PhDniX's blog

hmm prenasalisation or syllabic m?

Tibeto-Burman (and also Sino-Tibetan) are language that lean rather strongly on their prefixes. There exist some suffixes (most  commonly -s) but most things are encoded in the prefix.

For example the prefix *s- is a typical and well attested prefix that makes an intransitive verb transitive. For example in Tibetan:
'-grub pa 'to be made ready, to be finished, accomplished' besides s-grub pa 'to complete, to finish'. These are the famous a-chung / s- transitive-intransitive pairs, it is an extremely productive pattern well attested, also in other languages (though often less clearly since a-chung and s- have a tendency to merge).

An s- is also commonly found in body parts/animal names for example s-tag 'tiger' and s-ñiṅ 'heart'. It has been proposed, and I happen to find this convincing, that this prefix is a reduction of the word *sya 'animal, body, flesh' as found in Tibetan śa 'flesh, meat'.

It is lovely when a prefix seems to work so beautifully. It is less lovely when it doesn't work so well. Especially the *m-prefix of TB is very difficult. Matisoff in his Handbook of Proto-Tibeto-Burman {1} wants to reconstruct one prefix *m-  that could both function as prenasalisation and as a sesquisyllabic *mǝ- this would be great if languages really displayed some free variation between these suffixes. But as it happens it seems that, at least in Tibetan that prenasalisation (written with a-chung < ' > from here on written as <N> for ease of writing) and prefix m- never alternate with each other.

[EDIT] When I started writing this article, I had not yet written the previous post (Back to the Future eat your heart out!). At the time I ran into some problems which motivated me to write the previous post. At the time I was under the impression that N- and m- never showed any free variation. Some words would have N and some would have m-, but while searching through Jäschke's dictionary for good examples for my previous post, I actually did find several words that show variation between N- and m-.

Several examples follow:

• Nthug-pa / mthug-pa 'thick'
• Nthud-pa / mthud-pa 'make longer'
• mthol-ba / Nthol-ba 'to confess'
• mkhar-ba / Nkhar-ba 'staff'
  

And there must be many more. Far from all words display this free variation though, so it must be fossilized. I wonder if there is some kind of conditioning to be found, but I doubt it. It'd be nice if we could find some system in metrical texts. But all metrical texts consider both N- and m- nonsyllabic.

Nevertheless, the original free-variation seems undeniable now, though it might be worthwhile to have a look if there's certain consonants that do not show this optional variation. As you can see, I have only been able to find it with aspirates stops so far, which doesn't mean that it doesn't occur before others, but Jäschke isn't very systematic in showing these variations, so I would have to look over every single entry in the dictionary, something I'm not particularly looking forward to.

{1} There used to be a PDF version of this on STEDT it seems like they removed it. If anyone can find it back on that website in a legal manner, it would be great, it is a nice book.

[EDIT: Oops, I forgot to make this blog public, so it's a bit late. With 'previous post' I refer to Thibetan Aspirates]

More tal-k on stem extensions.

Some time ago, I wrote a blog post on a possible 'causative' Indo-European stem extension *p. I'm currently quite interested in the function of these stem-extensions and the phonetic alternations they seem to have. Soon I'll be writing more about 'gradation' that seems to occur in some of these extensions, but for now I would just like to point out a new group of words that I found.

In English, you have the words to tell, to hear and to steal.  Besides these words, you have a group of verbs based of these roots with a derived meaning, namely: to talk, to hark and to stalk. Semantically it is difficult to say anything useful, such a general term like 'intensive' could be used, but this is so wide that it doesn't help us much. What is consistent though, is the formation.

The formation is simply formed by taking the root of the original verb, putting it in the Germanic *a-grade (or PIE *o-grade) and adding the Germanic stem-extension *-k (PIE *-g or *-ǵ).

What is perplexing about this formation is that it seems almost completely unique to English. Frisian  has talken 'to talk', but besides that I cannot think of any other formations like this in West-Germanic, or Germanic in general. It is important to know that dutch tolk 'translator' has nothing at all to do with this word, it, in fact, comes from Old Russian tolk 'id.'.

I do not understand why a formation that looks so incredibly Indo-European can only occur in English. There is no proper internal motivation in English for such a formation to arise, which makes one think automatically of an older historical source, but with no comparative evidence, this becomes very difficult.

Can any of the people that read this think of any other languages (Germanic or not) that have *√e ~ *√o-g/ǵ alternation? I will look through LIV and see if it has any nice examples.

[EDIT]

One convincing example is *ieu- 'to bind' besides *ieu-g- 'to yoke, to bind'. The first one is found in Vedic Sanskrit yuváti 'he binds'. The second word is an extremely well attested root which gave rise to the PIE word for 'yoke' *iu-g-om.

Another example, but already more difficult is *leu-g- 'to make loose, to break' besides *leuH- 'to make lose, to cut'. *leu-g- is only found in Sanskrit rujati. The absence of the Laryngeal can be explained by Lubotsky's law: *H disappears before voiced non-aspirated consonant followed by another consonant. One has to assume that there were forms where a consonant directly followed the root, and that this formation then got extended to the rest of the paradigm. This is definitely possible, as we find a root aorist in Sanskrit.

*leuH- is the basis for the textbook example of a thematic verb when learning Greek (at least in Dutch text books): λύω 'to make loose'. So, by the evidence we have we should probably interpret this as a pair *leuH-g- besides *leuH-