Stress and tone in Mandarin

After googling for "stress" and "mandarin", I came across one forum where a member named Carlo notifies us of a dictionary of word stress for Mandarin. As he explains the situation in more detail with helpful examples, he adds: "So some say that putonghua has several lexical stress patterns: strong-neutral, strong-weak, and weak-strong."

Instinctively, I'm skeptical that our analysis of Mandarin stress must be so complicated (ie. strong-weak versus strong-neutral) and feel this may be an illusion caused by the interaction of stress with tone. I have my own language models in mind that would make things simpler than this, but I'll continue looking out for more info to test my thoughts.

The imaginary tonal language

Having read John Well's Phonetic Blog on Norwegian and Swedish pitch accent the other day, I see still that not everyone has the same ideas on how to convey a tonal language in the International Phonetic Alphabet (IPA). I suspect many haven't meditated on what they mean exactly when they describe a language as a "tonal language" or "stressed language". Being a native speaker of English, growing up with a Swedish-speaking grandmother, having learned French in school, and studying Mandarin in my adult years, my linguistic adventurism has driven me to a cheeky conclusion about this whole issue: There's no such thing as a tonal language at all.

Now, I realize that everyone can see that French leaves both native and non-native speakers the impression of a musical language that differs perceptibly from the stress accent that impregnates the sounds of English. And if anything else, Mandarin with four distinct tones may appear to logically conflict with my radical stance. Yet, my faith in the existence of a distinction began to erode when I'd speak with several native Chinese speakers who, in separate occasions, had informed me that Mandarin syllables are not to be pronounced with equal weight, a habit I picked up from French. Words like māma 'mother' (with "high level" tone) and bàba 'father' (with "high falling" tone) are easy examples of this uneven weight. The first syllable of each of these words, for all intents and purposes, may be described as... accented!


Imagine it: A tone-deaf world

Now, let's think about this for a moment. Mandarin has four tones and demonstrates an accent pattern overlaid on top of that. Accent and tone together?? Yes!! When I realized that tone and accent could co-exist, I found myself free-floating through an exciting, new linguistic landscape that I'd never noticed before. A sea of novel possibilities and combinations, beyond a simplistic contrast of tonal and non-tonal, beyond the us versus them. A world where either every language on our planet is a "tonal language" or none at all. All tongues become more familiar, less alien.

Taking from my lessons in Mandarin, I began to examine my English differently. No longer believing in the common lie that it was a non-tonal language, I started to see that English had both accent and tone too. Its one and only default tone could be described as a falling tone, equivalent to the "fourth tone" of Mandarin. I could now perceive in a word like happy that both of its syllables inherently contained a falling tone, but I also came to notice that the so-called "stress accent" (as well as the overall phrasal intonation) habitually distorts this monotonal equality by causing us to pronounce each syllable with unequal weight. In fact, I began to realize that the only reason why any language is said to have "stress accent" as opposed to "tonal accent" depends on what the default tone happens to be in the given language. So if we can describe English alternatively as a one-tone language with default "falling tone" (ie. a "stress language"), then we might similarly describe French as a one-tone language with default "level tone" (ie. a "tonal language"). To complicate further, I've noticed some varieties of French here in Manitoba and elsewhere in Quebec using a falling tone like in English (ie. some French dialects seem to employ "stress" as in English). I find it intuitive to perceive these dialect differences as a simple mutation of the default tone from "level" to "falling", a change identical to that having occurred in polytonal Cantonese, in fact. The imaginary barrier between languages of tone and those of stress begin to evaporate.


Tackling Swedish tone

Returning to John Well's examination of Swedish tone, he mentions a commonly cited minimal pair in Swedish that I'm familiar with. He represents the pair with special tonal marking as follows: /ˈandən/ 'the duck' versus /ˇandən/ 'the spirit'. He then goes on to write:

"Not only is the pitch accent difference often hard to describe succinctly, its notation is controversial. Although it is pretty standard to write the simple tone with a simple stress mark, the IPA has no firm guidance on how to notate the compound tone."

But as per my observations above, it may be misleading to represent this ultimately imaginary "compound tone" in IPA. Instead, there is, as usual, both accent on the one hand and tone on the other in Swedish. Since Swedish has a default "high" tone ("falling") with a second "low" tone in stressed syllables, it seems to me that representing the pair simply as /ˈàndən/ 'the duck' versus /ˈándən/ 'the spirit' is enough for most occasions. If speaking about many dialects at once with differing tonal patterns, one may want to codify a common set of tones with superscript numbers. One may further opt for more detail by marking every syllable with its inherent tone, but only if the need arises. Finally, I find alternative tonal notations like /ˈan˥˧dɛn˩/ far too unreadable and unaesthetic to encourage.

