A failure of parallelism, sort of

In this headline from the 21st:

The crucial part is the NP

(PA) child- and gang-rapes

a reduced variant of the coordination child rapes and gang rapes — with rapes “factored out” of the full coordination, leaving the two-conjunct constituent child and gang. What gives this reduced coordination the whiff of non-parallelism is the difference in the way the factor rapes is semantically related to the two conjuncts child and gang: the first conjunct, child, functions as patient, or affected participant, with the factor rapes (like a canonical syntactic object; in a child rape, someone rapes a child), while the second conjunct, gang, functions as agent, or active participant, with this factor (like a canonical syntactic subject; in a gang rape, a gang rapes someone).

The coordination of patient with agent has a mildly zeugmatic flavor. It probably adds a bit of processing difficulty to this example — and it’s certainly enough to make a linguist like me take notice of the headline.

My 7/10/05 Language Log posting “Still more Declaration of Independence” with three conditions on factorable coordination:

Category Likeness: the [conjuncts] must belong to the same syntactic category.

Distributivity: the factor must be [syntactically] well-formed in combination with each of the conjuncts.  (Crudely, X (Y + Z) = (XY) + (XZ).)

Factor Constancy: the factor must have the same semantics in combination with each of the conjuncts.

It’s Factor Constancy that’s violated by classic examples of zeugma (wth the factor in ALL CAPS and the conjuncts in boldface):

*He PUT OUT his cigar and the cat.

Here put out is ‘extinguished’ for the first conjunct but ‘expelled’ for the second.

In that Language Log posting I suggested that

Distributivity … might be strengthened by [also] requiring Relational Constancy, that is, by requiring that the factor bear the same syntactic relation to each of the factors.

Now I note that Factor Constancy might be strengthened by also requiring

Role Constancy: the conjuncts must have the same semantic relation to the factor.

Either Factor Constancy or Role Constancy would rule out things like

*I ate a mixed salad and Saturday.

(where the NP a mixed salad is a direct object and denotes a patient, while the NP Saturday is an adverbial and denotes a (period of) time).

But Role Constancy would rule out (PA) and other [ N1 Conj N2 ] [ N3 ] compounds where the role of N1 with resoect to N3 is different from the role of N2 with respect to N3, for example:

fish or lobster sauce; mink and other leather oils

(where fish sauce and mink oil are source compounds — sauce made from fish, oil made from minks — while lobster sauce and leather oils are use compounds — sauce for (use on) lobster, oils for (use on) leather),

Clearly these are very complex matters. For N + N compounds, however, there is a possible resolution that would disallow things like the mixed salad and Saturday example, but allow compounds like mink and other leather oils.

The idea is to take the semantics of subsective N1 + N2 compounds to be as unspecified as possible semantically: ‘N2 associated with N1’.  That would freely allow reduced coordination (except for individual compounds that are idioms).

But that can’t be the whole story, since there’s the familiar fact that ordinary (Type O) N + N compounds fall into a small number of types according to the semantic relationship between their parts, with the first N as source, use, agent, patient, etc.: cucumber soap is a source compound, body soap a use compound, gang rape an agent compound, child rape a patient compound, etc. If such content is conventional, but not straightforwardly a matter of (truth-functional) semantics, what is it?

The easy answer is that it’s a conventional implicature — perhaps, in more ordinary language, a kind of connotation — associated with a set of compounds, but that requires providing an account of how these implicatures are to be described. Now there’s a project.

One Response to “A failure of parallelism, sort of”

  1. javava2012 Says:

    Thank you, Arnold. Had I reversed the words ‘child’ and ‘rape’ I think the sense would have been clearer — “to a linguist ‘like’ you. I was unfamiliar with the ‘Zergma’ concept. It’s nice to learn something every day! (Today I also learned there is such a thing as a “Walton’s Mountain Museum,” housed in the town where the ’70’s series author grew up and worked.)

Leave a Reply

%d bloggers like this: