PhDniX's blog

Why do General Linguists only speak English?

As I'm currently doing my MPhil. Linguistics I am confronted with general linguistics more often than I used to back when I was still focusing solely on historical linguistics. While to some extent I feel that general linguistics brings an insight into language, which might be lost when focusing on historical linguistics, I am often unpleasantly astounded by the lack of knowledge of some of my professors.

Anglocentrism, or Eurocentrism is what General Linguistics suffers from. So many of the linguistic models currently being made are almost purely based on the behavior of the English language, sometimes Chinese is thrown in for 'foreignness'. Not to discredit Chinese, but as a SVO language just like English, it might miss the 'foreignness'  that could be brought by more syntactically exotic languages like Berber or Tibetan.

These days the theory on Syntax seems to be that VSO order is 'marked' and those languages that have this word order are currently being overanalysed to a point that it has nothing to do with the actual language anymore, just to make a point. Sure I'll give any of the anti-VSO'ers that yes, Irish might be argued to be far more SVO than VSO, nevertheless, there was without a doubt a stage that it was clearly VSO. But history of a language is of no interest to the general linguist.

But okay, these are subjects I still, feel very little for. Even if it's just that the classes I follow purely focus on English and Chinese, while they leave me with so many questions about Japanese, Berber,  Tibetan, which due to their Anglocentrism, my teachers are unable to answer. It's infuriating. How can a teacher of linguistics have all kinds of theories on language, when their language proficiency delimits itself to something so limited?

On a less syntactic note but still a knowledge note the following happened to me a couple of months ago. I followed a class of Advanced Experimental Phonetics and my teacher claimed that languages do not have the sound [h] in coda.

Obviously I was astounded. In fact at that time I was doing research on Minangkabau, a Malay language, and anyone who has some knowledge of Malay should know that rumah is a perfectly normal word for 'house', with, yes, an [h] in coda!

And that's not all, what about Arabic, sure, it's not that common, but whatever happened to ahlan wa-sahlan an undeniably common phrase with no less than two [h] in coda!

I could go on for ages on this, for example what about Sanskrit whose word final *s shifted to h (incidentally a shift Minangkabau underwent too), o wait! Historical linguistics! It's a dead language so it doesn't exist.

How can a teacher in phonetics not be aware of such fundamental fact of phonetics? No a coda [h] isn't common, but to say that it doesn't occur is frankly, absurd.

So this is a call to all people who study General Linguistics. Please please learn at least 10 languages, even if it's just a passing knowledge of phonology and grammar, it should greatly improve understanding of how language actually work in the real world where people don't just speak English.

Another thing, I followed one class on speech acquisition of babies. This teacher claimed that at the earliest stages, babies are able to produce every sound, and in fact do. Only after they have learned the phonological system of a languages they start to gradually reduce the amount of sounds that they babble to fit the language they will learn to speak.

I don't know about you, but I have never heard a baby produce [qˁʷʼ].  Has anyone every stopped to think how absurd this idea is? I have no knowledge of Ubykh (or any caucasian language for that matter) , but I at least have enough of a passing knowledge of their rather exotic phonologies to know that if I'd ever come up with a hypothesis about  speech acquisition of babies like this, that I'd know that I'd have to at least check up with Caucasian sounds and caucasian babies too.

God, I hope this doesn't  sound too ranty, and I really hope someone comes by and discards all my, without a doubt, grossly exagerated examples.

Tibetan Numerals

So tomorrow I have a test on Classical Tibetan, which means I'm spending some time studying again. And as I was doing this, I ran into something interesting.

These are the numerals from 1 to 10:

Numerals 1-10
1	gcig
2	gnyis
3	gsum
4	bzhi
5	lnga
6	drug
7	bdun
8	brgyad
9	dgu
10	bcu

 When we then look at the numerals for 11-19 We see that the word for 10 undergoes a change. It Alternatives between bcu and bco.

Numerals 11-19
11	bcu-gcig
12	bcu-gnyis
13	bcu-gsum
14	bcu-bzhi
15	bco-lnga
16	bcu-drug
17	bcu-bdun
18	bco-brgyad
19	bcu-dgu

What do bco-lnga and bco-brgyad in common with each other that the other numerals don't share? The answer would be a vowel a in the following syllable. Maybe we're looking at the remants of some old sound law. The silly thing is though, that an ablaut between u and o hardly occurs outside of the numerals.

