This study offers an explanation to a long-standing question in the typological distinction among languages with respect to the formation of wh-questions (i.e. interrogatives which use question words such as 'who' and 'what'). It is well-known that languages differ in the position of the question word(s) in a wh-question. It is proposed that both the availability of question particles and the properties of question words contribute to the typological distinctions found. In particular, the author argues that the availability of question particles correlates with the lack of fronting of question words. A theory of Clausal Typing is proposed to account for this correlation. More specifically, languages employ either question particles or a fronting strategy to "type" a clause as a wh-question. The theory of Clausal Typing together with the Principle of Economy of Derivation predicts that (a) no language has the option of alternating between the two methods of Clausal Typing and thus there are no languages with "optional fronting" of question words and (b) fronting of one question word is sufficient to type a clause as a wh-question. Apparent counterexamples to the predictions involving "optional fronting" languages such as Bahasa Indonesia and Egyptian Arabic as well as "multiple fronting" languages such as Hungarian and Bulgarian are discussed and accounted for.
The internal structure of question words is further shown to shed light on two particular issues in the literature: (i) the lack of scope ambiguity in "in-situ" languages such as Mandarin Chinese (a language without fronting of question words); and (ii) the question of whether "in-situ" question words undergo fronting at Logical Form. It is proposed that question words in Mandarin are indefinite noun phrases without inherent quantificational force, and that this contributes to the lack of scope ambiguity in the language. In addition, arguments for and against LF fronting of question words are examined. It is shown that evidence against such fronting does not hold and that the properties of question words in "in-situ" languages do not preclude fronting at LF.