Praisos #2

I was reading the latest post from Minoan Language Blog entitled Place-names on Cretan sealstones - A key to the decipherment of Minoan hieroglyphs? where in re of the artifact Praisos #2 the author observes: "Unfortunately there is no word separation; yet - if we follow van Effenterre's considerations - we can be almost sure that the word *inai was separate." This is a sound conclusion I can agree with given the likely division between two consecutive iotas. (For those unfamiliar with this artifact, please take a gaze Ray Brown's Eteocretan Language Pages from which I've borrowed the pictures below.)

A lack of word separation can be a source of headache for the would-be decipherer but it's common in ancient texts like this one. What could help is trying to deduce what are the likeliest rules of syllable structure and grammar this language might have had. For me, since I've strongly felt over the years that Eteocretan is related to Minoan and Etruscan, I'm guided by a generalized "Proto-Aegean" model of grammar and syllable structure. So let me explain what that is and how it leads me to separating the words as I do below.


Features of a common Proto-Aegean language family

As I've said before, I define a hypothetical ancestor of Minoan, Etruscan, Lemnian, Rhaetic, Eteocretan and Eteocypriot which I call Proto-Aegean. It would have been a fairly "syllabic" language (ie. no consonant clusters) with a mild stress accent lying by default on the initial syllable, although occasionally on the second. Judging by Etruscan alone, internal reconstruction affirms this conclusion about stress as it nicely explains the eventual development of initial clusters in Etruscan words that must have once had stress on the second syllable. I maintain there were no long vowels in its simple 5-vowel, V-shaped system consisting of *a, *e, *i, *o, and *u. Stops had no voice contrast and only a plain/aspirated distinction (ie. plain *t versus aspirated *tʰ). It had a default SOV word order.

Internal reconstruction also strongly suggests a Pre-Etruscan stage with the loss of word-final vowels (eg. Etruscan avil 'year' < *awilu). In Etrusco-Lemnian languages, there is an odd overabundance of word-final aspirated stops but this aspiration is explainable as a residue of the "whispered" word-final schwas as they disappeared beside word-medial plosives, eg. *ḳota 'four' > Etruscan huθ /hutʰ/. I also deduce that Proto-Aegean had certain grammatical features such as two tenses (unmarked present-future tense & a simple past in *-i) preceded optionally by modal markers like perfective *-ka (hence the perfective past *-ka-i becomes Etruscan -ce).


Enough! Let's parse and interpret!

So, long story short, based on considerations like the above, this is what I can currently pick out from this artifact:

[...]ona  desieme  tepimits  φa[...]
[...]do--iarala  φraisoi  inai[...]
[...]  restnm  tor  sar  doφ  sano
[...]satois  steφ  siatiun[...]
[...]anime  stepal  une  utat
[...]  sano  moselos  φraisona
[...]tsa  adoφ  tena
[...]ma  prainai  reri[...]
[...]irei  rerei  e[...]
[...]n   rirano[...]
[...]askes[...]
[...]it[...]

The most certain word or word stem repeated in this document by far is φraiso, the city of Praisos from where this artifact derives. Based on Etruscan vocabulary and grammar, I offer the following possible connections that I can perceive:

desieme 'with sacrifice' (= Etr tesiame [PyrT 1.x])
φraisoi 'in Praisos' (= Etr -i [locative])
φraisona 'Praisian, of Praisos' (= Etr -na  [pertinentive])
restn-m 'then wine lees' (= Etr restm-c 'and lees' [TCort A.ii])
tor 'to give' (= Etr tur [LL 11.iv])
doφ 'oath' (= Etr θuφ)
sar 'ten' (= Etr śar [TCort ii])
utat '(it is) served, (it is) delivered' (= Etr uθ 'to deliver' [LL 10.xiii])
une 'with libation' (= Etr une [LL 8.xvii])
tena '(they) present, (they) offer' (= Etr tena [CPer B.ii])

If my assigned values are even half on-track, it suggests that the topic of this artifact involves much the same as we might find on Etruscan stelae - a list of performed rites (presumably involving wine and lees, libations, oaths and animal sacrifice) performed in Praisos as a religious commemoration of a person, deity and/or event.


UPDATES
(2011 June 24) On Bayndor's advice, I corrected a typo that I'd copied and pasted from Ray Brown's website: *desime should be desieme and *tora should be tor. I've also changed the Etruscan comparandum for tor to reflect the newly apparent infinitive (ie. -a marks the present-future tense and an unmarked form represents an infinitive which has a meaning of 'to X' or 'X-ing' when translating into English).

