Non-concatenative morphology in Chinuk Wawa?
Non-concatenative (non-segmental) morphology is, I’d say, assumed to be exotic.
Compounds illustrated (Image credit: Pinterest)
Here I’m talking about those morphemes (i.e. those meaningful building blocks) of a language that are not just made of strings of phonemes (i.e. distinctive sounds). We tend to be taught scads of examples in linguistics classes that show morphemes have pretty fixed, simple-to-pronounce shapes, such as the Arabic prefix mu-, Salish s- Nominalizer, Hungarian -om for 1st person with a definite object, etc.
You’re left with the implication that morphemes not composed of phoneme strings, as seen from Chalcatongo Mixtec’s tonal inflection in káʔba “filth” ↔ káʔbá “dirty”, are unusual. (Thank you Wikipedia for the example.) Yet very large numbers of languages use such processes on a routine basis.
Incidentally, here’s a partial list of what a nonconcatenative morpheme can be like:
- a tone
- a stress
- a reduplication (repeating part or all of a word)
- a truncation (chopping off part of a word, to impart a new meaning to the word)
- a change in vowel quality, a.k.a. ablaut
- a change in consonant manner of articulation
- (signed languages also have their particular nonconcatenative morphological strategies)
My personal experience leaves me with the impression that we linguists, to the extent that we’re aware of them, tend also to assume nonconcatenative morphemes are as simple as that list would imply, e.g. “place stress here” or “just add low tone there” to convey a certain meaning.
However, it’s easy to find hundreds of nonconcatenative morphemes in the linguistic literature that are more multifaceted than that, e.g. English reduplication to express a canonical example of a phenomenon (as in “No, I want a car car, not a Tesla”) both repeats the stem and puts a special higher intonation on it. That’s 2 non-segmental processes at once. I’ve often noticed that even excellent grammatical descriptions of a language will tend to tell that e.g. “there’s reduplication of the shape CVC” (consonant plus vowel plus consonant), but without providing important details of stress etc.
Furthermore, we characteristically suppose that nonconcatenative morphemes are purely nonconcatenative. And yet, just to look at one of the languages I happen to research, I can easily give you examples from Chinuk Wawa’s co-input language Lower Chehalis Salish of morphemes that are simultaneously concatenative sequences of segments and non-concatenative. There’s a cute little unstressed Diminutive suffix -uʔ that’s just 2 phonemes attached after the word stem — but it also includes a glottal stop that gets infixed inside the word, directly following the stressed vowel. (An example that my readers here might enjoy is skʷəkʷáʔmuʔ ‘an annoying kid’, which is the Diminutive of skʷəkʷə́m ‘a monster’ — the source of “skukoom” in CW. The original stressed schwa of skʷəkʷə́m, you may notice, also gets mutated to á.)
Chinuk Wawa disproves still another typical and wrong presumption of linguists, that contact languages are “simple”. Creole CW does not merely rattle off sequences of root-word after root-word, in a babytalk style. It has, not just grammatical affixes (such as hayu- ‘Progressive action’), but also nonconcatenative morphology, in the form of productive reduplication, the ‘Distributive action’ form that places an unstressed copy of the stem after the stressed first occurrence of it. (E.g. nánich-nanich ‘to be looking around’.) I’ve previously suggested that CW’s reduplication is likely to trace back to Salish influence.
This discussion leads up to my proposing that both creole & pidgin Chinuk Wawa include at least one more example of nonconcatenative morphology. Here I’ll be using different wording than I have previously, but talking about something I’ve already examined many times on this website: compound words. (Which we conventionally signal by a joining dash, –.)
In my analysis, the prototypical compound in CW is a Noun + a Noun; there are Verb + Noun compounds etc. as well, all fitting a single pattern. Here, a stem that otherwise can be a verb (whether Active or Stative) is turned into the descriptor of an immediately following “head” noun, whose primary stress gets reduced to a secondary stress. Simple examples:
(the CW name of naturalist JK Townsend)
The addition of this stress-reduction morpheme, call it V̀, indicates that a relatively simple noun has become a semantically more complex compound. Stated another way, V̀ flags the immediately preceding noun or verb as having lost its potential to be a predicate, thus depending on the following noun for its proper interpretation.
Without V̀, you’d have these equally valid, but quite different, CW expressions:
- kə́ləkələ táyi
‘a/the chief is a bird’
- nánich lakasét
‘look at the box!’
Now to ask my usual question about everything in Chinuk Wawa — where did this come from?
Tentatively, I’d connect this operation in CW with at least 2 of its major input languages. I’m not really aware of compounds in French working like this, nor of there being many compounds in Chinookan.
But we can pretty obviously look to English, where compounds undergo a similar stress alteration; anyone with a fluent grasp of English grammar knows that a compound “bird-chief” would have secondary stress on the second word: bírd-chìef.
The other potential model is in Lower Chehalis Salish, where I’ve recently found a comparable process, albeit a more complex one. There, I was wondering why unstressed ə becomes stressed í in quite a few instances. (An instance of the “ablaut” mentioned above.) The effect seems to be to turn a noun root into a “combining/modifying form”, i.e. the first element in a Noun+Noun compound. (The second, “head” noun, as in CW, is demoted to secondary stress.) Examples I can easily think of:
(compare ʔál̓əs, x̣áʔqaʔ
‘children of royalty’
(compare ɬəw̓ál̓məš, (s)y̓ə́q
(compare ɬəw̓ál̓məš, q̓ʷəč̓əstəy̓əp
(compare yám̓əc, qʷə́nɬ
(compare ?, skʷən̓túʔ
So there you have it, a little lesson on how Chinuk Wawa is more complex than it has historically been described by folks who hadn’t paid terribly close attention to how it operates.
