Tag questions using wh-words are quite common in American English. The usage can be divided roughly into two categories: a highly productive form which may involve any wh-word, and a frozen form using "or what", which appears to be idiomatic. Following Merchant's analysis, we can see that the productive form matches the characteristics of other constructions that have been classified as sluicing, with one exception. The form can therefore tentatively be considered a type of sluicing. I will only briefly address the non-productive form, after the main discussion.
The most productive form of wh-tag is a genuine question, reiterating the constituent being queried. The intonation is typical for a question, high overall, rising at the end of the main clause, with stress on the queried constituent. The tag phrase receives secondary stress, and is somewhat lower in pitch. Examples using various wh-words are in (1). Following Merchant, I indicate stress in CAPS and deaccented text in italics. [1]
| (1) | a. | Are you planning to wear a SUIT, or what? | |
| b. | Is he seeing LIZZIE, or who? | ||
| c. | Are you arriving TOMORROW, or when? | ||
| d. | Are you arriving TOMORROW, or when? | ||
| e. | Are you traveling by TRAIN, or how? | ||
| f. | ? | Did you leave because of the NOISE, or why? |
Various characteristics of sluicing proposed by Merchant are addressed in the following sections.
Sluicing is a phenomenon in which a wh-question is reduced to just the wh-phrase, as in (2).[2]
| (2) | a. | Jack bought SOMETHING, but I don't know what. |
| b. | SOMEONE called, but I can't tell you who. | |
| c. | BETH was there, but you'll never guess who else. | |
| (Merchant) | ||
Merchant analyzes sluicing as IP deletion.[3] The sentences in (2) are semantically equivalent to those in (3), in which the second IP is deaccented rather than deleted.
| (3) | a. | Jack bought SOMETHING, but I don't know what Jack bought. |
| b. | SOMEONE called, but I can't tell you who called. | |
| c. | BETH was there, but you'll never guess who else was there. |
In an analysis similar to that used for VP-deletion, Merchant proposes the following Focus Condition on IP-Ellipsis (my paraphrase):
| (4) | An IP can be deleted if it entails and is entailed by its antecedent. |
It can be seen that the sentences in (1) are semantically equivalent to those in (5).
| (5) | a. | Are you planning to wear a SUIT, or what are you wearing? | |
| b. | Is he seeing LIZZIE, or who is he seeing? | ||
| c. | Are you arriving TOMORROW, or when are you arriving? | ||
| d. | Are the Terps playing at HOME, or where are the Terps playing? | ||
| e. | Are you traveling by TRAIN, or how are you traveling? | ||
| f. | ? | Did you leave because of the NOISE, or why did you leave? |
The relationship between the elided sentences in (1) and the unelided sentences in (5) is apparently similar to the relationship between (2) and (3). Thus, wh-tag questions satisfy the Focus Condition on IP-Ellipsis given in (4).
Merchant quotes Ross (1969):
| (6) | All and only predicates that s-select questions and c-select CPs allow sluiced wh‑phrases. |
This is the canonical use of sluicing, demonstrated in (2).
At first glance, this is a problem, since none of the wh-tag questions are licensed by such a predicate. However, Merchant implicitly acknowledges that these are not the only constructions in which sluicing can occur. He uses the examples in (7) and (8).
| (7) | Some of these problems are solvable, but which problems is not obvious. |
| (8) | A: Lois was talking (to someone). B: Really? Who to? |
In (7), the sluiced element which problems is in the subject position. In (8), the sluice is in a matrix clause. So, rather than (6) being a requirement in all cases, we rather interpret it as applying only when the sluice is in a subordinate clause as object of a predicate.
In the sentences in (1), it makes sense to analyzed the sluiced element as a matrix clause, coordinated with the initial clause using the conjunction or. Thus, these sentences are eligible for consideration as sluicing in the same manner as (8).
| (9) | The agreement marking on a verb with a sluiced subject is typical of the marking with any CP subject, and independent of the marking of the wh-phrase itself. |
This principle is demonstrated in (7).
Since the construction in (1) does not treat the sluiced element as a subject, this characteristic is irrelevant.
| (10) | In case-marked languages, the wh-word agrees in case with an antecedent NP, if any. |
Merchant points out that the only case-marked wh-word in English is whose, as in (11).
| (11) | Somebody's car is parked on the lawn, but we don't know [whose/*who]. |
He correctly remarks that it is not clear exactly what is going on in these cases, since there may also be NP ellipsis, as in [whose car]. In any case, wh-tag questions behave the same as the sluicing example, as in (12).
| (12) | Will you be taking LISA'S car, or whose? |
| (13) | The positions available to a sluiced wh-remnant are always the same as the positions available to full interrogative CPs, not the positions available to non-moved wh-phrases. |
| (Merchant) |
Again, we analyze the sluiced position in (1) as a coordinated clause. This position is readily available to a full interrogative CP, as the sentences in (5) make apparent. Thus, wh-tag questions satisfy (13).