Reconstructing Egyptian n(n)egation

This is what Antonio Loprieno says about Ancient Egyptian negation on page 127 of Ancient Egyptian: A linguistic introduction (1995):

"From an etymological point of view, nn is presumably the result of the addition of an intensifier to the nexal nj, much in the same way in which similar predicate denial operators developed in Indo-European languages: Latin non < *ne-œnum 'not-one,' English not, German nicht < *ne-wicht 'not-something,' etc."

From the immediate sound of it, I'd gather that Loprieno's use of "presumably" introduces a theory that the author himself considered possible but unverified. (The direct comparison with Latin non is naturally not to suggest that Egyptian and Indo-European negation are somehow related.) However this comparison still strikes me as awkward, perhaps because the direct comparison of the intensifying -n of Latin non with a merely implied suffix *-n in Egyptian nn is just *too* conveniently similar to me.

I also question that this transliterated nn properly reflects the intentions of the Egyptian scribes to write two instances of actual /n/. The simultaneous use of a negation nj suggests another interpretation to me. I find that in a wide number of languages, a negational morpheme will have both full and reduced variants which may share a common etymology. In French, for example, there is both unbounded non (when answering questions or negating adjectives, for example) and the reduced form ne used before verbs (normally with added pas, plus, guère or jamais after the verb, of course).

So can I get away with imagining something like Middle Egyptian *anī with reduced *ani behind spellings nj and nn respectively? How have others reconstructed this word? In this view, the terminating -n in the latter transliteration would not be actually phonetic but rather the byproduct of the lenient rules of the Egyptian writing system in how speech sounds are expressed. Consider that the Egyptian writing system evolved not just pragmatically as a writing system but also as a kind of sacred art form with much room for artistic creativity. As we see by how negation was often written (see image above), the second glyph (water symbol) may simply be meant to reinforce the /n/ already represented by the double-arms. And when we find, for example, the negational double-arms followed by not one but two water symbols, does this really prove two literal /n/'s? Or is it meant as a clever rebus showing a "dualized" /n/, thereby to be read nj (nb. the dual marker happened to be -j)?

Loprieno does in fact confess back on page 125:

"[...] the second one shows the same logographic sign accompanied by the phonogram n /n/ [...] and is conventionally transliterated nn, although it probably exhibited just a single /n/; [...]"

When I think about it more, these choice of glyphs to write negation in hieroglyphs is likely to have come about simply to disambiguate from the word 'of' which was already written as a single n. In actual speech, the difference between words 'of' and 'not' then should depend only on the choice of accompanying vowels surrounding the single /n/.

If so, Loprieno's example of Latin non and the suggestion of an alleged intensifying suffix amounts to a bit of a distraction from a likelier reality, as far as I can piece this together so far.

UPDATES
(2011 June 20) I repaired a teensy error I just noticed in my post, "the masculine dual marker happened to be -j", which I corrected to "the dual marker happened to be -j" since technically the masculine dual in all was properly -wj while it was -tj in the feminine dual.

Let us try to be on our guard against all that sort of thing

Duane at Abnormal Interests lately quoted an English translation of Plato's Politikós 263d:

"But indeed, my most courageous young friend, perhaps, if there is any other animal capable of thought, such as the crane appears to be, or any other like creature, and it perchance gives names, just as you do, it might in its pride of self oppose cranes to all other animals, and group the rest, men included, under one head, calling them by one name, which might very well be that of beasts. Now let us try to be on our guard against all that sort of thing."

It's an interesting quote when placed alongside modern quests to understand language origins in proto-humans. Do some of us likewise group modern humans separately from all other antecedents of our species and assume in shame that they must surely have been lacking in the ability to communicate in complex ways? And do we assume that oral speech, using a developed modern larynx, is the only way to convey rich ideas?

In the dumbed-down hype over the discovered FOXP2 gene a few years back, I could only be irritated by a persistent media bias towards spoken communication when discussing it, in blanket ignorance of the possible role gestural communication (including ancient-but-now-dead sign languages) might have played in our development towards what we now label "language" but which we improperly imply to mean only "vocal language" or "speech". If this gene were honestly at the very root of our capacity to understand grammar, it would also affect an ability to express in sign language or learn writing while having no effect on our other abilities somehow. Then again, isn't every form of thought, even outside of language proper, formed from some kind of set of "grammatical rules"? Circuits are structured by the grammar of logic (a combination of fundamental AND, OR and NOT gates), for example. Is one grammar truly different from another? If so, how? How can we then separate a capacity to decipher communicated ideas via consistent rules from a capacity to reason itself? If we admit what's suggested here, we admit that language is in effect universal to all living things in a plethora of forms and complexities. Language, when defined less biasedly as mere information exchange, turns out to be surprisingly commonplace.

