Do there exist languages with wh-prepositions?

Question

I can imagine a language where instead of "what did you put a toy on?" one says something like "whon did you put a toy?". Do such languages exist?

Answer 1

German does have something like this:
_{(list of abbreviations see below)}

Wo-r-auf     hast du  ein Spielzeug gelegt?
where-ITF-on have you a   toy       put.PSTPTCP
Where did you put a toy on?

Wo-von   ernähren sich    Hasen?
where-of live.on  REFLPRN rabbits
What do rabbits live on?

Wo-nach     riecht das?
where-after smells that
What does this smell like?

Wo-durch      entsteht ein Tornado?
where-through develops a   tornado
Due to what does a tornado develop?

This is due to Pied piping: When wh-fronting an NP, you also have to move another element it goes with (like prepositions); this is often optional in English, but obligatory in German:

Sie hat sich    [mit  XYZ] getroffen
she has REFLPRN [with XYZ] meet.PSTPTCP
She met [with XYZ]

-->

[Mit  wem]     hat sie sich    __ getroffen?
[with who.DAT] has she REFLPRN __ meet.PSTPTCP
[With whom] did she meet __?

Leaving the preposition at its initial position after the NP it belongs to has moved to the front makes the sentence ungrammatical:

* [Wem    ] hat sie sich    mit  __ getroffen?
  [who.DAT] has she REFLPRN with __ meet.PSTPTCP
Who did she meet with?

The preposition must be moved to the front as well in order to make the wh-question grammatical. This is what is called pied piping.

When the questioned NP is not a person ("who"/"wer"), but an object ("what"/"was"):

Die Schüler  haben [mit XYZ]  experimentiert.
the students have  [with XYZ] experiment.PSTPTCP
The students experimented [with XYZ].

Du  hast ein Spielzeug [auf XYZ] gelegt.
you have a   toy       [on  XYZ] put.PSTPTCP
You put a toy [on XYZ].

you could colloquially say:

[Mit  was]       haben die Schüler  __ experimentiert?
[with what.DAT]  have  the students __ experiment.PSTPTCP
[What] did the students experiment with __?

[Auf was]  hast du  ein Spielzeug __ gelegt?  
[on  what] have you a   toy       __ put.PSTPTCP
[What] did you put a toy on __?

However, the "correct" form is turning the question word "was" into "wo" (literally "where", but it loses its local meaning and severes only as a basis to indicate the wh-word) and attaching the pied-piped preposition to it:

[Wo-mit]     haben die Schüler  __ experimentiert?
[where-with] have the  students __ experiment.PSTPTCP
What did the students experiment with __?

[Wo-r-auf]     hast du  ein Spielzeug __ gelegt?
[where-ITF-on] have you a   toy       __ put.PSTPTCP
What did you put a toy on __?

This results in a wh-word which is at the same time prepositional. It looks like "where" + preoposition, but as said, the "wo" has no local meaning in this context (that the "toy" example is about locality is only a coincidence; this is not indicated in the "wo" part), it behaves more like "w"+preposition where the "o" is only inserted for pronouncability, so I would guess this is exactly what you were looking for.

---

ITF interfix
PSTPTCP past participle verb form
REFLPRN reflexive pronoun
DAT dative case

Answer 2

My assumption from Turkish is that agglutinative languages have this property.

For example:

• Oyuncağı nereye koydun?

Word by word:

• TheToy whon youput

In Turkish "Ne" means What, but also is used as wh. Ne + rede means in Where, Ne + den means for What.

Answer 3

Hindi kind of has this feature:

किस (चीज़) पे खिलौना रखा?
kis (cIz) pe khilonA rakhA?
What did you put the toy on?

किसने किया?
kisne kiyA?
Who did it?

किस is the oblique form of the word क्या (kyA) meaning "what".

Answer 4

Preposition by definition means that it is a functional word put in front of another word (typically nominal in nature) that it modifies in some way.

The wh- type words are pronominal (or more precisely shifter-like) in their nature, thus they need to stand for full sense words, e.g. nouns (what/who) or adverbs (where/when/how).

Thus to have a wh- type preposition is sort of contradictory (to be honest, when I read the title, I thought you meant something like "WH-prep table did you put the plate?", i.e. asking in which position towards the table the plate was put - beneath it, on it, next to it,...).

What you are referring to (Wh-on did you put the toy) would most likely not be analysed as preposition but just a simple contraction of two words/morphs that can exist apart but in this specific combination, they merge into a single unit. This obviously happens frequently with prepositions and pronouns, e.g. French du, des, ès for de le, de les, en les, Czech nač, proč, doň,... for na co, pro co, do něj,..., but the resulting unit is typically still just univerbised realisation of two otherwise separate words/morphs or it turns into an adverb by the process of grammaticalisation (the Czech case - proč meant originally what for and now it means why and most likely the case with the English adverbs as well, coming from some PIE kʷo- and spatial expression of location, direction, origin etc. like wh-ere, wh-ence, wh-ither,...).

Answer 5

Not only do many languages have wh-prepositions, English itself has at least two wh-prepositions. These are the words when and where, which, although classified as adverbs in traditional grammars, are recognised as prepositions by such grammars that allow for intransitive prepositions (for example, The Cambridge Grammar of the English Language, Huddleston & Pullum, 2002). The words when and where pass all of the tests for preposition-hood in English and none of the tests for being an adverb.

