Notes on Question-Answer Clauses

Background material to: Caponigro, Ivano and Kathryn Davidson. 2011. Ask, and tell as well: Question-Answer Clauses in American Sign Language. Natural Language Semantics. DOI 10.1007/s11050-011-9071-0.

These notes are intended as an introduction to the Caponigro and Davidson reading on question-answer clauses. As usual, I'm zooming out from the topic of immediate concern to give some general and cross-linguistic background.

Fun fact to know and tell: Kate Davidson was a linguistics major at Penn in the 2000's or so. She went on to grad school in linguistics and got interested in sign linguistics, and she's now a leading researcher on the formal semantics and pragmatics of sign languages. She teaches at Harvard and directs the Meaning and Modality Laboratory there.

Focus

Some of the readings for the last few classes have introduced the notion of focus, and these notes will go into a bit more detail on this concept. An important thing to understand is that one and the same sentence can have a different focus depending on the preceding discourse. This is most clearly illustrated when the preceding discourse consists of a question. Notice that although the answers to the questions in (1)-(3) all consist of the same string, the part of the answer that is in focus is different in each case. This is clear from the fact that a short answer to each of the questions would consist of just the focused phrase. (Notice incidentally, that the long answers are pronounced exactly the same, with the same intonation and stress, even though the focus differs. Thus, intonation is not a completely reliable clue to identifying focus, at least in English.)

Here and throughout, the focus is highlighted in boldface. Stress is indicated by underlining. Note that only part of the focus is audibly marked by stress. This is typical of spoken languages. In sign languages, it's more typical for any intonation to mark a relevant phrase in its entirety.

(1) a. Question: What will the kids eat?
b. Answer: They will eat peanut butter and jelly sandwiches.
(2) a. Question: What will the kids do?
b. Answer: They will eat peanut butter and jelly sandwiches.
(3) c. Question: What will happen?
b. Answer: The kids will eat peanut butter and jelly sandwiches.

In English, the focus doesn't have to be the final constituent. This is what Hall et al. mean when they say that English is a [+plastic] language. We can see this clearly in connection with the answers in (4). As usual, "#" marks a sentence that is syntactically well-formed, but infelicitous in context. (4b.ii) is infelicitous because the stressed syllable is not part of the constituent that is in focus given the preceding discourse.

(4) a. Question: What will the kids eat for lunch?
b. i. Answer: ✓ They will eat peanut butter and jelly sandwiches for lunch.
ii. # They will eat peanut butter and jelly sandwiches for lunch.

Two types of plasticity

Since English is [+plastic], it doesn't actually need rules like topicalization to move constituents "out of the way" to allow the focus to occupy a particular position the way that ASL does. As a matter of fact, howver, English does have topicalization, as we saw last week, just like ASL. So it allows a variant of (4) as in (5).

(5) a. Question: What will the kids eat for lunch?
b. Answer: ✓ For lunch, they will eat peanut butter and jelly sandwiches.

The fact that English allows both (4) and (5) suggests that it would be helpful to distinguish two types of plasticity. We could call these stress plasticity and word order plasticity. Stress plasticity is the same as the plasticity mentioned in Hill et al. and concerns whether a language requires stress to fall on a particular position in the sentence. English doesn't have this requirement, but ASL does, and so they have the values [+] and [-], respectively, for this feature. [+ Word order plastic ] languages would be defined as ones with rules that move constituents to positions other than the basic position (whether to the left, as with topicalization, or to the right, which is possible in some languages). ASL would be highly [+] with regard to this property, but English is also [+], though not to the same degree.

One could imagine a language that is [+ stress plastic], but [- word order plastic]. Such a language would the inverse of ASL - a language with completely rigid word order, where focus is signalled exclusively by the position of stress in the sentence. Such a language seems possible as a means of communication, but I am not sure such a language actually exists.

There's a final combination - the language type with [-] for both properties. This type seems like it should be impossible or at least highly disfavored, since it would allow no way of overtly marking focus in question-answer pairs like those in (6) vs. (7). (The question in (6) has no wh- movement because by assumption, there's no word order plasticity.)

(6) a. Question: The kids will eat what?
b. Answer: The kids will eat peanut butter and jelly sandwiches.
(7) a. Question: Who will eat peanut butter and jelly sandwiches?
b. Answer: The kids will eat peanut butter and jelly sandwiches.

By assumption, the language is [-stress plastic], and so the stress will always fall on the same position, whether that's the end of the sentence or the beginning or somewhere else. But the stress would have to move in order to differentiate between (6b) and (7b). In a [+ word order plastic] language, the workaround is to move the constituents of the sentence around, but that option isn't available, again by assumption. So it seems like this type of language shouldn't exist.

Clefting

It clefts

Paradoxically, this type is possible, though, because human languages more generally have a strategy called clefting, which allows the doubly [-plastic] languages under discussion to get out of their bind. Clefting splits simple sentences into two parts, both of which are consistent with the two [-plastic] values and which combine to express the desired focus marking via stress. A well-known example of this language type is French. Stress falls on the last constituent. So the counterpart to (7a) is not a problem. (I've left out the peanut butter and jelly, because I'm not sure whether French kids would eat peanut butter and jelly sandwiches.)

(8) a. Question: Les enfants ont mangé quoi?
the kids have eaten what
'What did the kids eat?'
  b. Answer: Les enfants ont mangé les tartines.
the kids have eaten the sandwiches
'The kids ate the sandwiches.'

But now what the counterpart to to (7b)? Neither answer in (9) is possible because (9b.i) one requires stress plasticity, and (9b.ii) requires word order plasticity (leftward movement of les tartines and rightward movement of les enfants).