In the verbal system you have i/u ablaut, a/o and e/o but no u/o.

The etymological origin of o in Tibetan is quite diverse it can come from PTB *o, *aw, *aːw, *ow, *wa. The origin of u is PTB *u, *əw and *uw *aːw

As you might have noticed *aːw can yield both o and u. We even have words that show the variation. nu-bo ~ no-bo both mean 'younger brother'.

We also find some words in Tibetan that alternate between o and wa.

So is the word for ten in PTB maybe *bcaːw? And then the a in the next syllable might indeed be a hint at the old conditioning. Maybe the word for younger brother used to be nu-bo besides no-ba (bo/ba are quite interchangeable masculine noun suffixes). Obviously we'd want more examples where my hypothesised sound law would apply without having to assume analogies, but it's an interesting thought.

Matisoff has a completely different root for 10 than the one we find in Tibetan so *bcaːw would be Pre-Tibetan rather than PTB. Which is too bad, because it makes checking the theory by comparative evidence rather difficult.

But there are more situations where a vowel changes occurs. The aternate forms of 2-7 are:

Alt. Numerals 2-7
2	gnyer
3	so
4	zhe
5	nga
6	re
7	don
8	gya
9	go

The ones of itnerest are 2,3,4,7 and 9
2 and 4 display a i > e shift
3, 7 and 9 display a u > o shift. With gsum the vowel u can definitely be reconstructed, it's well attested all accross Proto-Sino-Tibetan. Nobody in the right mind would consider PTB *saːwm for 'three'.

Also the conditions of these variants are different from the the one that causes the alternation in bcu.

From thirty onwards a rather odd reduplication of the numeral takes place. Examples:
gsum-cu-so-gcig 31
bzhi-bcu-zhe-gnyis 42
dgu-bcu-go-dgu 99

I have no idea why gsum-cu-so-gcig suddenly loses the b- prefix for ten. What is clear though is that these variants are not conditioned by an a in the next syllable. It can also probably be proposed that there's something going on related to 'unaccentedness'. But it's rather vague, and it is remarkable that the vowel shift is consistent with the vowel shift we see in bcu. But bcu could probably not be said to have been unaccented in a word like  bco-lnga. Why would words with the vowel a attract the accent more than with, say, the vowel u?

Ah! The wondrous world of a ridiculously archaic classical language that somehow managed to not weed out all bizarre irregularities. Unlike Sanskrit which lost much of it's usefulness to historical linguists after the Vedic period.

I'm going to stop my rambling here, because I have absolutely no idea what to do with this information, it's a lot to take in and rather confusing. But any idea's and suggestions are more than welcome

No native word for earthquakes?

So, I was wondering today, what the native word for Earthquake would be in Japanese.

As you might know, an enormous part of the Japanese vocabulary consists of Chinese loanwords. So the word for earthquake is 地震 jishin, whose individual elements translate perfectly as EARTH-QUAKE(or SHAKE if you want to be antsy).

But considering where Japan is stationed and its frequency of earthquakes, it seems ridiculous to assume that they do not have a native word for Earthquake. Especially common things that are native to Japanese culture have only a Japanese name, or both a Japanese and Chinese name, but this word, nothing.

I tried checking my dictionary, I tried searching in the gogenyuuraijiten (語源由来辞典）　(Lovely etymological dictionary site if you know Japanese), and it didn't give me anything useful.

I find it perplexing that the Japanese only have a native word for tidal wave (the well known tsunami) but seem to lack a native word for earthquake.

Only reason I can think of why this word would fall out of use is because of some taboo, but I'm not a big fan of the whole taboo explanation, it's a bit easy.

So, is there anyone out there who has more knowledge of Japanese/Old Japanese than me who could tell me that there is a native word for earthquake, because curiosity is going to kill me like this.

EDIT: Glen Gorden has solved this mystery in the comments, be sure to check it out!

Plural glottal stops in Zénaga

So I figured I would start of looking at the plurals in Zénaga that have a glottal stop, while the singulars don't, since this turned out to be the smallest group.

First and most striking point was, that all these class three nouns are all feminine. Feminines in Berber languages are generally quite easily recognised by a t- -t circumfix in the singular, and a t- prefix in the plural with a plural suffix that is usually -n or -ən.

