Gothic for Goths: Some Thoughts on Pronunciation

You know, no matter how many times i say it, people keep asking me why i pronounce things a certain way in gothic, or "Shouldn't it be ____ instead of ____?" or "How come you say ____ when that other guy on YouTube pronounces it like ____?"

Of particular contention among all gothic scholars are the diphthongs (and/or digraphs) ai and au.  So let me put this to rest here, once and for all.

First of all, no one is "right" or "wrong" about this.  Nevermind, that's not true.  There are a lot of "wrongs," but it's impossible to know if one is right, because gothic just plain doesn't exist anymore.  So let me present the theories, the evidence, and my own personal take on the situation.

Ideas about gothic pronunciation really fall into three basic camps, one of which is just plain wrong.  I fall somewhere between the second and the third, if we think of these three camps on a sliding scale.

Theory #1:  ai and au are pronounced as [ai] and [au], respectively.

This theory is the easiest to explain, and the easiest to discredit.  It's just plain not the case.  Look at any greek word with an epsilon that is borrowed into Wulfilas' bible and you'll find it transliterated as "ai," and epsilon was just plain old never pronounced as [ai], but always as [ɛ].  And this shouldn't be construed as the mispronunciation of a non-native greek speaker trying to render the words: It is thought that Wulfilas' mother was possibly greek, and that he spoke it as a child.

Theory #2: ai and au are prounounced [ɛ] and [ɔ] in some cases, and sometimes [ai] and [au], respectively.

The trick here is to decide what constitutes "some cases," and what doesn't.  My theory of gothic pronunciation falls pretty firmly into this category, as do most scholars, but they are far from uniform on opinions about which are which.

Some of the nicest gothic lessons out there are those by Slocum & Krause, and they fall squarely into this category.  They fall into what we might call "Theory #2a," which would be that ai and au are usually pronounced [ai] and [au] except in very specific circumstances, like in borrowed greek words or when occurring before r, h, or hw.

My own theory i would classify as "Theory #2b," which I'll spell out in some more detail below, but it may be summarized to say that it is Slocum & Krause's pronunciation with mora loss in unstressed diphthongs, particularly in noun and verb endings.  Pronouns are a different story, and i explain more about them below.

Theory #3: ai and au are pronounced [ɛ] and [ɔ], respectively, in all cases.

There is some interesting evidence to back this up, at least in the case of au, particularly in the use of certain transliterations in gothic like the name of the Apostle Paul, where Wulfilas renders Παυλος as Pawlus (instead of the expected Páulus), or the latin "cautio," transliterated as "kawtsjō."

Voyles falls strictly into this camp, and extends his germanic ai/au-to-ē/ō rule to apply to all instances of ai and au in east germanic.  In particular, the germanic rule:
Monophthongization of unstressed ai, au
Stage 1: ai,au → ɛ̄,ɔ̄ when in a word-final unstressed syllable immediately preceded by an unstressed syllable in east germanic only.
Stage 2: ai,au → ɛ̄,ɔ̄ when unstressed, then → ē,ō in northwest germanic only.
Later he expands this rule in gothic.
Monophthongization of ai, au
ai,au when stressed → ɛ̄,ɔ̄ when word-final or before a non-vocalic consonant.
So there you have a run-down of the classic theories.  Now allow me, if you will, to spell out my own, which as i mentioned earlier is "Theory #2b."

I think that Voyles was pretty darn close, but i would simply cut out his later Gothic Monophthongization of ai,au rule and attribute his evidence for it to orthographical conventions rather than phonetics.  I also have little doubt that this later rule actually did take place, but i don't believe that it had completed at Wulfilas' time.

So in a nutshell, i would posit that:
  • ai,au = [ai],[au] 
    • whenever stressed, with a few exceptions, below.
  • ai,au = [ɛ],[ɔ] 
    • always when unstressed.
    • always before r.
    • always before l. (This would explain away the rendering of Pawlus.)
    • before h(w) except where stressed and from germanic [au], instead of from germanic [u] via the gothic expansion of first umlaut.
      • For example, PGmc hauhaz → hauhs [hauxs], but PGmc luhō- → lauh- [lɔx].
One aspect which stumped me for a while are monosyllabic words with standard endings which ought in most cases to be pronounced [ɛ] and [ɔ], such as bai, twai, etc.  After much consideration, i've come up with the following rather ad hoc rule which is still open to some debate, and is largely based on what seems the most natural to say:
  • ai,au = [ai],[au]
    • when final, as in bai, þai, twai.
  • ai,au = [ɛ],[ɔ] 
    • when followed by a consonant, as in þaim, twaim.
A problem word:

raida:  I've been saying [rɛða], based partially on the much later spelling of the letter R as rēda (from the 9th century Codex Vindobonensis); it is uncertain if the proto-germanic was raidō or rēdō.  It is very likely that the later spelling is reflective of Voyles' later rule.  In my first lesson, i pronounced it [raiða], which stands an equal chance of being correct as far as i can tell, but i've grown to like the sound of [rɛða] better.