March 24, 2005

Retrodiction confirmed

In an earlier post, I made an educated guess about Chippewa ("Anishinaabe") verbal morphology:

... I suspect on general principles that the system is quasi-regular: in other words, the relationship between the meaning and the combination of stem-forming elements, prefixes and suffixes is often regular and therefore predictable, but can also often be more or less opaque and "idiomatic".

Within a couple of hours, John Lawler sent a link to a 25-year-old paper confirming the guess: Richard Rhodes and John Lawler, "Athematic Metaphors", CLS 17 (1981). It's not very impressive to predict something that turns out to have been documented in a paper published 25 years ago, but I'll take my successes where I can...

As Rhodes and Lawler explain:

...after some time working in Ojibwe, one of us elicited an item mdwesjiged in a text. From our general knowledge of Ojibwe, we knew that the stem of this word has the morphemes madwe- ‘be/make a sound (at a distance)’, -sid- ‘cause to be in a location/state’, and -ige ‘unspecified object’. By the regular rules of Ojibwe semantics, this means, altogether, ‘make a noise at a distance by moving things around’.

However, our native speaker insisted that this word only meant ‘ring the church bells,’ even though he admitted that the form inflected for definite object, mdwesdood, could be used by someone in one room commenting on noise emitting from another, as, for example, when one sits near the kitchen in a restaurant. It was only after some prodding that he allowed that, in fact, mdwesjiged could refer to other situations in which things were being moved around and were making noise, but were out of sight.

Now one might think, as we at first did, that this was simply a reaction to the situation in which this word was being discussed, but subsequent elicitation has shown that one of the “meanings” of mdwesjiged is ‘ring the church bells,’ although people who come from Protestant areas, where church bells are less prevalent, don’t get this reading as strongly. Similarly, enormous numbers of Ojibwe words whose semantics are clear from their structure and which can be used in those meanings have important restrictions on their “meanings” in normal usage.

You could compare this to an English compound like "chair lift". Most people will tell you that it refers to a particular kind of contraption found on ski slopes, which carries people up the hill on chairs suspended from a cable. But according to the regular semantics of English, "chair lift" could be used to refer to many other things, for example to a kind of exercise done while sitting in a chair, or a scheme for storing extra furniture by hanging it on the wall. Similarly, a "spark plug" could be a device to keep live embers from escaping from a chimney -- but it isn't, it's a very specific part of certain types of internal combustion engine. Or consider conscription, which literally means "writing together", but normally refers to being drafted into military service.

Rhodes and Lawler give some other examples:

zhisjiged put things in a certain place set the table
zhising be laying in a certain way be written
gshkitod be able afford
aanjpizod change s.t. tied change a bandage
dbaaknigaazod be judged be in court
miijgaazod be given s.t. be on welfare

They interpret these as cases where extra meanings arise from a "prototype context" whose properties are carried along as a particularization of a more general, compositional sense. They contrast these with idioms where the compositional meaning is lacking

giiwsed *walk around hunt
mzinhigan *s.t. to carve/write on/with paper, book
mzinhiged *carve/write get credit
namhaad *greet pray
zaaghigan *s.t. to get out with lake
waawaaskone *light flower
baashkzang *burst/break s.t. with heat shoot s.t. (with a gun)

and/or where (they feel that) the meaning is not a predictable consequence of any "prototype context":

zaaghaad be stingy with s.t. (animate) love s.o.
waabgookookoo white owl wedding cake
mnidoons little spirit flying insect
baasod be dry be thirsty

Whatever the right taxonomy is, it's clear that the Chippewa/Ojibwe system of stem-forming elements, prefixes and suffixes is a rich combination of regularities, subregularities and irregularities -- just like every other such system I've ever learned about.


Posted by Mark Liberman at March 24, 2005 11:39 PM