Over on Language Log, Mark Liberman reported a headline that lots of people find at least a bit off:
(1) Unrest Spreads, Some Violently
As so often happens with attested examples, there are several things going on here at once, and they need to be disentangled: the adverbial violently (what does it modify?), the verb ellipsis in some violently (in which spreads is ellipted), and the ellipsis (or whatever it is — see below) within the M NP some (understood as some of it or some unrest). The last point is what I’m primarily interested in here.
A few words on the other two points, though.
On the function of violently, Barbara Partee’s comment on LLog summed up that part of the discussion in the comments section:
One point seems clear: the headline says that some of the spreading was violent but the writer probably meant that some of the unrest was violent.
Probably an effect of compression to fit space (remember that this is a headline), but otherwise it looks like a transferred epithet, or hypallage (most recent discussion on this blog here), with violence attributed to the spreading rather than the unrest.
As for the verb ellipsis, it’s unremarkable. The way this works will be clearer if we avoid the violently issue and shift to a PL C (Count), rather than SG M, noun, as in
(2) Protests erupt, some ___ violently
(antecedent VP underlined, ellipsis site marked by underscores). Here erupt violently is unproblematic, and so is ellipsis (or whatever) within the NP.
On to the NP some in (1) and (2). I know of three somewhat different treatments of cases like this:
(a) ellipsis of the head N (here, protests), leaving a determiner (here, some) as remainder, so that some is related to some protests;
(b) ellipsis of the (“partitive”) complement PP (here, of it or of them) to a head pronoun (here, some), so that some is related to some of it/them;
(c) fusion of determiner and head (or head and partitive complement) into a single word, some, that represents the sequence (see the note on CGEL‘s fusion analyses at the end of my “Combos” posting, here).
Fortunately, I don’t have to decide here which of these analyses is to be preferred, since any one of them will have to be sensitive (for a great many speakers) to whether the NP some is PL C or SG M.
[Background on C/M: check out my 2001 Stanford Semantics Festival paper, “Counting chad” (handout here), and my 2006 Language Log posting, “Plural, mass, collective” (here). The most important background point is that the classification of noun lexemes as C or M is based on their morphosyntactic characteristics: pluralizability (without semantic shift), selection of determiners, subject-verb agreement, and selection of anaphoric pronouns. The connection between this classification and the semantics of the lexemes is a complex one, which is why I prefer to use the labels C and M, rather than more transparently semantic labels.
(Countable and non-countable are particularly hazardous labels, because there are subtypes of M nouns that can be used as the objects of the verb count — among them, the “functional collectives” recently treated by Grimm and Levin (some discussion here); you can count the furniture in a room, the mail in your mailbox, the luggage that you check before a plane flight, etc., just as you can count a team or a group).
To make these matters even more complex, C/M interacts in intricate ways with SG/PL (in particular, through C nouns with zero PLs and C nouns that are PL-only, and also in proper names, like the band name Dead Tongues, that are PL in form but denote entities) and +COLL/-COLL (in collective C nouns and functional collectives, which act like collective M nouns); there are schemes for converting C nouns to homophonous M nouns, and vice versa; and some noun lexemes have a dual classification, as C or M.]
Now back to elliptical (or fused) some, in the contrast between pairs like the following:
(3a) Protests erupted, some violently. [PL C some]
(3b) Unrest erupted, some violently. [SG M some]
Not everyone finds (3b) problematic — I myself prefer (3a) to (3b), but would not object to (3b) — but many people do. The contrast carries over to examples that don’t have verb ellipsis, as in
(4a) Fluids, some polluted with diesel fuel, surged through the gap. [PL C some]
(4b) Seawater, some polluted with diesel fuel, surged through the gap. [SG M some]
Example (4b) was supplied by LLog commenter maidhc, who wrote that it “doesn’t sound right to me”; (47a) is a counterpart I concocted.
(Most seem to work like some, and so does all, although there are other possible uses for all in these contexts.)
Mark Mandel then commented perceptively on maidhc’s example:
I agree about that example. But consider [(5)] “Seawater, some polluted with diesel fuel, has been blown to the tops of many shoreline cliffs”? For me, your example sounds wrong because it’s about a single mass of seawater, where pollution physically can’t be limited to “some”. Where there are separate masses, the “some” can be read as referring to some of these distinct and countable units. At least for me, that implicit countification licenses the “some”.
This struck Barbara Partee as hitting on a crucial distinction, and I agree, though neither of us has any idea how it can be fitted into any existing syntactic or semantic theory. What’s going on is certainly not countification in the usual sense, conversion of M to C, since seawater in (5) is still clearly a (SG) M noun (note the subject-verb agreement). So the “countification” would seem to have to do with the referent of the noun in particular contexts, with the extent to which we can “parcel up” the referent in those contexts — what you might label “collectivization” rather than “countification”.
Leave a Reply