Since this argument only applies in German, it is not of use in our discussion. If it turns out that German also has wh-tag questions, we could return to this point.
Merchant gives an extensive discussion on the semantics of IP-ellipsis. I will highlight only one key point that is relevant to this discussion.
| (14) | An elided IP has a feature which can only be checked by a [+wh, +Q] C head (i.e. a wh‑word). |
The sentences in (1) have such a head available.
Note that other forms of tag questions are possible, as in (15).
| (15) | You're not leaving already, are you? |
Without a wh-word head, these are not categorized as sluicing.
| (16) | In sluicing, no non-operator material may appear in COMP. |
| (Merchant) |
In other words, only the wh-phrase itself can remain after ellipsis. For example, (17).
| (17) | A: Max has invited someone. B: Really? Who (*has)? |
The same characteristic is exhibited with wh-tag questions, as in (18).
| (18) | a. | ? | Have you invited your girlfriend, or who (*have)? |
| b. | Did you invite your girlfriend, or who (*did)? |
The most direct parallel, (18a), is marginal even in the acceptable form, for reasons that I would take to be pragmatic.[4] I included (18b) as a case where the simple tag question is perfect, but with the auxiliary is horrible.
In English, an inverted preposition can remain as part of the wh-phrase, as in (19).
| (19) | a. | Lois was talking (to someone), but I don't know who to. |
| b. | A: Lois was talking (to someone). B: Really? Who to? |
Notably, this is not true for wh-tag questions, as in (20).
| (20) | Was Lois talking to her mother, or who (*to)? |
This is the first piece of evidence we have seen that distinguishes wh-tag questions from known sluicing examples. Instead, they class with non-sluiced interrogatives, as in (21).
| (21) | a. | * | I don't know who to Lois was talking. |
| b. | * | Who to was Lois talking? |
Note that all five examples can pied-pipe the preposition in uninverted position, as in (22).
| (22) | a. | ? | Lois was talking (to someone), but I don't know to who. |
| b. | A: Lois was talking (to someone). B: Really? To who? | ||
| c. | ? | Was Lois talking to her mother, or to who? | |
| d. | ? | I don't know to whom Lois was talking. | |
| e. | ? | To whom was Lois talking? |
Oddly enough, the non-sluiced interrogatives require the whom form, at least in my idiolect. If we accept that whom is an archaic or formal term, this may indicate that the to whom form is unnatural. In this regard, the wh-tag question is like the sluicing examples in preferring to who.
It is not clear how much weight to give to this evidence. It may cast doubt on the analysis of wh-tag questions as examples of sluicing, but does not seem strong enough in itself to emphatically dismiss the analysis.
This phenomenon only applies to languages other than English. In some languages, a complementizer can co-occur with a question word in non-sluiced interrogatives, but not in sluiced constructions. I do not have data regarding wh-tag questions in those languages.
Sluicing has the remarkable ability to repair island violations. Unfortunately, it is not easy to construct pragmatically plausible examples of potential island violations using wh-tag questions. However, in the examples that I was able to construct, the island violations seem to disappear.[5] [6]
| (23) | Relative clause | ||
| Are you looking for someone who speaks MANDARIN, or which language? | |||
| (24) | Adjuncts | ||
| a. | Would you like it if I wear the BLUE suit, or which? | ||
| b. | Were you upset because BRUNO insulted you, or who? | ||
| (25) | Complement to nouns | ||
| Did the administration issue a statement that it is willing to meet with the YOUNG ANARCHISTS, or which group? | |||
| (26) | Sentential subject | ||
| (27) | Embedded question | ||
| Did Sandy say she needs to know which students had solved the TRAVELING SALESMAN PROBLEM, or which (problem)? | |||
| (28) | Coordinate structure constraint | ||
| a. | ? | Did Bob and GLORIA go to the ballgame, or who? | |
| b. | ? | Did you have pizza and BEER for supper, or what? | |
| (29) | COMP-trace effects | ||
| a. | Did you say that you are leaving TOMORROW, or when? | ||
| b. | Did Barbara say that she was going to the store, or where? | ||
| (30) | Left-branch (attributive adjective case) | ||
| (31) | Derived position islands (topicalizations, subjects) |
||
This is perhaps the strongest evidence so far that wh-tag questions should be classed as a form of sluicing. At the very least, whatever it is that allows sluicing to repair island violations appears to apply here, as well.
| (32) | The sluiced wh-phrase must bear the case that its correlate bears. |
| (Merchant) |
This only applies to case-marked languages.
| (33) | A language L will allow preposition stranding under sluicing iff L allows preposition stranding under regular wh-movement. |
Because this feature applies to both sluicing and non-sluiced interrogatives, it does not help us distinguish between the categories for wh-tag questions.
