Zero relationships
« previous post | next post »
My posting a while back on countification (M(ass)>C(ount) conversion of nouns, with accompanying individuating semantics) elicited e-mail and blogging about other cases of zero relationships in English (of which there are a lot, though all pretty much irrelevant to my topic in that posting), and now Bill Poser's posting on moose has set off a comments thread on zero plurals (moose being an example of a noun with a zero plural).
There's an important point here: formal relationships — like phonological identity ("zero relationship"), suffixation by /z/, and systematic vowel alternations, are "just stuff". They have no intrinsic meaning on their own, but are available to serve all sorts of grammatical ends.
[To follow up on the moose comments: zero plurals are treated in every reputable reference grammar of English. The Cambridge Grammar of the English Language has a particularly nice succinct survey on pp. 1588-9.]
Phonological identity, or zero relationship, is the simplest sort of formal relationship there is for expanding on the lexicon (though from the semiotic point of view it's suboptimal), so it's no surprise that it's all over the place in languages, and particularly visible in languages (like English) that are not especially rich morphologically. (Other high-frequency strategies for expanding the lexicon include combining different stems, into compounds of various sorts, and combining duplicate copies of a single stem.)
Category-changing zero derivation. English has by-now fully lexicalized nounings of verbs (run, talk, and many, many others) and verbings of nouns (hammer, query, and many, many others); and there are other nounings and verbings "in the pipeline" (the verb verb itself, for example, plus things like the count nouns ask 'a request' and swear 'swear word' that I've posted about here and here; there are endless numbers of others that can be (and are) made up on the spot ("We'll Chevrolet to California", "They gave it a long contemplate").
[Digression: these "spontaneous" verbings and nounings have a profile that's familiar from N+N compounding and non-predicational Adj+N composites:
(a) There's a morphological configuration (N+N=N, Adj+N=N, N or Adj = V, V or Adj = N) that has, when you look at the whole range of actually occurring examples, very general and unspecific semantics (a N2 related to a N1 in some way, etc.).
(b) But there are recurrent subtypes, or patterns, of this configuration with much more specific semantics, so that they can be produced, and comprehended, "off the shelf", so to speak, without a lot of taking context, background knowledge, and so on into account. These recurrent subtypes have been catalogued in some detail for N+N compounds and for verbings.
(c) Nevertheless, novel instances, with very context- and knowledge-dependent interpretations, are freely available (e.g. N+N canoe wife, Adj+N diabetic insole 'insole for diabetics', to bookend 'to place between two paired items').
I suspect that this is the way "word formation" (including both various kinds of compounds and derivational formations) works a lot of the time: great vagueness in the associated semantics, a number of more specific templates for interpretation, and a big zone of open context-dependent interpretations. This is the pattern for N>V derivation in -ize and for duplicative compounds that have characteristic exemplars ("It was (not) a DATE date", on which there's now a fair literature) — and, I'm sure, for other types of word formation.
There's obviously a lot more to be said about category-changing zero relationships, but but I'll pass on to zero relationships between different subcategories of the same "part of speech", in particular, relationships between C and M uses of nouns. Here I'm going to focus on the (b)-type cases, where there are patterns of conversion with conventional semantics associated with them.
(This is going to be just a sampling of what's out there, and it doesn't pretend to give detailed semantics of the various types. I'm posting this to give a sense of how much diversity — and specificity — there is.)
M>C: countification. A M noun that refers to an assemblage or collection of elements has a C use to refer to one of these elements: spam, e-mail, porn, slang, folklore. Often commented on here, most recently here.
C or M>C: situational metonymy. Based on an association between pairs of referents in some specific situation — for example, the association between a customer in a restaurant and the item that customer has ordered. The metonymy can take C to C ("Table 4 has two hamburgers and two hotdogs sitting at it") but it can also take M to C ("Table 4 has two spaghettis and two lasagnas sitting at it"). (I'm paraphrasing examples from Geoff Nunberg's Ph.D. dissertation.)
M>C: variety individuation. A M noun has a C use to denote varieties or subtypes: "There are many lavenders growing in my garden" (not 'many lavender plants', but 'many types of lavender', e.g. Lavandula angustifolia, Lavandula dentata, Lavandula stoechas.
M>C: serving individuation. A M noun denoting something drinkable or edible has a C use to denote a serving of it ("I drank three Tsing Taos", "We ordered three Veal Orloffs").
C>M: the universal grinder. A C noun has a M use to denote a thing or things transformed (actually or imaginatively) into stuff ("The child took the flower and smashed it into pulp, and soon the table was covered with tulip" [invented]; "Biological 'success' simply means something like: there are many more pounds of dog in the world than there are pounds of wolf" [attested example supplied by Joel Wallenberg]).
C>M: expanse massification. A C noun has a M use to denote an assemblage of things presenting themselves en masse ("We rounded a corner in the Dutch countryside and were confronted with a huge expanse of tulip").
C>M: substance massification. A C noun denoting an individual has a M use to denote a generic substance or totality, usually in construction with a quantity determiner ("That's a lot of horse", "That's more elephant than we can handle").
I've distinguished two types of massification here, but the difference between them is not always clear, as in this attested example: "When huge deposits of ruby and sapphire were found several years ago, Cushman was one of the first dealers to venture into the often violent mining towns".
C>M: meat conversion. A C noun denoting a creature has a M use to denote the edible flesh of that creature — "There's not much rat in it!" ("We had chicken/lamb/eel for dinner").
(Some dictionaries list some of these M uses as subsenses of the entries for the C nouns, even though meat conversion is productive. [Digression: some meat conversion M noun uses are pre-empted by lexical items borrowed (long ago) from French: pork, for instance, normally pre-empts pig as a meat name.])
C>M: sex-part conversion. A C noun denoting a sexual part has a M use denoting this part as a generalized object of lust ("I'm looking for some cunt/pussy/cock/dick/ass").
All of the C~M conversions above can pretty easily be seen as metonymies. One last type is based on a metonymical idiom give head 'provide oral sex', but is not itself metonymical:
C>M: give (great/good) N. For instance: "I can still give great lecture"; "You give great interview!"; "He can really give rant".
(Older speakers often report finding the give-N idiom extensions to be too sexual for use in anything but the most informal contexts.)
Semi-final note. The C~M conversions are notably CONVENTIONALIZED: although, given enough context, you can make up a new instance on the spot and get away with it — people will see what you're trying to convey — and although maybe some of these inventions will become fixed as idioms, most of the occurrences fall into types, which people (in general, or in certain groups) will process as familiar patterns, without having to "work out" other people's intentions.
What this means is that, however "natural", intuitive, and obvious some of these conversions might seem to you, they are things you've LEARNED about the use of the language, and there is no guarantee that other languages will have the same conventions.
And in fact they do not. Jerry Sadock reports to me that conventionalized metonymies are virtually absent in Greenlandic Eskimo (and that creative metonymy is conversationally problematic), and a quick survey even of familiar European languages will show that some of them lack some of the conventions that English speakers find so natural (while having others that are initially strange to us). I'm not competent to survey such facts (though I've taken part in entertaining discussions with various multilingual linguists about them), so I won't pursue the topic here. But it would be nice to see someone take an extended look at it (and maybe someone has; I miss a lot of stuff).
Final note. What I've said here barely scratches the surface of zero relationships. There are zero relationships between adjectives and adverbs; between intransitive and transitive verbs; and, spectacularly, between different inflectional forms (as, for example, in English zero plurals, alluded to above: one sheep, two sheep; one moose, two moose; and some people's one croissant, two croissant). And more.
Zero relationship is a purely formal matter, with no overarching content of its own; it can be pressed into service to do all sorts of work.