In the early 1970s, much research was done on multiple-wh (MWh) interrogatives. Some of this research, notably Wachowicz (1974), concerned coordinated-wh (CWh) interrogatives, in which two or more wh items are coordinated, such as the coordination of wh-adjuncts in Where and when did this happen? Browne (1972) observed that some CWh interrogatives in Serbo-Croatian and Czech were unusual, in allowing wh-complements to be coordinated with wh-adjuncts, as in What and when did he give you—ungrammatical in English, but grammatical in Serbo-Croatian. In a response to Browne, Bechhofer (1976) noted that such coordinations were ungrammatical in Turkish, but that coordination of wh complements for different subcategorizations of a verb was possible, e.g. Where and with whom do you live? By and large, however, these findings were ignored or forgotten in the literature on interrogatives, as the focus shifted to the crosslinguistic syntax and semantics of MWh interrogatives.
In recent years, however, there has been a resurgence of interest in CWh interrogatives, particularly those in Slavic languages. Comorovski (1989/1996) noted that coordination of wh-complements filling different thematic roles was possible in Romanian, as in Who and what told you? (meaning, "Who told you something and what did they tell?")— again, ungrammatical in English, but grammatical in Romanian. This kind of complement-complement CWh, as well as the adjunct-complement CWh noted earlier, has now been observed in Russian (Kazenin 2002), Hungarian (Lipták 2003), Chinese (Zhang 2006), and to a limited extent in English (Whitman 2002), in addition to the above-mentioned languages. However, answers to basic questions about CWh interrogatives remain to be established.
There is the typological question of what kinds of languages CWh interrogatives can occur in, particularly the unusual adjunct-complement ones exemplified by What and when did he give you? and complement-complement ones exemplified by Who and what told you? Comorovski (1996:139) states that "...such coordinations of wh-phrases appear to arise only in languages that allow multiple wh-fronting," but the Hungarian, Turkish, English, and Chinese data show that much crosslinguistic research is needed in assessing the truth of this proposition. In addition, there may be interesting typological generalizations to be made about which languages use both MWh and CWh constructions (or only one or the other), and how each is used. For example, as can be seen in the data collected here, several languages have only complement-complement MWh and adjunct-adjunct CWh, but no language in the collection thus far has the opposite: only adjunct-adjunct MWh and complement-complement CWh. Research is needed to find such languages or explain why they are unattested.
There is also the syntactic question of the structure of CWh interrogatives. One approach has been to analyze them as a kind of sluicing. That is, a CWh interrogative such as When and where did this happen? and the coordination of a wh-interrogative with a sluice in When did this happen, and where? are both underlyingly two conjoined interrogatives: When did this happen and where did this happen? The difference lies in which of the clauses (minus its wh part) is deleted/elided/ellipsed. Such analyses have been proposed by Browne (1972), Giannakidou & Merchant (1998), and Camacho (2003), and disputed by Bechhofer (1976), Kazenin (2002), Whitman (2002), and Lipták (2003).
Another hypothesis is that the wh items are coordinated in the same way as anything else is coordinated. If this is the case, then one is faced with the question of what category the coordinated wh-items have, especially in cases of adjunct-complement coordination, where the wh-items presumably have different syntactic categories. Is simply being wh-items enough to allow two words or phrases to be coordinated, as proposed by Bechhofer (1976), Bolinger (1978), Grimshaw (1978), Comorovski (1996), and Camacho (2003)?
Furthermore, if wh coordination is like other coordination, it raises questions for the interaction of coordination with wh-movement (for theories that have wh-movement). To wit: Does the coordination occur before wh-movement, as proposed by Grimshaw (1978)? Or does it occur after wh-movement, as proposed by Kazenin (2002), Lipták (2003), and Zhang (forthcoming)? Or does it vary according to language and kind of CWh?