Wh-tag questions behave like known examples of sluicing in every manner except the inability to hold onto a preposition in inverted position in the wh-phrase. They even allow the repair of strong island violations, a distinctive characteristic of sluicing. We can therefore tentatively conclude that wh-tag questions are, indeed, a form of sluicing.
Tag-questions with queried constituents are largely productive. However, in other similar constructions, or what is the only applicable phrase.
For instance, or what can express exasperation or impatience, as in (34).
| (34) | a. | Are you coming, or what? |
| b. | Have you decided, or what? |
It can also express amazement or awe, similarly to an exclamative, as in (35).
| (35) | a. | Is she hot, or what? |
| b. | Am I smart or what? |
All of these sentences (34)-(35) share similar intonation, uniformly low.
One might suppose that these sentences class together, expressing a meaning such as "emphasis" or "emotional involvement". However, there are some problems with this analysis.
A key difference is that the sentences in (34) are not comparable with unelided sentences such as (36).
| (36) | a. | # | Are you coming, or what are you doing? |
| b. | # | Have you decided, or what have you done? |
The sentences in (35), however, can be treated as elided forms, comparable with the unelided sentences in (37).
| (37) | a. | ? | Is she hot, or what is she? |
| b. | ? | Am I smart or what am I? |
On the other hand, the speaker does not really expect an answer. The only felicitous answer is "Yes". In this the sentences in (35) behave like rhetorical questions.
In contrast, the sentences in (34) really do seem to be a demand for information. The listener is expected to respond as in (38) or (39).
| (38) | a. | Yes, I'm coming! |
| b. | Give me five more minutes. | |
| (39) | a. | Yes, I'll take chocolate. |
| b. | I'm still thinking. |
Finally sentences like (34) and sentences like (35) appear to have different distributions. Rhetorical questions like (35) most naturally occur with copular sentences. So, we can have (40a), but not so easily (40b).
| (40) | a. | Was he fast, or what? | |
| b. | ?? | Did he run fast, or what? |
Sentences like (34), expressing exasperation, apparently occur only in second-person progressives and perfects.
| (41) | a. | * | Will you hurry up, or what? |
| b. | * | Are you ready, or what? |
While different distributions do not in themselves prove that the classes are different, it provides additional evidence in that direction.
Wh-tag questions as query constituents appear to be highly productive with all possible wh‑words. These sentences match almost perfectly with the known characteristics of sluicing, and can therefore be considered a type of sluicing.
Two other forms are used in limited settings, and are restricted to the tag or what. One of these forms looks like sluicing, and has the semantics of a rhetorical question. The other form expresses exasperation, allows an answer, and does not look like sluicing.
This analysis suggests that or what may represent three distinct constructions, each with unique syntactic, semantic, and pragmatic characteristics.
Merchant, Jason. 2001. The syntax of silence: Sluicing, islands and the theory of ellipsis. Oxford: Oxford University Press.
Ross, John R. 1967. 'Guess Who', in R. Binnick, A. Davison, G. Green and J. Morgan (eds.), Papers from the 5th Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society, 252-86, quoted in Merchant.
[1] The judgments in this paper are all mine. I informally presented the data to other native English speakers, and was surprised to find that less than half of them were able to get sentences such as (1) at all. I will leave for a later discussion any speculation about the reasons for the varying judgments.
[2] There are also cases of sluicing which lack an overt antecedent, but these are not important to this discussion.
[3] Merchant argues for an analysis of ellipsis as PF deletion, rather than LF copying. The distinction is not crucial to this discussion.
[4] I can only speculate on what these reasons are. Offhand, it seems that the past perfect presupposes the intended action, as in, "Have you washed the dishes yet?" It is then infelicitous to indicate uncertainty about the intended action with the tag question.
[5] Following Merchant, I do not provide the paradigmatic examples to describe these islands.
[6] Alas, I did not have the opportunity to run these examples by other native speakers for their judgments.