16 Oct 2011

Egyptian vowel reconstruction and other gripes

Occam's Razor is a valuable tool to the student and scholar. It forces us to think hard about the assumptions we hold on to and whether they are absolutely justified or whether there's room for doubt. Linguistics seems to be one of those studies where this methodical principle is still not respected to the level that it should be and, as a result, there are many ancient languages being reconstructed with too much artistic flair to properly reflect the data.

Diversity of plausible theories or diversity of empty opinion?

I've been very busy collecting data on Ancient Egyptian after growing dissatisfied with the lack of profound discussion or clarity on its vocalism. Egyptologists constantly write words with only their consonantal values to reflect how the Egyptians themselves wrote these words. This is how it's always been. However I find that it often does more to obstruct and obscure the proper reading of these texts than aid us. It doesn't take a rocket scientist to figure out that Egyptians themselves wouldn't have thought of words purely in terms of consonants. Some of the clever word puns exhibited in Egyptian texts require our knowledge of the vocalism too in order to grok its fullest meaning and pattern. After centuries of Egyptomania, why is there no clear consensus on the Ancient Egyptian vowel system? What's the hold up? Are we interested in Egyptian or not?

To illustrate the point, let's take the word for 'cat' which may be represented consonantally as mỉw. Here is the mountain of possible reconstructions for the utterly confused outsider to select from:
  • Albright *mắȝĕʔ
  • Callender *máȝejvw
  • Garnot *mṓȝei̯
  • Smieszek *må̆ȝjᵉw
  • Vergote *māȝuy
Obviously they can't all be correct. Notice that a lot of these scholars seem to delight in masking their representation of the language with a bunch of unnecessary diacritics. (I've ranted against this before many times.) To aid in our investigation, we see that the plural form of the word is reflected in the Greek name Πανομιευς which represents the Egyptian phrase *pȝ-(n)-nȝ-mȝj.w 'He of the cats'. Of course, Egyptian shares with Arabic the use of broken plurals and so the plural vocalism is not necessarily the vocalism of the singular. In order to keep my sanity, I find myself forced to develop my own testable opinions on the matter with a conciliatory reconstruction of *māya /'mɑ(ː)jə/ for the period around 1500 BCE and it seems sufficient to account for later Coptic form moui agreed upon by Sahidic, Bohairic, Akhmimic and Fayyumic dialects.

Back to Occam's Razor, one thing that frustrates me when I see this kind of diversity of opinion and no consensus is that the reasons why these individual scholars have arrived at their differing ideas appears to be grounded less in linguistic science and more in artistic whim. To me, phonotactic analysis is unavoidable in this task. We need to be absolutely conscious about how syllables are put together in our language of interest, not just the individual phonemes. We need to start with the most universally commonplace rules and meet each contradiction with adaptation from a simple and commonmost state to a more complex and exotic one, not vice versa. Sadly linguists often don't demonstrate this rigour but it's vital in creating a coherent theory that obeys the KISS principle (ie. Keep It Simple Stupid). So, to me, the diversity of opinion in the example of 'cat' is not so much the result of coherent theories clashing for competition, but a bunch of lazy theories made by scholars ignoring Occam's Razor in their idiosyncratic ways.

And how to handle those unstressed syllables?

Focusing just on how different scholars treat unstressed syllables Egyptian, there doesn't appear to be a justification for how one decides which vowel it is, aside from appealing to outside branches of Afro-Asiatic like Semitic. Callender for example reconstructs *pAsīḏaw for 'nine' with wildcard symbol A whereas Loprieno chooses *pisī́ɟvw (nb. Loprieno's i = Callender's A) with yet another wildcard symbol v in the final unaccented syllable. In this case, Proto-Semitic having only *tišˁu has no equivalent cognate to enlighten our efforts on the matter.

Neither the Babylonian inscription EA 368 which records 
pi-ši-iṭ nor the later Sahidic Coptic form psis gives us much evidence of what the first vowel was because an unstressed vowel is often less audible than a stressed one. Coptic has already dropped the vowel while, for all we know, the Babylonians interpreted a garden-variety schwa as a lax -i-. I still search for precise evidence that justifies this need for more than one vowel quality in unstressed positions. Until I do, I reconstruct *pasiḏa /pə'siɟə/ where unstressed *a is nothing other than the generic schwa /ə/ which we would find in all unstressed positions. Notice too that I choose to avoid unnecessary diacritics like the plague, as I believe we all should if we strive to be good little linguists.

Naturally if there is indeed unambiguous evidence of other possible vowel qualities in unaccented syllables, I'd love to hear about it. But until I do, Occam's Razor must be my guide.


  1. I think people avoid using common symbols such as a for an unknown vowel because it sketches a 'warped' image of how certain we are of its value.

    Of course, you expect people to actually read your explanation on the transcription, and see that by a you mean ə or maybe another uncertain reduced vowel (unlikely), but can we really assume that someone who does, for example more, deeper Proto-Afro-Asiatic reconstruction reads a whole book on ancient egyptian, and doesn't just copy a and says "Gorden (2011) reconstruct /a/ in this word that undoubtedly goes back to Proto-Afro-Asiatic".

    Of course, we would hope scholars are not this sloppy.

    But the fact remains that, a scholar will be forced to look up what you mean by your transcription a before they can do anything with it. This would also be true had you used v. But in the case of v they would be more quickly inclined to look it up.

    Not exactly professing an optimistic view of the professionalism of scholars, but I can understand why people would be using more obscure symbols than plain a. This of course doesn't mean that a symbol like å̆ or isn't silly unless you have a very good reason to write it as such.

  2. Phoenix: "But the fact remains that, a scholar will be forced to look up what you mean by your transcription a before they can do anything with it."

    Um, the very purpose of any half-decent scholar is precisely to "look things up", regardless of the model chosen. That's just not the issue.

    The issue is with gutless, non-committal theory that unnecessarily abuses meaningless wildcards like "v" (= any vowel) to avoid commitment to a premise that may be tested against the facts. Simply put, by using "v", the theorist avoids the fundamental issue of Egyptian phonotactics and leaves it to the next scholar to devise those coherent rules.

    So why not me? Why not you? Let's debate on what constitutes the set of possible unstressed Egyptian vowels.

    In my model, I note that the nominative and accusative case endings preserved in Proto-Semitic, *-u and *-a respectively, have merged already by Old Egyptian. This merger can be explained by a pre-existing system of three unstressed vowels (as per Proto-Semitic) collapsing to generic schwa.

  3. Sorry for barging in here three months late, but why not write schwa if you mean schwa? Why use "a" if it's bound to be misunderstood?

  4. "Why use 'a' if it's bound to be misunderstood?"

    This is as inane as saying that writing "sofa" in English causes mass confusion because "a" happens to be often pronounced /ə/.

    In Egyptian, from all evidence that I can see, there are only three phonemic vowels in unstressed position: *a, *i, and *u. So it hardly matters in a system like this whether you pronounce it /a/ or /ə/. It only matters that your pronunciation of *a is [+low], [-round] and distinct from *i and *u.