The main semantic question regarding CWh questions is whether they call for a single-pair or pair-list answer. It has been claimed that CWh questions are a means of asking a multiple-wh question without the requirement of a pair-list answer that a MWh question carries (Kazenin 2002, Lipták 2003, Whitman 2004). To illustrate with a well-known contrast, consider the infelicity of (1), which seems to ask for a list of time-location pairs, implying that someone was born more than once, and the felicity of the CWh in (2), which is easily answered with a single time-location pair.
However, it is not clear that this distinction, noted for English (Wachowicz 1974, among others), Hungarian (Liptak 2003), and Russian (Kazenin 2002), holds for all languages. In light of the existing research deriving PL answers for MWh questions in their formal semantics, more thorough and systematic crosslinguistic examination of this claim is needed.
The Coordinated-Wh Project began as an attempt to test the above-mentioned hypothesis regarding MWh and CWh interroagatives and pair-list answers. Its aim is now more general: to collect a sampling of MWh and CWh interrogatives in a diverse sample of languages, along with semantic judgments on whether these interrogatives require pair-list answers. It is hoped that the collected data will serve as a resource to linguists interested in CWh interrogatives, providing data from a wider range of languages than in typically seen in individual publications, thus allowing typological generalizations to be made, and syntactic and semantic hypotheses to be tested. To allow as close a comparison as possible, the same questions (with occasional, minor variations) are used for each language. These questions cover the three broad types of MWh/CWh interrogatives already laid out: complement/complement, adjunct/complement (or complement/adjunct, if you wish), and adjunct-adjunct. In addition, where published MWh/CWh data is already available, it is included as a supplement to the data gathered in the course of the current project, so that this collection can also serve as a clearinghouse for all available data on CWh interrogatives.
Obtaining translations for the set of questions is tricky on several counts. One problem is that some questions that are grammatical in English, such as When did she sing where?, may be ungrammatical in the language under investigation. The translator, therefore, may propose an alternative translation to convey the desired meaning, possibly something parallel to "When did Mary sing, and where?" This is a reasonable thing to do in ordinary translation tasks. The trouble here is that the proposed alternative is not an MWh question, but a coordination of a full question (When did Mary sing?) with an elided question/sluice (where [did Mary sing]?). Though it is useful to know that this is the most natural way to phrase this question in the target language, we still need a translation without any coordination, to see whether MWh is indeed ungrammatical.
Furthermore, in trying to obtain such a translation, care needs to be taken to ensure that the translation is ungrammatical because of being phrased as an MWh or a CWh, not because of some other reason the researcher made was unaware of. To illustrate: a Tagalog speaker translated Who read what? into Tagalog for me. I then tried to construct a translation of What did who read? by swapping the Tagalog words for what and who, and the result was ungrammatical. Later, however, I learned that in Tagalog, the first position in a question is the focus position. By swapping the focused who with the nonfocused what, I was beginning the sentence with a nonfocused element, and placing the focused one where it wasn't allowed, so naturally the question was ungrammatical. But maybe if I had put the focus marker on the what, the nonfocus marker on who, and changed the verb's morphology appropriately, the question would have been grammatical. And if not, the ungrammaticality could be more confidently attributed to a superiority violation in an MWh question.
A second problem is obtaining translations for questions that are ungrammatical in English. For instance, even if something analogous to What and who read? is grammatical in the target language, the translator may not know what to do with it if they cannot parse it in English. Or they may give it the undesired but only available parse ("What read something, and who read something?"), translate that, and assign a judgment accordingly. In these cases, I try to construct my own translation based on already-obtained translations and submit it for judgment (though this has its own problems, as noted above).
Third, the translator should realize that grammaticality judgments should be given assuming that these are regular, answer-seeking questions, as opposed to echo questions or REF questions. An echo question is one expressing surprise in response to something already said. For example, if someone says, "John came to work late yesterday," and this is so surprising to you that you say, "WHO came to work late WHEN?", this is an echo MWh question. A REF question is one asking for a repeat of an imperfectly heard or understood utterance. For example, someone says, "So he finally did it!", assuming you know who they're talking about and what it was that the person did. But you have no idea who or what they're talking about, so you ask, "Who finally did what?" This is a REF question. It is important for informants to know we're not interested in echo ro REF questions, because sometimes an utterance that is grammatical as an echo or REF question is ungrammatical or at least questionable as an ordinary question. For example, if someone asks you, "Well? What did he say?" and you don't know who he refers to, you might ask the REF question, "What did WHO say?" But if this question were asked in a situation where several people made different statements and you wanted to match up speakers with statements, it would be a little strange, with Who said what? being preferred.
