PhDniX's blog

Nostracists and their crazy theories

For some time now, I've been converting a Nostratic etymological dictionary to be put on-line on www.ieed.nl, it's the Nostratic dictionary written by Allan R. Bomhard, and while I'm extremely sceptical towards the Nostratic hypothesis even I must now admit that there is definitely 'something' going on at least in some of the branches. I personally though see no problem in saying that this is due to loans, or just common onomatopoetic words, or words using other types of soundsymbolism.

A problem that I see in his dictionary is that he keeps on making diconsonantal roots for a proto-language whose descendants tend to have triconsonantal roots (or even more). It's great that you can match up the first two consonants of words with vaguely similar meaning in both Semitic and Indo-European, but this begs the question, why does every single Nostratic language use a different suffix? Why is an apparent schwebeablaut perfectly okay? Vowels? Who needs vowels!

Simply put, there's absolutely some really interesting stuff, but I'd say at least 700 of the 900 something entries are a very big stretch.

But now I came to an entry which absolutely baffled me. Namely:
"346. Proto-Nostratic *kʰa- 1st person pronoun stem (stative):"

First he brings up Afro-Asiatic, Arabic anaħnu, where /ħ/ is a reflex of *kʰ, and Coptic anok where the /k/ is a reflex of the *kʰ (Where did the *a run of to?). Well sure these words are incredibly similar, I'm not gonna argue with this.

But then he brings up Indo-European, and actually tries to argue that the κ in Greek perfects is a reflex of the Indo-European 'suffixed first person pronoun *kʰa-'. Are you guys still with me?

Greek has a unique formation, the so-called “first perfect”, which would be better named the “κ-perfect”. As noted by Sihler (1995:576): “Its inception must belong to prehistoric G[reek], for it is already established, within limits, in Hom[er] and in the earliest records of other dialects.” Moreover, Sihler notes (1995:576): “In Hom[er] the formation is found in some 20 roots, all ending in long vowel (from the G[reek] standpoint), and in all of them the κ-stem is virtually limited to the SINGULAR stems which actually contain a long vowel… Later the formation, by now more accurately a κα-perfect, spreads to other stems ending in a long vowel, then to stems ending in any vowel (including denominatives), and finally to stems ending in consonants, and to all persons and numbers.”
[...] The -k- forms are also found in Tocharian, as in 1st singular preterit active tākā ‘I was’, and, as in Greek and Latin, the -kis found in all persons and has given rise to secondary formations. [...]

And by this he concludes that the *-k- must be from Ind-European. Sihler points out that it's typical for Greek, and must be a prehistoric Greek development. Why then Bomhard felt the need to somehow claim that it's a Nostratic development is absolutely beyond me. The example that he brings up for Tocharian is absolutely ridiculous.

First tākā doesn't mean 'I was', but takāwa means 'I was', at least in Tocharian B. tāka is 'He was'. Why Bomhard isn't distinhuishing between Tocharian A and B is beyond me, it could be that he's referring to Tocharian A (of which I don't know much); What I do know though is that the stem tak- is a suppletive preterite to the present stem nes-. It seems like a stretch to say that there's a *k suffix put behind a stem **ta- to give it a preteritive meaning in the first person singular if that stem **ta- is completely unattested! And I don't know which language is able to have an intial mutation from **ne > **ta by replacing suffix **-s with suffix **-k, but Tocharian definitely is not one of them.

All together this etymology makes no sense whatsoever, we solely have a -κ suffix in Greek, and where Greek got that one from is indeed quite mysterious, but it's without doubt not a feature of Tocharian, or Indo-European for that matter.

Wow I hope that made any sense at all, but debunking a theory that makes so little sense is actually quite difficult, since it's not proving anything at all.

[edit]
Just because it's nice to give references, it seems to me, though the page numbering seems different, that this is the Etymological dictionary I'm talking about:
The Nostratic Macrofamily

Maybe I'm working with a newer version, I'm not sure, never got the title page with it.

Functionality of the *o-grade

While several days ago I argued that Indo-European must have had the vowel *o for a long time already, and isn't somehow derivative from the *e, there's still a problem with the *o-grade in verbs.

The perfect or hi-verbs or whatever you want to call them, had an *o-grade. In the earliest stage of Proto-Indo-Euroepan, as is seen in Hittite, there were no productive means to convert a mi-verb to a hi-verb or the other way around. In later Indo-european, as is clearly attested in many languages, it became possible to create an *e/o ablaut. To motivate this radical shift of verbal semantic is incredibly difficult to explain.