(9) a. Question:   Qui a mangé les tartines?
who has eaten the sandwiches
'Who ate the sandwiches?'
   b. Answer: i. * Les enfants ont mangé les tartines.
the kids have eaten the sandwiches
ii. * Les tartines, ont mangé les enfants.
Intended meaning for both answers: The kids ate the sandwiches.'

So what French does is split up the answer into two parts: (1) a very simple sentence whose only real content consists of the intended focus, and (2) a relative clause which contains the rest of the content. This general strategy is called clefting, and it yields the result in (9).

(9) a. Ce sont les enfants qui ont mangé les tartines.
it are the kids who have eaten the sandwiches
'It is (lit. 'are') the kids that ate the sandwiches.'

Each of the two parts counts as a sentence with its own final position. So now les enfants 'the kids' can be stressed as required by the discourse context and without violating the [-stress plastic] property.

The clefting option is also possible for the unproblematic sentence in (8b), and it yields the clefted variant in (10).

(10) a. Ce sont les tartines que les enfants ont mangé.
it are the sandwiches which the kids have eaten
'It is (lit. are) the sandwiches that the kids ate.'

As is evident from the translations of the French examples, the clefting strategy is available in English as well. Because the first part of the cleft construction is introduced by it, this particular type of cleft is called an it cleft (and the term would be used for the French examples also, even though ce really corresponds more exactly to 'this' than to 'it').

Wh- clefts

In addition to it clefts, English also has wh- clefts. (Since English, unlike French, is [+ plastic] for both types of plasticity, it doesn't actually need either type of cleft, but for some reason it has two cleft constructions, while French only has the one). Like an it cleft, a wh- cleft contains two parts, but the order of focus and non-focus material is reversed. As (11) illustrates, the first part of the cleft is a wh- clause containing the non-focus material, and the second part contains the focus. Both parts are connected by a form of the copula be.

(11) a. Question: What will the kids eat for lunch?
b. Answer: ✓ What the kids will eat for lunch is peanut butter and jelly sandwiches.

Wh- clefts expressing focus on the subject should be possible, but for some unknown reason they sound awkward, at least in English.

(12) a. Question: Who will eat peanut butter and jelly sandwiches for lunch?
b. Answer: ? Who will eat peanut butter and jelly sandwiches for lunch is the kids.

At first glance, the English wh- cleft construction looks identical to the ASL Question-Answer Clauses discussed by Caponigro and Davidson. However, as Caponigro and Davidson argue, the two constructions, though equivalent from the point of expressing focus, are not identical. The first part of the wh- cleft in English is a headless relative clause (see immediately below for more discussion). But as Caponigro and Davidson point out, ASL doesn't have headless relative clauses so it isn't possible to extend the English analysis to ASL.

Background reading

Headed vs. headless relative clauses

The Caponigro and Davidson reading mentions and distinguishes two types of relative clauses:s headed relative clauses and headless relative clauses (also known as free relative clauses). A headed relative clause modifies a noun that is explicitly present in the discourse or text. English allows three kinds: Only the wh- marked kind is relevant in what follows, but all three are illustrated in (13) for completeness. In (13), the head noun that is being modified is in italics. It is followed by the marker that introduces the relative clauses (and counts as part of it). The rest of the relative clause then follows. It contains a gap (that is, the clause wouldn't be able to stand on its own). We indicate the gap with "___" in the same way that we did last week for topicalization.

Type Head Marker Rest of relative clause
(13) a. wh- the doctor who(m) we trust ___ most nowadays
b. that the doctor that we trust ___ most nowadays
c. zero the doctor ø we trust ___ most nowadays

In contrast to headed relative clauses, headless relative clauses don't modify an overt head noun. Nevertheless, there's an understood modifiee. Thus, a headless relative clause like (14a) is semantically roughly equivalent to the headed relative clause in (14b).

  Type Head Marker Rest of relative clause
(14) a. Headless relative: ø who(m) we trust ___ most nowadays
b. Roughly synonymous headed relative clause: the person who(m) we trust ___ most nowadays

Morphosyntactically, headless relative clauses as in (14a) function as noun phrases. In other words, they play the same role as headed relative clauses together with their head and any associated material like determiners, adjectives, and so on. This is illustrated in (15) and (16).

Object of preposition:
(15) a. They arrived with the equipment which we needed ___ most today.
b. They arrived with ø what we needed ___ most today.
Subject of sentence:
(16) a. The equipment which we need ___ most is due to arrive tomorrow.
b. ø What we need ___ most is due to arrive tomorrow.

Two further points are worth noting in connection with the two types of relative clauses that we've just described. First, although they have a different function, their form is identical to indirect questions as in (17). (See the background material on questions for more details.)

(17)     Indirect question: I'm not sure what we need ___ most today.

Second, and more relevant to the issue of Question-Answer Clauses, headless relative clauses permit a narrower range of wh- pronouns than either their headed counterparts or indirect questions, as Caponigro and Davidson mention. As far as I know, it is not known why this is so.

(18) a. ✓ The house which they want to buy is a fixer-upper.
b. * ø Which they want to buy is a fixer-upper. (cf. ✓ 'What they want to buy ...')

Since Question-Answer Clauses in ASL do not exhibit this restriction, they are not plausibly analyzed as headless relative clauses (there have been such analyses in the sign language literature). In fact, there are no wh- constructions at all that exhibit this restriction in ASL, leading Caponigro and Davidson to conclude that ASL doesn't even have headless relative clauses. (So an analysis of Question-Answer Clauses as headless relative clauses is ruled out even more strongly.)