Early on in the project, semantic judgments were sought at the same time as the translations and syntactic judgments. After doing the 24 translations, the translator would simply be asked, for each interrogative judged to be grammatical, if it required a PL answer. Furthermore, it was assumed that whether required or not, a PL answer was always an acceptable response. There were several problems with this approach, however, First, by the end of creating and judging the 24 translations, the translator might well be bored or fatigued. Second, it turns out that some interrogatives that seem to require PL answers can nonetheless be given an SP answer in the right context, while others must be given a PL answer regardless of context. One speaker might imagine an SP-friendly context for a question more easily than another, resulting in apparent differences in judgment.
Therefore, the current procedure is different. To begin with, semantic judgments are obtained from a different speaker than the one who provides the translations&emdash;or if from the same speaker, after an interval of several days following the translation task. Since a possibly different person is giving semantic judgments, this person is presented with ALL the translations, not just those judged to be grammatical. For each group of questions, two contexts are presented: an SP-friendly one, and a PL-friendly one. The translator's task is to say whether each question could be asked in one or the other or both contexts. Ideally, those judged ungrammatical will prove unacceptable in either context, though this doesn't always happen.
Some of the earlier-gathered translations are still tagged with labels such as "SP OK" with no context given. These are awaiting the availability of a native speaker to provide semantic judgments in conjunction with the SP- and PL-friendly contexts mentioned above. More recently obtained translations are presented with the two contexts that were given to the translator. These are shown with each set of questions, rather than in the background material here, since these contexts have evolved in response to informants' questions about them, and therefore are not completely alike for each set of questions.
The questions that each informant was asked to translate are as follows (with occasional slight variation):
|
|
| Chinese | Czech | English | Estonian |
| German | Greek | Hebrew | Hindi |
| Hungarian | Japanese | Korean | Macedonian |
| Russian | Spanish | Tagalog | Vietnamese |
If you cite data from this source that was not originally published elsewhere, please list this source in your references as follows:
[Name of person who gathered the data for the specific language], in Whitman, Neal, ed. The Coordinated-Wh Project. Version of [date of latest page update]. Accessed [date of access]. <http://literalmindedlinguistics.com/Coordinated_Wh_Project.html>
Data from other sources that has been included here should be cited with the original publication data (available in the Bibliography).
If you would like to help expand this data collection, I welcome your additions.
If you are a native speaker of a language that is not yet represented here, I would be delighted to add your translations and judgments to the collection. In fact, I would be delighted to do so even if your language is already represented, as intuitions are subtle, and the more speakers who contribute their judgments, the clearer a picture we'll get. Contact me and we'll arrange the details.
If you have collected MWh/CWh translations and judgments from a native speaker of a language that ought to be in this collection, and would like them to be incorporated, I will be happy to do so, giving credit to both you and your informant.
Thank you to all the native speaker informants who have provided translations and syntactic and semantic judgments for this project:
Ben Chudnovsky, Ilija Doneski, Anna Feldman, Calixto Gonzales, Betya Goykhman, Jirka Hana, Hyeon-Seok Kang, Soyoung Kang, Yusuke Kubota, Sun-Hee Lee, Ilse Lehiste, Dmitry Levinson,
Anikó Lipták, Xiaofei Lu, Arantxa Martín-Lozano, Detmar Meurers, Bettina Migge, Mineharu "JJ" Nakayama, Roberto Orci, Panayiotis Pappas, Mike Puchovich, Hongqi Rouzer, Jane Rubin-Kurtzmann, Le Nhan Thanh,
Giorgos Tserdanelis, Shravan Vasishth, Amanda Whitman, and Niina Zhang.