Thankfully Geoffrey Pullum debunked much of this brown sugar drivel six years ago on Language Log. FOXP2's not a grammar gene afterall. It's terribly unlikely in fact that DNA codes for something as abstract as "language" or "grammar" by way of a single gene. For that matter, it's unrealistic to expect that the double helix codes such an abstract thing even explicitly at all. Check out new theories on how "emergence" plays a role in everything from gene function and computer programming to human language and grammatical rules.

So let's be on guard against unscientific human chauvinism; that somehow we modern humans gained the corner on language. Plato's pagan writings of long ago are correct. We're not that different from all the talking beasts around us. Science is merely reminding us of what was already gestured before we were born.

The house of Armna

In inscription ET Vs 1.133 I notice the last name of an individual written out as Armnes. I can't be terribly certain from only three names in this rather brief inscription but it seems at first blush to me that vel : armnes : vipes : should be read with a directive case form of Armna rather than with a genitive of Armne. In this way it reads in English: "Vel (son) to Armna, of the (gens) Vipe." We find the use of the directive case in -is signifying 'to, towards' in order to specify descent in TLE 321 as well. The name Vipe, by the way, was known to Latin-speaking Romans as Vibius.

I then wonder. Does Armna bear relation to the gloss *arim 'monkey', identified by Strabo as a Tyrrhenian word? Is this name a later form of *Arimna '(He) of the monkey'? Great Tinia only knows, stranger names had existed in the Etrusco-Roman record. Further, when I think of monkeys, my thoughts are drawn to Carthage from whence such exotic imports would sail to Etruria.

Ancient African adstrate in Etruscan

This subject, I feel, doesn't get enough attention and yet I think it's a fascinating topic: African loanwords in the Etruscan language. Considering what little Etruscanists currently seem to know about the language, unable even to explain conjugation and declension as I have attempted on this blog, it's probably unrealistic to expect that I should find a published mention of Punic or Berber adstrate in Etruscan, and yet, surely some African loans wandered their way into the vocabulary of ancient Italy, no?

The relationship of Etruria with Carthage is well proved ever since the Pyrgi Tablets were discovered. These are bilingual artifacts written in Etruscan and Punic. Punic, a dialect of Phoenician, was the official language of Ancient Carthage in what is now Tunisia. Berber dialects must also have been spoken in the Carthaginian region but Berber is a separate although related group to the Semitic and Egyptian languages traditionally to the east.

There is at least one word that may be considered Berber, at least based on the popular etymologies of the name Africa which source it either to afar 'dust' or to ifri 'cave, cavern' in allusion to local cave-dwelling. The name Afircina is recorded in ET AT 3.2 in the form of the type-I dative Avhircinasi 'to/for Afircina'. This mirrors Greek Ἀφρική 'Carthaginian region' and Latin Āfrica 'Carthaginian region'.

Now what about other African words, hmm? So far I've spied my eye on Proto-Berber *a-kal^[1] which bears a curious resemblance to Old Etruscan cal(u) 'earth', attested by calus 'of the earth' [TCap xv] and calusi-m 'and to the earth' [TLE 99]. Later, Etruscan a regularly rose to e before resonants like l. It should be noted that in Berber languages, noun stems are completed by additional gender prefixes like masculine singular *a- and feminine singular *ta-, so there's an unusual wealth of nouns with vowel onsets in their citation form.

Coincidentally Phoenix has been exploring the Berber language in great depth on his blog and I've been reading it avidly to guide me through this oft-neglected but fascinating and historically important language group.



NOTES
^[1] Phoenix in a comment below justifies Proto-Berber *a-ʔkal instead.

The Etruscan name Ramnuna

In inscription ET Vs 1.60, we find a name written out as Ramnunas. This is the Etruscan genitive of Ramnuna used as a male praenomen in this inscription. This is probably a reduction of earlier *Ramniiuna composed of *Ramniiu and the pertinentive suffix -na commonly used among other things to form family names.

At that, we can turn our attention to the Roman nomen Ramnius, likely the ultimate source of the Etruscan name. Ramnius may mean 'of the Ramnes' and the Ramnes were an ancient Italic tribe said to have been instrumental in the initial founding of Rome.