You could say *o-grade verbs were simply a type of stem with an original *o vowel in the root. And the *e-grade verbs had an original *e. If just any verb could have an *o-grade and any verb could have an *e-grade though, the radical shift of the verbal system would be incredibly difficult to explain.

So that's why it is often believed that the *o-grade verbs generally had a more 'perfective' feel, while the *e-grade verbs were more 'imperfective'. But this would already imply a very early productive ablaut in Pre-Indo-European. Not that these words could shift from *e to *o, but there was a definite link between the vowel grade and the semantics of the verbs.

Now this would be perfectly acceptable, if there weren't special endings for both these types of roots. If you have both different vowel grade in the root and radically different endings to distinguish these two semantical types of verbs, the verbs have a double marking of their semantic field. Both the vowel grade and the ending.

Typically speaking any language tries to get rid of such double marking quickly, since it's terribly superfluous. When there is a double marking, as is the case here, this is highly suspicious. You would come to expect that when these classes were created, their wasn't a double marking, but a single marking, and the second marking is secondary due to phonetical reasons.

It's hard to come up with any plausible phonetic reasoning why  an ending *-m/*-mi (earlier *-mə) would change into *-h₂e (maybe earlier phonetically /ʕə/). So it should be the other way around, the endings influenced the vowel grade.

Glen Gordon proposes quite a nice theory in his fantastic PDF Diachrony of Pre-Indo-European . So let us just assume that this problem is 'solved'. The assumption that the *o-grade in verbs brings up a lot of questions. I'll try to summarize them before going into them.

First, there's the question 'Why is there no *e and *o vowel distinction in the verbs, while it's clearly there in the nouns?'. Second 'Why doesn't the Middle voice, whose endings are very similar to the Perfect endings give an *o-grade?' and last but not least 'Why does a suffix like *-eie which has little to do with the Perfect semantically or morpohologically give an *o-grade in the root?'.

Those are three enormous questions, and that will probably show from the length that this post is going to be. But hold on tight, it's going to be a fun trip!

Why is there no *e and *o vowel distinction in the verbs, while it's clearly there in the nouns?

My answer for this question is actually quite simple, I have no idea. More elegantly put, I do not think it is possible for a language to technically have one distinctive vowel in the verb, while it has two distinctive vowels in the noun. If this is, in someway incredibly common in a good number of languages, please point it out to me, because I'm missing it completely.

All roots seem to have been formed from the 'dictionary form' as we like to write it these days, with an *e-grade in the root. Where's my recently established phoneme *o?!

A possible explanation could be, that there were in fact *o's in verbs, and when the vowel harmony showed up, tons of *e-grades were being turned into *o's, while *o grades stayed *o's.

Then all the verbs with Perfect endings would have an *o-grade, while the verbs with Present endings would have mixed *e- and *o-grades. Was this mixed analogically levelled, because the Perfect had all *o-grades? Why weren't the Presents levelled to *o rather than *e?

Any suggestions are welcomed. Don't feel afraid to feel stupid, I'm feeling it too, it's perfectly natural when talking about such a difficult subject.

Why doesn't the Middle voice, whose endings are very similar to the Perfect endings give an *o-grade?

Beekes reconstructs the perfect and middle endings as follows:

Perfect	Middle
*-h₂e	*-h₂
*-th₂e	*-th₂o
*-e	*-o
*-mé	*-medʰh₂
*-(h₁)é	*-dʰue
*-ḗr	*-ro

 

Personally I (for now) reconstruct these forms:

Perfect	Middle
*-h₂ə	*-h₂ər
*-th₂ə	*-th₂ər
*-ə	*-ər
*-ə́rs/ər	*-r

Which I feel is more in line with the Anatolian and Tocharian evidence. Quite clearly the Middle was in some way derrived from the Perfect, notably by spreaking the 3rd plural throughout the paradigm as a middle marker(1st and 2nd person plural remain incredibly problematic, and I'm open to a nice theory to explain those forms).

So if the endings come from the perfect, why didn't these endings influence the vowel grade in the root? The only viable explanation is, that the middle endings were formed only after the arisal of the *o grade after these endings had become phonemisized, rather than a case of vowel harmony. This explanation works quite well.

Why does a suffix like *-eie which has little to do with the Perfect semantically or morpohologically give an *o-grade in the root?

Here we're talking about a much later stage of Indo-European. One where Anatolian had long split of, nevertheless there's a funny *o-grade popping up in Presents stems, where the language first seems to've excluded them completely from these paradigms.

