A little more on nonduality

« previous post | next post »

In my recent posting on uses of non-dual (outside the domain of the philosophical/religious position of nonduality or nondualism), I (informally) characterized the meaning of the expression as follows:

a non-dual X is simply something (of the appropriate category) that is not a dual X

This characterization incorporates an important observation about expressions of the form non-dual X, like non-dual citizen: they exhibit a "bracketing paradox", in that these expressions have one syntactic bracketing,

[non- + dual] + [X]

but a different composition for the purposes of semantics,

[non-] + [dual + X] 'something that is not a dual X' (e.g. 'someone who is not a dual citizen')

(and not 'a X that is not dual', e.g. 'a citizen who is not dual'). If you were hoping that semantic interpretation could build directly on morphological and syntactic structure, then cases like these are problematic.

There's quite a literature on bracketing paradoxes, of various types — for example, the problem presented by expressions like transformational grammarian, which are interpreted not as 'grammarian who is transformational', but as 'someone who studies or espouses transformational grammar', that is as transformational grammar plus -ian.

Non-dual citizen and transformational grammarian share another feature: both are built on adjective + noun combinations in which the adjective is not understood as simply predicating a quality or property of the noun; a dual citizen is not a citizen who is dual, and transformational grammar is not grammar that is transformational. This is "non-predicating modification".

There are many different sorts of non-predicating modification — including things like electrical engineer, Vietnamese war, indigenous language, marital bliss, and daily prayers — and there is a gigantic literature on their analysis, in a number of languages. Like noun-noun compounding, non-predicating modification shows some recurrent and frequent patterns in the semantic relationship between their parts, but (as Geoff Pullum noted recently) these relationships can sometimes be distant, unpredictable, and highly dependent on context. (I have a posting in preparation on one such case, the expression indigenous nudity.)

Some final words about resolving bracketing paradoxes: linguists have made a number of suggestions (sometimes different ones for different cases), but the approach I favor turns on looking at the (morpho)syntax-semantics interface a bit differently from the way I talked about it above. Instead of trying to read semantics directly from morphosyntactic STRUCTURES, we should be seeing both the details of morphosyntactic form and of semantics as specified by RULES. (This is an idea with a long history; it is scarcely original with me.)

In the case at hand, we say that some rules can license versions of complex expressions by associating a function on the semantics of the contributing expression with a function on the morphosyntax of the contributing expression. In particular, a rule associates (a) (a particular kind of) negation with (b) the prefixing of non- to the first word of the contributing expression. If the contributing expression is only one word long, then this operation isn't very interesting. But if it's dual citizen, then we get non-dual citizen 'someone who is not a dual citizen'.

Other rules will have morphological effecs on the LAST word of the contributing expression: for instance, a rule associating (a) the meaning (roughly) 'one who studies or espouses X' with (b) the suffixing of -ian to the last word of an expression denoting X, as in transformational grammarian.

Two somewhat more complex examples. First, a rule associating (a) the meaning 'ordinal' with (b) the use of the ordinal variant of the last word in a complex cardinal number: four hundred (and) fifty-first as the ordinal version of cardinal four hundred (and) fifty-one. And then a rule associating (a) plurality as a property of a complex nominal expression with (b) the appropriate plural variant of the last word in that expression: Christmas cookies as the plural of Christmas cookie, but also musk oxen for musk ox, snowmen and snow geese for snowman and snow goose, and swordfish for swordfish. (Again, I'm reporting on familiar examples, not breaking new ground. And there are more complexities, not covered here, in both the ordinal and plural cases.)


Comments are closed.