Thoughts of Protolanguage Models

Iwas told once, that the laryngeal model of the Protoindoeuropean phonetics was absurd: a single vowel, three laryngeal consonants. I would not say this is true. First, it simplifies the model – a single phonetic stems pattern instead of two, a single ablaut series instead of six, sonants are sonants, vowels are vowels and not both ways. It is no more absurd than the material point or ideal gas in the physics.

The neogrammarian consonant model is far more untrue. It happens, that on Earth there is no language with voiceless, voiced plain and voiced aspirated phoneme in a set of stops. Additionally, the plain phoneme *b is very rare, and is questionable in all reconstructed Protoindoeuropean stems: this is just the contrary to the fact that the b consonant happens to be one of the most widespread phonemes all over the Earth.

For this reason people hesitate using general definitions and serious scientists use terms like neogrammarian model, laryngeal model, glottal model – and the model (in physics and especially in control theory) is a thing that we create ourselves and consciously simplify in order more easily to study the real object. To what extent the model reflects the actual object, in the case of reconstructed languages we do not know and will never know. And if we read N. S. Trubetzkoy’s article about the Indoeuropean problem , we would start doubting if all theories of a Protoindoeuroean language are not a kind of a big self-delusion. Most theories today are built so that they do not contradict neither the hypothesis a divergent development, nor the one for convergent development (e.g. The Big Soviet Encyclopaedia goes into the origin of the IE languages the following way (translation mine): The actual interpretation of the similarities of the Indoeuropean languages may be concluded in either postulation of a kind of a common source of the known Indoeuropean languages (Indoeuropean protolanguage, basis language, a variety of ancient Indoeuropean dialects) or acceptance of a union of languages which being initially different developed a set of common characteristics. The two possibilities do not contradict each other but belong to different chronological perspectives).

The neogrammarian model builds a theory of a language with no counterparts on Earth – a language with plain voiceless, plain voiced and aspirated voiced phonemes in a set of stops does not exist. Indian languages do have voiced aspirated but they also have voiceless aspirated. Additionally, as Trubetzkoy notes, three sets of stops have been preserved only in languages which are surrounded by non-Indoeuropean languages. In all other cases the sets have been reduced to two. The glottal theory stipulates aspirated stops only as allophones of non-aspirated ones.

According to the laryngeal theory the vowel is not one, but three – *a, *i, *u (а is that vowel that lately gave e/o); *r, *l, *m, *n between consonants form a syllable, i.e. they also have vowel values (in this case they are noted with a circle below: *, *, *, *). Languages with three vowels and three laryngeals not only can exist, but do exist. For instance, classical Arabic: three laryngeals – ain, alef and h, and three vowels – a, i, u. One would say: there are long a, i, u. This is not the case. Long a is short а + alef, long i is short i + y, long u is short u + w. Vowels i and u between vowels become sonants – in both Semitic and Indoeuropean languages. Today’s Arabic dialects have lots more vowels but they are reduced to allophones of these three. The Nostratic model, which unites in a single superfamily the Indoeuropean, Semitic, Altaic and lots of other language families, is not a baseless hypothesis.

I have heard of a language with one single vowel.

In general, lots of assumptions both of neogrammarians, laryngealists and glottalicists are just to explain something. As are the palatovelars. There is no Indoeuropean language, extinct or extant, with clearly shown descendants of these palatovelars. But we have parallels deca — десет. Without the abstract palatovelars we have no explanation why some Slavic s correspond to Latin c, and others – to s (soror — сестра). Gamkrelidze’s theory also introduces such abstractions – voiced aspirated stops however exist in some contemporary Indoeuropean languages but there is no trace of glottalized ones. His theory just matches more accurately to the overall language appearance of the Earth. We model all the time. And change the model in order to get better explanation through it of more and more facts.

Of course, nothing contradicts a hypothesis that both neogrammarians, laryngealists and glottalicists are right: it is not impossible that the three models reflect three different states of Indoeuropean phonetics development. Howsoever, the Indoeuropean protolanguage (or protolanguages) has not appeared from nothing – investigations reach about 6000 years back in time but a human had learnt speaking longer before. Actually we cannot know which periods of Indoeuropean predecessors the development each of the three theories is applicable for. But we can suppose: the oldest should be glottal model. After it the laryngeal model comes (the Hittite language which has duly documented one laryngeal, i.e. we stop supposing and work with facts, shows some strange properties compared with the other Indoeuropean languages). Last should be neogrammarian model. It is possible that some groups have separated from the others earlier – e.g. the Germanic languages would not need Grimm’s Law if they have split before the glottalic phase reached the neogrammarian one (voicing of vowels is a normal and easily explainable phenomenon, but the contrary – devoicing, postulated by Grimm’s Law, especially between vowels, sound quite strange). This would also explain the significant percentage of Germanic roots with no counterparts in the other Indoeuropean languages.