*-eie when functioning as a causative always has an *o-grade in the root. *men- --> *mon-eie. Looking at the shape of the suffix it looks like the thematic vowel *e + thematic root suffix *-ie. What I don't understand though, is why the *o-grade arose. Semantically a causative is nothing like a perfect, and phonetically *-eie is nothing at all like the common verb endings we find in a perfect. In post-anatolian languages this formation is very well attested though. In Sanskrit it was still 100% productive, and even in English we still find some forms which go back to this formation.
For example lie/lay sit/set etc.

I'd love to play around with schwa's and having them colour according to their environments, but this is a clear post-anatolian formation, a stage in which the schwa's were gotten rid of nicely. I believe some new ones may have arisen when reduplication was introduced, but these seem to've only changed into *i and *e, and there's actually no reason for a schwa to be in the root, at the point this suffix was introduced accent was no longer needed to have full grades.

Suggestions to explain this weird *o popping up are very welcome.

So there you have it, a lengthy post about the problems with the function of the verbal *o grade. Almost no answers, and a lot of questions.

Never thought it would happen...

Never thought it would happen, but I've come to disagree with some of the theories of Beekes. He is still considered one of the most important Indo-Europeanists over here in Leiden, but not everything he says should be taken for granted (which some people do).

His idea of the Indo-European vowel system seems wrong to me. He identifies several stages of vowels and their distribution in Indo-European.

First there's one vowel *e which appears both accented and unaccented.
The stage after that the unaccented *e develops into both *e and *o.
In the last stage the unaccented *o spreads to the accented syllables.

I dislike this theory from the start, since I find a 'One-vowel hypothesis' rather unrealistic. But this dislike should only be an indication not a dismissal, so I shall continue to elaborate.

I believe that in the first stage there were vowels *e and *o (or probably more accurately *e and *a), while eventuall neutralised when unaccented into *e, which then subsequently split into *e and *o again.

The spread of *o from unaccented to accented syllables most definitely did happen, it helps to explain alternations of suffixes like the nom sg. *-tṓr and *-tḗr. Probably the orignal paradigm alternated between accented *é and unaccented *ó, and it was neutralised both ways (some paradigms now only have *o, and some only have *e). But this analogy theory simply isn't enough to account for all accented *o's.

The most important of these unexplicable *o's are the one's we find in the perfect. There's no way to get unaccented *o's in the root in verbal paradigms. So the analogy theory would never work. Also in root nouns there's quite a few *o's flying around that don't 'just' explain themselves. Take the word *pṓds that actually has an unaccented *e in the root rather than an *o when we look at the genetive *pedós.  Still though, the amount of true accented *o's are quite sparse, and it does seem that Indo-European was out to get rid of them for a while.

But even that is not certain, maybe the unequal distribution of *e's and *o's is because of a prior collapse of the vowel system where more vowels merged to *e rather than to *o.

The bottom line is, I believe there is an abundance of proof that in fact the *o didn't exclusively come from the unaccented syllables, but is in fact native to Indo-European

nyanko's and neko's

I was having a discussion with FTPlus about how funny it was that all over the world, there's about three types of Cat words. you have 'CaT', 'PuSS', 'MIaW' (and NIaW) as patterns, the consonants can differ somewhat but that's about all the cats there are. And then I said: And then there's the Japanese with... neko.

Neko doesn't seem fall in any of these categories whe you look at it at first. There's an informal word for 'kitten' though, that does fall in these catagories. Namely 'nyanko'. Which would translate to 'meow-child'.

Then it dawned on me, it is hypthesizes that the modern Japanese /e/ in Proto-Japanese had two origins bot from -aiy and from -ya. While in Old Japanese this distinction was lost after most consonants, still some retained it, namely after /p/, /b/, /k/, /g/ and at a very early stage also after /m/. This is the moment we wish that a neko was called a meko, because what I'm going to suggest later on would be proven without a chance of discussion.

Some nice indications of the -a-i in words like fune 'ship'  which in compounds becomes funa like in funatabi 'trip by boat'. Japanologists hypothesise that this shows that there was a nominal suffix -i which was lost when compounding the word.

Proponants of Japanese being an Altaic language will tell you this is a 'clear' case of vowel harmony though (I'm not convinced).

Another reason that the origin of these e's  are hypothesised as diphthongs is that in Old-Japanese teksts, we find some contractions to form the right corresponding e's.

So now to get back to our neko. In Proto-Japanese this neko could have been both a nayko or a nyako. That's right! There we have our nyanko from modern Japanaese (save the /n/ of course). Even in modern day japanese the onomatopoetic word for meow meow is nyaa, nyaanyaa or nyannyan. Isn't that odd? It's almost as if, in the lexical 'meow-child' the contraction took place, but in the things that were still clearly onomatopoetic, the change never took place. This actually isn't all that odd, if onomatopoetic words would undergo sound changes, they'd quickly stop being onomatopoetic. Can you imagine Greek sheep saying [viːviː] rather than [bɛːbɛː] all of a sudden?