We’ve also seen that CW’s non-concatenative morphology may be an inheritance from its Indigenous ancestors.
1) I’m not sure why redup would by definition be seen as non-concatenative, though. I think of it simply as affixation with a phonologically underspecified affix.
2) Might not the loss of stress in the compunds be a phonological affair rather than a morphological one? With ”the chief is a bird”, you are dealing with several words, but with ”bird-chief” there is only one, and then a phonological rule (I am obviously just guessing here!) disallows several instances of stress within the confines of a word.
3) ”Where did this come from?”. OK, I admit being predictable, but whenever you ask that question, I usually say something to the effect that only idiosyncratic features can be meaningfully traced to the input languages. For things that are exceedingly normal, that makes little sense. ”CW has consonants and vowels because languages X, Y and Z did” – that wouldn’t seem like a very interesting or meaningful claim to me. I suspect that the loss of stress here belongs to the category ” exceedingly common”.
4) ”Chinuk Wawa disproves still another typical and wrong presumption of linguists, that contact languages are “simple””. Two things here:
a) few if any linguists believe that ”contact languages” are simple. The category after all also includes intertwiners like like Michif. Some of us (but we are becoming a minority, I fear) believe that pidgins and creoles (but not all contact languages) are less complex than traditional languages (all languages are of course complex, but some even more so than others).
b) The presence of a complex feature in a pidgin or creole does not disprove the idea. That would be like saying that someone isn’t poor because they possess two dollars. Or that nuclear wepons aren’t dangerous just because knives can also cause damage. The ”less complex” claim of ours must obviously be taken in relation to other languages. ”Less complex” is not an absolute attribute, but a relational one. A complexity in CW proves that there is complexity in CW, but NOT that there is as much complexity in CW as there is in the average language.
LikeLiked by 1 person
I second Dr. Parkvall’s comment. I like to supplement one thing. One cannot make generalisations about “contact languages” because they are a diverse group: pidgin, creoles, intertwined languages, converted languages. Pidgins are the only ones that are purely simple, and that is also the purpose of pidgins, and the most rational choice that pidgin creators can make in the situation they end up in. Johanna Nichols 1992 book was perhaps the first in which language contact was mentioned as a factor that complexities languages. But the social circumstances (long-term multilingualism) are different where pidgins emerge.
LikeLiked by 1 person
All excellent comments, naturally! Many thanks to you, Mikael and Peter.
My article today is a continuation of my dissatisfactions with previous Chinuk Wawa scholarship (up to about 1980) that drove me to do my dissertation research. Huge advances have definitely been made in contact-language studies since then.
I enjoyed using today’s piece to argue over the boundaries between morphology and phonology. That’s not a settled question.
Prosodic / suprasegmental / intonational / etc. phenomena are notoriously under-researched by us linguists, so we have few clear ideas about the functions, and functional load, associated with them. In Chinuk Wawa, I perceive that prosody plays important roles in delineating phrase structure, so isn’t that a grammatical function — a morpheme or morphemes even?
Similarly, my understanding of CW is that what we’re standardly taught to call “syntax” is precisely what supplies the speaker/listener’s core grasp of clause structure, telling us what’s a predicate and what are the various kinds of arguments. Those functions are fulfilled by the addition of obligatory morphemes to roots/stems, in most other languages. This “word order” stuff is among the only mandatory elements in grammatical CW utterances, whereas most of the overt prefixal aspect marking etc. is optional. So, rather than seeing CW as uncommonly deficient in transitivity-related morphology, I sometimes conceptualize word order as having morphemic status in Chinuk Wawa.
Every language uses a unique combination of the known strategies for communicating various functions. For instance, expressing “wanting to” do something is done by a verb in some languages, an affixal morpheme in others, juxtaposed clauses in others, and so on. So, might it be that a pidgin-creole like Chinuk Wawa morphologizes some things that we don’t often see morphologized in other languages?
On that subject, it’s really interesting to consider what a unique array of things are pretty incontestibly grammaticalized into dependent morphemes in CW, prominently:
Distributive action (full-root postposed reduplication),
Causative voice (mamuk-),
Inceptive and Progressive aspects (chaku- and hayu-),
Intensity (hayas(h)-), and
Lack of control (in the northern dialect) (t’ɬap-).
On one hand, that could seem like a deficient selection, leaving a number of functions unfilled. On the other, though, it reflects speakers’ concerns, and a pretty typical range of such for the Indigenous Pacific Northwest. And in fact other concepts are successfully conveyed by non-morphemic means, what RMW Dixon calls “strategies” (e.g. a conventionalized way to imply a Passive voice in CW is to use a transitive subject “they”). Languages are like that, yeah?
Special thanks for reminding me that not all phenomena in a young language necessarily have an external source. This deserves serious recognition, alongside the need to consider potential source languages.
For the benefit of my non-linguist (and non-specialist) readers, I should say that Dr Bakker’s work has been really important in proving that pidgin languages, for example, are not quite so simple as was once thought.