscribe at yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/