But what's really interesting is that the near identical formation of the word cat was created with Onomatopoia twice. I just thought I'd point it out.

There's one problem though, as I said, the two e variants were not distinguishable in Old-Japanese, so neko could in fact be Proto-Japanese *nayko, but let's be honest, it does sound likely that it was nyako right?

Just to close off an appropriate Haiku by Kobayashi Issa:
鳴く猫に赤ん目をして手まり哉

 

Obrigado gozaimasu?

A very persistent, and blatantly wrong popular believe seems to exist that the Japanese word Arigatou 'thank you' is actually a loan from Portuguese 'Obrigado', yes these two look similar, and sound similar, but by no means is the one derived from the other, it's pure simple coincidence.

I ran into it again when I googled 'History of the Japanese Language' where I ran into this page:
http://www.todaytranslations.com/language-history/japanese-language-history/

Since the mid 18th century the Japanese have adopted a huge amount of "gairaigo": foreign words mainly from English. These include "teburu" (table), "biru" (beer), "gurasu" (glass), "aisu" (ice), "takushi" (taxi) and "hoteru" (hotel).

There are also a few words from Portuguese, Dutch and Spanish, such as "pan" (bread) and "arigato" (thank you), from the Portuguese "pão" and "obrigado". Such words arrived in Japan mainly during the 16th and 17th centuries, when missionaries and merchants started to visit the country.

Sure teiburu is from table, aisu is from ice, takushi is from taxi and hoteru is from hotel. Assuming that biiru and gurasu are from English though, is a bit short sighted. It could have just as well been from Dutch.

Glas is the Dutch word for glass, and while these days it is pronounced [χlɑs] (at least in my dialect), when the Dutch first had contact with the Japanese it was most likely pronounced [glɑs] or [ɣlɑs]. But this one is still open for discussion, and probably actually is english (Did we have glasses in the 18th century?).

Biiru though, is unmistakably Dutch, just like kouhii 'coffee' is.

The Dutch word for biiru is bier. What makes it so convincing that it Dutch then? Because it has Chinese characters:

麦酒

麦 ' Barley'

酒 'Sake'

Back in the days of the Dutch contact with Japan, it wasn't common for the katakana to be used for loanwords, and it was common for every noun to have characters, thus they gave beer just that.

Same goes for kouhii

珈琲 Which in meaning of the chracters has nothing to do with it, but seems to vaguely try to imitate the word for coffee (koffie in Dutch) as 'kahai'  in it's reading. How it ended up with a more accurate pronunciation than its characters is a mistery to me.

kouhii das in fact reveal an interesting fact about the pronunciation of the 'h'  at the time, that it must have been the hypothesized [ɸ] and by that emulate the Dutch word quite closely.

One who wants to mention that the [h] was simply used to replace the unfamiliar [f], this could be true if the word for Holland (alternate word for the Netherlands) was horanda in Japanese. But actually it is oranda, showing that the Japanese were unfamiliar to the [h] sound at the time.

Here I have greatly diverged from the original topic, but it all made some remote sense, honest!

A good way to check whether a word is a loanword or not, is to see if it has any morphological complexity within the language. For example alcohol is just one word in English, but if we see it as Arabic we clearly see the definite article al+kohl, which should be its origin.

Same goes for arigatou, or more complete arigatou gozaimasu

有り難う御座います

The fact that it actually has characters belonging to it, with conjugated endings should already have your alarms ringing.

In the spelling at the time arigatou was actually written arigatahu, a rather archaic (but still existent) conjugation of arigatai　有り難い (arigatahi at the time), a compound adjective consisting of 有り 'having'  and　難い　' difficult'

Arigatou gozaimasu  therefore translates to 'it is difficult for me to have'. I'm guessing the implication is ' it is difficult for me to have/accept your (whatever the person is giving you)', which is an appropriately humble way of saying thank you in Japanese.

So arigatou actually makes morphological sense in Japanese, making it incredibly unlikely that it is a loaned word.

So maybe the Portuguese loaned arigatou into their languages as obrigado then? No, no such luck either. obrigado is from the same source as the English word 'obliged', and with my limited knowledge of Latin I'd say the Latin for is obligatus. Note the incredibly typical but irregular l>r shift! Obligado thus means 'much obliged' or something along those lines (I don't actually speak Portuguese).

So now it's clear that it's just an odd twist of languages that two words turned out to be very similar.