Analytic languages and their function. (3)

jess tauber phonosemantics at
Sat May 27 08:46:21 UTC 2006

Aya Katz k-i:kama:n-ude:

>But animal cries, even those of our closest relatives, tend to be holistic, in the sense that a whole sentence is communicated with a single word, as is the case in the most synthetic of human languages.<

Not so fast...  Single phonological 'word' yes, but still morphologically complex, to the analyst (even if the speaker is unaware of the constructive pre-packaging of the finished product).

Let's also consider the good number of historically fusional but synchronically isolating/analytical languages exist (such as in Tibeto-Burman) where ancient derivational affixes have first merged with other phonemes sometimes creating consonantal contrasts, which then were lost to tonal development, and so on. No longer synchronically analyzable. There are many languages where morphemes are at the level of individual distinctive features, so worn down are the original forms. In some of the San languages, despite a monosyllabizing tendency, a number of ancient morphemes still appear to be present. 

Then in many analytical/isolating monosyllable languages you need larger collocational context in order to pin down which lexeme you're actually dealing with, because too many of its old features have been lost or rendered unrecognizable- it might be a little odd from the POV of usual thinking, but one might argue that in such cases the morphological word is in some way obligatorily larger than the phonological word (even when not dealing with 'affixes' per se). This would be the flip side of the situation in polysynthetic languages. So the phonological and morphological word are in an endless struggle to try to contain each other. Embedding versus serialization, etc, might also be part of the same type of struggle, in the larger arena. Sometimes you get the tiger.. sometimes the tiger gets you.

Animal cries seem to display the same sort of typological variations, along certain parameters, that human languages do- some are more synthetic and 'holistic', but others seem much more analytical. One size doesn't fit all.
Research done in animal communicative ethology shows the same sorts of automatizing processes we see in human languages- bits and pieces of pre-existing behavior, often anticipatory or initial movements ('Modal Action Patterns', or MAP's), get shortened, fused, etc. Eventually shorter still, the older pieces may no longer be evident. In other cases its the final products (glandular, wastes, etc.- which by the way are the origins of the vast majority of body pigments) or terminal motions in an action. Some chains are serializations (normally sequential motions a, b, c, d become much reduced chain a'b'c'd')- I don't know if others are 'embedded', or there are head/dependent relations that would determine which old signal gets 'inflected' or otherwise modified. The researchers aren't thinking along these lines, since they slavishly follow Old Chomsky, and have not kept up so far as I know with later developments or movements in linguistics- for those in the 'animal language' biz the only focus seems to be on what counts as *syntax*, which is unfortunate.

All human languages have phonemes- ranging from the minimum (with compensatory or substitutive prosody and large allophonic variation) in Piraha~ to maximum in San languages (@120-140 depending on what counts as a contrast) or Ubykh with 80+. With only so many distinctive features to play with, we get the higher numbers by piling them onto each other. It's something we're able to do, yet not that many languages have the highest or lowest numbers. Obviously there is a preference zone (IIRC somewhere between 20-40 phonemes). But that's us humans. Animals may have different preferences. Interestingly, if one extrapolates either up or down, one ends up with a near continuum featurally, from the perspective of signal quantization.

If we go below the Piraha~ phoneme number we leave recognizable segmentality altogether, and end up with a multidimensional continuum of prosodic effects. If packaged properly, with lots of context to help you along, the system may still be communicatively viable- it just takes a bit of effort to imagine it from the human perspective. In the other direction imagine how many phonemes one could generate if one used all human contrasts maximally. The resulting system might leave click languages in the dust in terms of jawbreaking articulations.

Now vary the number (and form) of feature contrasts we're allowed to use and you end up with an even larger playing field. Each species will have its own, limited by its anatomy and neurology in absolute terms, and to a probable subset of the possible for actual communicative purposes. Interestingly, the articulators themselves (and this goes beyond vocalism to ALL possible signal generators) often shift their own forms and motional possibilities in order to increase the effectiveness of the signaling. It may be hard for us to recognize some of the signals we are looking at as being 'phonemic', 'morphemic', 'phrasal', or 'clausal' if we limit ourselves to seeing through the blinders of human norms. Or looking only at vocal signals (versus more global communication). I'm not saying that what animals are doing is 'language', but perhaps what humans do is just a special case of a more generalized set of communicative behavioral norms.

Jess Tauber

More information about the Funknet mailing list