Arnold Zwicky's Blog

In a comment on my “Quoting modestly” posting, “phidauex” wrote:

The incongruity of seeing all those “fucks” in an otherwise fluffy obit-piece is where all the humor stems.

Jan Freeman queried “where all the humor stems”, noting that she couldn’t say that without making it “stems from”, and I agreed, adding that phidauex’s version looked like an example of P omission where the P is selected by the verb (stem, in this case), or “P absorption”, as I’ve sometimes called it. The idea is that some verbs so strongly select particular Ps (stem from, adhere to, etc.) that people sometimes feel free to omit the P because it can be supplied from the context. The result is non-standard but not at all uncommon.

Now to explain where Jan Freeman’s and my judgments come from.

(Note: essentially none of this is original with me. I’m talking about very well-known facts, though (for pedagogical purposes) not always in the most common terminology.)

First thing. Some Vs take direct objects, with no element marking their objecthood; other Vs take oblique objects, with a P marking the object. (Some Vs allow both possibilities.) Obl-obj Vs range from, at one end, those occurring with a variety of Ps in some semantic domain to, at the other end, very choosy Vs, with a strongly selected P as marker, so that dictionaries will treat them (rather misleadingly) as “phrasal verbs” and list them as units. That gives us adhere to and stem from. (These associations of V and P aren’t absolute — adhere on(to) and stem out of are possible, and do occur — but they’re strong.)

More important, obl-obj Vs for the most part don’t have dir-obj counterparts:

*The chewing gum adhered my shoe.

*Our problems stem your intransigence. [with stem ‘stem from’]

As far as I can see, this is a fact about syntax, not semantics. Indeed, there’s a fairly brisk trade in supplying dir-obj counterparts to earlier obl-obj Vs, via “transitivizing P-drop” (as in agree a solution ‘agree on / about / over a solution’); and obl-obj counterparts to earlier dir-obj Vs, via “intransitivizing P-addition” (as in glimpse at the scene) — some discussion here.

Second thing. English has a variety of “extraction constructions”, in which some clause-initial element (call it In) is linked to some missing constituent within the clause; in metaphorical language, the clause-internal constituent is “extracted” from the clause and “fronted” to initial position. (In what follows, I’ll focus entirely on extractions of non-subjects, mostly objects of various kinds.)

Sometimes the extraction is straightforward, as in

Beans I hate ___.

where the underlined expression is In and the empty underlining marks the linked “extraction site”. Sometimes the extracted material appears in pronominal form, in particular as a WH-pronominal:

— as an indefinite WH-pronominal in an interrogative construction (Who did you see ___?  I know who you saw ___)

— as a definite WH-pronominal in a relative construction (the person who you saw ___)

Third thing. An extracted WH-pronominal has to match the extraction site, in the sense that the pronominal has to satisfy the requirements on an expression at the extraction site. So there’s an intimate relationship between properties (syntactic category and syntactic function) of an expression at the extraction site and the particular extracted WH-expression. (The wider topic here goes under the name “filler-gap” relationships; the extracted material is the “filler”, the extraction site the “gap”.)

In simpler cases, this relationship is uncomplicated. Human objects (dir or obl) correspond to who (or whom):

[Q] Who did you see ___?  [Rel] the person who you saw ___

[Q] Who did you talk to ___?  [Rel] the person who you talked to ___

Non-human objects correspond to what in interrogatives, which (or the complementizer that) in relatives:

[Q] What did you see ___?  [Rel] the event which you saw ___

[Q] What did you come across ___?  [Rel] the object which you came across ___

Time adverbials correspond to when:

[Q] When did you go to Rome ___?  [Rel] that time when you went to Rome ___

And location adverbials correspond to where:

[Q] Where did you put the frammis ___?  [Rel] the place where you put the frammis ___

(There are many more details, but this is the basic picture.)

Putting these things together, we predict that Vs like adhere and stem will have their selected P preserved when the object of that P is extracted, in interrogatives:

[Q] What did it adhere to ___?  I don’t know what it adhered to ___

[Q] What does this stem from ___?  I don’t know what this stems from ___

and in relatives:

[Rel] the thing which it adhered to ___

[Rel] the source which this stems from ___

This is standard English. The P-less counterparts (with P-absorption) are attested but non-standard:

[Q] I don’t know what it adhered ___  [Rel] the thing which it adhered ___

[Q] I don’t know what this stems ___  [Rel] the thing which this stems ___

And now the complications. So far so good, but there are further wrinkles. In particular, objects of P that denote locations can have where rather than what or which as the extracted In:

[Q] Where did it adhere to ___?  [Rel] the place where it adhered to ___

[Q] Where does this stem from ___?  [Rel] the place where this stems from ___

At this point, adhere and stem part company. Ps denoting simple location (like at, in, on, and in some contexts to) can usually be omitted with where, while Ps denoting source (in particular, from) cannot:

[Q] Where did it adhere ___?  I don’t know where it adhered ___  [Rel] the place where it adhered ___

(X) [Q] *Where does this stem ___?  *I don’t know where this stems ___ [Rel] *the place where this stems ___

There’s a huge iceberg of facts below the tip I’ve explored here. But this is enough to get us to why Jan Freeman and I queried where all the humor stems.

The crucial poiont is that P-absorption interacts with the what/which vs. where choice and with V choice. One result is that even people who wouldn’t absorb the P with extracted what/which in non-standard examples like

[Q] *What does this stem ___?  *I don’t know what this stems ___

[Rel] *the source which this stems ___

might still accept, or even prefer, P-absorption with extracted where and the verb stem, at least in interrogatives; these people accept the asterisked Q examples in (X) above. (Relative examples like the one in (X) might not follow the interrogative examples.)

That is, some people are moving some Vs (like stem) into the class of Vs (like adhere) that can sponsor P-absorption with extracted where.

Notice how many factors seem to be at work here: (a) the V involved; (b) the obl-marking P involved; (c) extraction of what/which vs. where; and possibly (d) extraction in interrogative clauses vs. relative clauses. (In fact, there may be people who distinguish the facts for main-clause interrogatives from those for embedded interrogatives, or even for different embedded interrogative constructions, of which there are several.)

This is a situation in which just collecting some people’s judgments on a few examples is pointless. except to get some initial sense of the envelope of variation — which Vs, which Ps, which extracted WH-elements, in which WH-constructions. Figuring things out is a huge project, requiring multifactorial analysis of both naturalistic data and carefully controlled judgment studies. And I haven’t even mentioned social, contextual, and pragmatic factors, having to do with who says what to whom in which contexts for which purposes.

But there’s clearly variation out there.

This entry was posted on November 15, 2011 at 11:48 am and is filed under Syntax. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.