Wh-Words as Variables
Wh-words — who, what, where, Japanese 何 nani/nan, 誰 dare — behave less like ordinary nouns and more like variables: the sentence builds as if a noun phrase filled the slot, and the question asks for the value that fills it. A learner who models なに/なん as “x in algebra” has landed on the analysis formal semantics settled on in the 1970s. Karttunen’s terminology footnote calls wh-questions “search questions”, because “semantically these questions involve a search for a suitable value for a variable (single search questions) or a set of variables (multiple search questions)”. 1
The variable analysis
Hamblin (1973) let every question denote the set of propositions expressed by its possible answers: “who came?” denotes the propositions John came, Bill came, Mary came, and so on. Karttunen kept the set-of-propositions shape and changed one clause — a question denotes the set of propositions expressed by its true answers rather than all possible ones — and gave the wh-phrase the job of supplying the variable those propositions vary over. 2 On this analysis an answer is the value (or value set) that fills the variable, which is what makes the algebra analogy exact: the question word marks a slot, and the reply names what goes in it.
The slot is an ordinary argument position
The variable sits where a noun phrase would, and takes the marking any occupant of that position receives. Japanese shows this with its particles: だれが きましたか “who came?” (が marks the doer of くる), だれを みましたか “who did you see?” (を marks the target of みる), なんですか “what is it?” (predicate complement). The question word carries no case of its own; the predicate assigns it. 何 fills subject, object, oblique, and predicate slots alike, and its reading follows what it fuses with: nani before separate particles (を, が, に, か, も) and word-finally, nan before です/だ, quoting と, の, d/t/n-initial sounds, and every counter. 3
Languages differ in where the variable is pronounced. English moves the wh-phrase to the clause edge (“who did you see?”); Japanese and Mandarin leave it in the argument position. Cheng’s Clausal Typing Hypothesis gives both strategies one job — a clause is typed as a wh-question either by a wh-particle (Mandarin ne; Japanese か) or by wh-movement — and economy prevents a language from using both: the languages with in-situ wh-words are exactly the languages with wh-particles. 4 In Mandarin the in-situ wh-words pattern as indefinites: no quantificational force of their own, polarity sensitive. 5
Quantity questions: the variable fused to its unit
Asking for a quantity welds the variable to the counter: なんえん “how many yen”, なんにん “how many people”, なんじ “what time” — 何 fused with the counter for yen, people, and clock readings (the counter system itself is in japanese-counter-words). The nani→nan shift is the phonological trace of the fusion: before a separate particle 何 stands alone and keeps nani; inside a compound it shortens. 6 One compound is homographic: 何人 is nan-nin “how many people” and nani-jin “what nationality”, written identically — one prompt carrying two answers, the interference shape worked through in flashcard-design.
Interrogative and referential readings
whos-on-first runs on the same lexical fact the analysis separates. Abbott’s players — Who, What, I Don’t Know, Why, Tomorrow — are interrogative forms used as names, so every declarative sentence reaches Costello as a question. The routine needs both readings of the form live at once: the variable reading (asking for a value) and the referential one (picking out an individual).
Related pages
- japanese-counter-words — the counter system the quantity questions plug into
- flashcard-design — interference pairs and prompt design around homographic forms
- whos-on-first — interrogative forms used referentially
- assessment-integrity — the fill-in blank as a variable slot, and what an item must state about it
Sources
- Karttunen, L. (1977). “Syntax and Semantics of Questions.” Linguistics and Philosophy, 1(3), 3–44. Lauri Karttunen 1977 — Syntax and Semantics of Questions
- Cheng, L. L.-S. (1991). “On the Typology of Wh-Questions.” PhD dissertation, MIT. Lisa Lai-Shen Cheng 1991 — On the Typology of Wh-Questions
- Trinh Ngoc Dieu (2026). “Nan or Nani: Why the Kanji 何 Changes Its Reading.” Unihongo. Trinh Ngoc Dieu 2026 — Nan or Nani: Why the Kanji 何 Changes Its Reading