In relation to the sentence:

• I put the toy on the table.

... the preposition phrase on the table may be replaced by the interrogative preposition where:

• Where did I put the toy?

Answer 6

The answer to your question is no. If a language had a word like whon? standing for what on?, first, that word should, if anything, be described as a 'wh-postposition', not as a 'wh-preposition', and secondly, and above all, if such a word existed in any language, its wh+Q-operator would have to 'question' just the preposition on, not its complement, and the result would not be a 'partial' wh+Q-question, as in your What did you put the toy on__?, but a polarity (= a 'yes/no') question similar in meaning to what B replies to A in the following dialogue (say, between somebody who has arrived at his shop at 8 a.m., before his shop attendant, to find all the lights on and the shop attendant supposedly in charge of closing the shop and turning them all off on leaving the day before):

A: Do you know how you left the lights when you went home last night?

B: On? (= Did I leave them on? = a yes/no question)

The reason why presumably no human language has an interrogative word like whon? meaning what-on (< on what?, after 'P-stranding', a very rare feature of English) is that, as Katz & Postal in their (1964) book argued, a) the wh+Q- morpheme behaves as an interrogative determiner, b) what wh+Q- 'determines' is necessarily a (pro)nominal element (i.e., it cannot be a 'predicate', say a preposition - or a verb, for that matter), and c) what the wh+Q- operator correspondingly 'questions' is the identity of a) a human individual, as in who (approximately analysable as the surface form of *wh+he), or b) a non-human individual or substance, as in what (approximately the surface form of *wh+it), or c) a time interval, as in when (the surface form of *wh+en), or d) a place, as in where (< wh+ere), or e) a way of doing things - as in how (roughly, < *wh+(s)o)), or,finally, f) a reason for doing things, as in why (< wh+y), basically. In sum: what can carry an interrogative wh+Q- determiner is, at bottom, an NP/DP or an equivalent pro-form (which may happen to be a 'bound morpheme') denoting an 'entity' (an 'object' in Frege's sense), and, since a preposition does not denote an 'object', it cannot carry a wh+Q operator. (In this, I fully agree with Eleshar's answer above). Note that the (informal) Logical Forms of Who wrote Hamlet?, When was Shakespeare born? Where was Shakespeare born?, etc. are, respectively, Which x, x = a human being, Wrote (x, Hamlet), Which x, x = an interval of time, Was Born (Shakespeare, at/in x), and Which x, x = a place, Was Born (Shakespeare, in x)].

That classical analysis of interrogative wh-items has never really been challenged because several of the abstract bound morphemes on which it depends, particularly -en, -ere, -ow, and the -y of why, do occur elsewhere: in the corresponding th- 'deictics' then, there, thus (< *th + so < Middle English swa), and in the Early Middle English instrumental thy - roughly meaning 'for this' (note the contrast wh + y <> th + y).

Hence, I am sorry to have to say that neither German wo+von, wor+auf, wo+mit, etc., nor English where+to, where+abouts, etc. (nor their Dutch counterparts) are, in fact, examples of what the OP is interested in. To start with, because all of them would be post-positional, not prepositional expressions, but above all, because in none of such cases is the preposition itself 'questioned' by the wh+Q- operator. Spanish por qué, French pourquoi, Italian per che, etc. would at least be 'prepositional', but again, even in those cases, what the wh+Q- operator 'operates' on and 'questions' is not the preposition, but its '(pro)nominal' complement, whereas in the OP's hypothetical wh+on example the preposition has no complement and the wh+Q-operator would have to operate on the preposition itself (= a two-place 'predicate'), which, as far as I can tell, is impossible on conceptual grounds.

That does not mean that it is impossible for a predicate, in general, to be the apparent focus of a question, of course. In a yes/no question like Did he marry her? the VP marry her certainly counts, informationally speaking, as the focus of the question, but what the Q-operator really 'binds' (informally: 'questions, focuses on') in that clause is not the VP itself, but the polarity variable - ranging over the two t-type values {'true', 'false'} - associated with all clauses. Hence, the meaning of that question can be paraphrased as 'Is it true (or false) that he married her?', and, even more tellingly perhaps, the natural 'tag' that can be added to it is didn't he (or just not), as in Did he marry her or not/didn't he?, which clearly points to the clause's polarity value as the real focus of the Q-operator.

Interestingly, in Old English, presuppositionally 'marked' yes/no questions could still be introduced by the wh+Q-word hw+aether (a direct ancestor of our present-day whether) that could be transparently analyzed as the wh+Q- operator (spelt hw- at the time) plus a determiner, either, that introduced a proposition with the canonical structure of a positive statement (as in E. Closs Traugott's example [HES, p. 73] Hwaether ge nu secan gold on treowum? = 'Is it (really) true, or false (as I believe), that you now seek gold in trees?'), which made the logical structure of all OE questions, both polarity-focused and non-polarity-focused, look more uniform than it seems in Modern English (although, of course, we still have whether, possibly followed by or not, introducing 'indirect' polarity questions with the syntactic SVX form of statements (= propositions).

The only case in which a predicate can be the real focus of a Q operator, then, is in 'alternative questions' like Did he marry her or leave her?, but note that those are neither 'partial' nor 'polarity questions': they resolve, though, into two coordinated (contraposed) polarity questions, i.e., Did he marry her or did he leave her?, in each of which a separate Q-operator 'binds' the polarity variable (not the VP!) of the respective coordinate clause.