Looking at the class 3 nouns, it all fell into place, which is quite exciting. There are two types of class 3 nouns.

Class 3a [1]

• sg. tkuṛṃuẒ pl. tkuṛṃuẓ̄aʔn 'pinch' (note: capital letters stand for long consonants)
• sg. taʔräS pl. tiʔrəššaʔn 'tear'
• sg. taʔmmärt pl. tuʔṃṃuṛaʔn 'beard'
• sg. tīnəS pl. tīnəššaʔn 'season'

If you know that it's unusual to have long consonants at the end of a word, and that Zénaga feminines usually have a t- -ḏ the original situation of these nouns should be obvious. All the plurals would point to an original cluster in the singular of CCC or even CCCC, which were simplyfied, losing the ʔ in the proces.

So in Pre-Zénaga these pairs would look as follows:

• sg. *t-kərməẓʔ-t pl. *t-kərməẓʔ-an 'pinch' (*ʔ can only appear in coda in Zénaga and therefore metathesises)
  
• sg. *ta-ʔraš(š)ʔ-t pl. *t-əʔrəššʔ-an 'tear'
• sg. *ta-ʔmmarʔ-t pl. *t-əʔmmərʔ-an 'beard'
• sg. *t-əH*nəš(š)ʔ-t pl. *t-əH*nəššʔ-an 'season'

Although these forms are probably quite close to Proto-Berber, I'm pretty sure there has to have been some syncopé period for Zénaga to get these final clusters in the first place.

So, to conclude, 3a is in fact regular, but just obscured by sound laws and lack of analogy. Now on to the next class!

Class 3b

• sg. tiʔrih pl. tiʔraʔn 'cry, call, shriek' [2]
• sg. täguḅḅah pl. tguḅḅaʔn 'petit delou (pour le puits)' (help me out french speaking readers, what is a delou, my dictionary knows nothing)
• sg. tgäʔrih pl. tgiʔraʔn (though also a pl. tgäʔrən) 'water taken from the well'
• sg. täyäkkäh pl. tiykaʔn 'flash'

Sooooo. This is a lot weirder isn't it? Although the system itself is clear, there doesn't seem to be a wonderful obvious regular explanation.

First of all, let's mark what's weird. First of all, all the members of class 3b are feminines that do have the t-/ta- prefix but all lack the -t/-ḏ suffix. On top of that the h shows that there is no word final glottal stop. The plurals are pretty regular and do show a glottal stop.

How do we solve this? One might be tempten to suggest that  a *ʔḏ > ø shift occurred, but besides this being phonetically implausible, there's also direct evidence that this can not be right.

For example the word (one of many):
sg. t(ə)nəgmiʔḏ pl. t(ə)nəgmaʔn (also has plural in -ayn and -adən) 'young girl, 14 to 17 years of age' where it's clear that there is nothing wrong with the cluster ʔḏ.

But this remains the only problematic class of the plural glottal stops in Zénaga. I am not sure about it's origin, but at least it's uniform.

Of course, because languages have this terrible habit of never listening to anyone, there is one oddball.

sg. taʔð̣uð̣ pl. tuʔð̣uð̣aʔn 'wool' (Font problems disallows writind d with line and dot (ḏ̣ see?) so went for the ð with dot approach). Luckily, there is also a rare variant of the singular taʔð̣uḌ the regular class 3a form from *taʔḍəḍʔt. I have no idea why this variant exists though, 'frequency of use' is such a frequently used excuse, and quite terrible, because any historical linguist should know by now that the frequently used words tend to be the ones that don't change rather than do (take French être, few verbs in the French language that retained such an archaic conjugation).

[1] For the sake of completeness I have chosen to use Taine-Cheikh's transcription which is extremely phonetic, rather than phonemic. For reconstructions I will use a more phonemic transcription, but not all phonemic values are exactly clear, for example in some cases ṃ seems to be phonemic, just like u isn't always an allophone of ə.

[2] One thing I tihnk I forgot to mention is that not all glottal stops have their origin in a Proto-Berber *ʔ. Some of the Zénaga glottal stops come from *γ which is obvious from the Zénaga root ʔNʔ (Aorist: yaʔni) 'to kill'. which is related to Tarifiyt nəγ 'to kill' . And also the word tiʔraʔn has its second glottal stop from an original *γ, though in the plural and verbal system this does not seem to make a difference in the way they behave, but I have included it into my database just in case. 

Zénaga and Hittite

I'm currently doing research on the Zénaga language. A Berber language spoken in Mauritania by about 200 people. The language is fascinatingly and quite radically different from all other Berber languages. First of all it has retained all proto-berber vowel values *i *a and *u yet it lost length distinction, in most Berber languages the distribution is pretty much reversed, where most vowel values are lost but length is retained quite well.

But the most fascinating of all is that Zénaga takes up a place in the reconstruction of Proto-Berber in much the same way that Hittite does. First of all it is quite obvious that Zénaga split of earlier than the rest, and just like Hittite, Zénaga provides proof for Laryngeals that were hypothesised to exist but unproven. I find it fascinating how (pre-)history repeats itself like that.

The realisation that Zénaga /ʔ/ was indeed the missing link, is quite a recent discovery discovered by Maarten Kossmann and Catherine Taine-Cheikh around the same time (2001 and 2004 respectively). This evidence was mostly based on the so called CC* verbs (terminology by Kossmann, a bit cryptic but it will do).

CC* verbs are a special class of roots with two root consonants that have a different inflection in the Perfect than other verbs with two root consonant (CC verbs).

For example there are the Tarifiyt roots NN 'to say' and G*/GG* 'to do' whose paradigms in the perfective look as follows:

	CC	CC*
1sg	nnaγ	ggiγ
2sg	tənnad	təggid
3sg m	yənna	yəgga
3sg f	tənna	təgga
1pl	nənna	nəgga
2pl m	tənnam	təggim
2pl f	tennant	təggint
3pl m	nnan	ggin
3pl f	nnant	ggint

As you can see, in CC verbs the final vowel is consistent in CC* verbs the vowel is i in closed syllable and a in open syllables. In Zénaga these paradigms look like this:

	CC	CC*
1sg	ənnag	əgag
2sg	tənnad	təgaʔd
3sg m	yənnah	yəga
3sg f	tənnah	təga
1pl	nənnah	nəga
2pl m	tənnam	təgaʔm
2pl f	tennəmmad	təgaʔmmad
3pl m	ənnan	əgaʔn
3pl f	ənnənnad	gaʔnnad

I need to add a note that I am not completely clear on the feminine plurals, difficult data to find, and I don't have any articles that neatly describe the paradigm.

So what can we tell from this? Well that CC* used to have a third root consonant *ʔ!
How does that work? As you can see in open syllables the *ʔ does not manifest but rather just ends in the vowel, it can still be differentiated from verbs that don't have a final glottal stop though, since they end in a consonant h. When a consonant follows The *ʔ appears, it can only appear in coda before another consonant.

Only oddball is the 1sg, Kossmann (2005) notes that a *ʔ before g is lost, though the evidence for it is rather marginal, there can not really be another way to explain this.

So, if the verbal paradigm can have its Proto-Berber *ʔ reconstructed on basis of Zénaga verbs, surely this should work on the nouns as well. No extensive research has been done on this yet, but it's quite clear that it does not work as beautifully as the verbal system.

And this is where my current research kicks in. I have looked for all Nouns that have:

1. Glottal stop both in singular and plural (So, the expected pattern)
2. Glottal stop only in singular
3. Glottal stop only in plural

After having collected this data, I'm going to further analyse it to see if there is any kind of system going on. From eyeballing my data (based on Taine-Cheikh's fantastic Zénaga-French dictionary) it seems like there is something of analogy going on with some of the more common plural suffixes like -an and -ayn.

Just to give you a little illustration here follow 3 examples of the above mentioned three groups:

1. sg. təffəʔd pl. təffaʔn 'present, gift'
2. sg,  əʔllə pl. əʔllan 'Achilles tendon'
3. sg. təʔmmart pl. təʔmraʔn 'Beard' (A berberologist favorite, since it's so well attested in Berber languages)

But the biggest problem of all is that many of the verbs consist of both class 1 and 2, as in they have several plurals, and within class 2 often have several different plural suffixes, and in the case of the example for 3 there are no less but 4 different plurals, of which 2 present a glottal stop and the other two don't.

Alert readers might already have a hunch about the origin of 3. I'll write about that coming up blogpost.