- The syntax of yes and no by Anders Holmberg
This monograph is a crosslinguistic investigation of the syntax of yes/no questions and their answers. The idea is that answers to these questions, even when they consist of a single word, are derived by ellipsis from a full sentential expression. Because the elided constituent is identical to the constituent in the preceding question, the syntax of answers turns out to be very much the same as the syntax of questions. The discussions and proposals are based on data from 136 languages, out of which a handful (English, Swedish, Finnish, Chinese, Thai, and Welsh) are selected for closer analysis. The languages have been examined in relation to two parameters of variation, namely, if answer particles (as in 1a) or echoed verbs (1b) are used, and how negative questions (2) are answered.
The idea that verb-echo answers are derived from sentential expressions by movement and ellipsis is uncontroversial and receives support from the fact that in many languages the verb is inflected for tense and agreement. A movement-and-ellipsis analysis is also available for other questions, including alternative (Do they want tea or coffee?) and wh-questions (What do they want?). The fact that single-word answers to these questions are case-marked indicates that they are part of sentential structure. Verb-echo answers to yes/no questions may then look like 3, while single-word answers to alternative and wh-questions look like 4.
The idea that even particle answers are derived by ellipsis from a sentential expression may seem less straightforward, and it contradicts the idea put forward by, for example, Krifka (2013) that answer particles are clause substitutes. Holmberg aims to show that the ellipsis analysis has wider coverage and more explanatory power, since it is able to capture both verb-echo and particle answers to yes/no questions, as well as answers to other types of questions. Based on Hamblin 1958, 1973, H hypothesizes that knowing the meaning of a question means knowing the set of alternative statements that can constitute its answers. In neutral yes/no questions, the set of alternative statements is restricted to p and ￢p, a positive proposition and its negation, and in negative questions to ￢p and ￢(￢p), a negative proposition and its negation. To capture the idea that a question puts a set of alternative propositions before the addressee and asks them to choose the [End Page 240] one that is the true alternative, H proposes that a question contains an open question variable in the IP. The variable moves, overtly or covertly, to the C-domain in order to take sentential scope. In direct questions, the CP is embedded under Q[uestion]-force, which contributes an invitation to the addressee to assign a value to the variable. The addressee does this by answering the question: the answer contains something that makes value assignment possible. In yes/no questions, H argues, the relevant property is polarity: answer particles like yes and no are viewed as sentential operators that can apply to a question with an open polarity variable [±Pol] and assign a value, [+Pol] or [−Pol], to it.
What answers are possible in negative questions is shown to follow from the idea that a yes/no question contains a polarity variable [±Pol] and that the answer to the question assigns a value, [+Pol] or [−Pol], to the variable. The division of languages into those that follow the truth-based as opposed to the polarity-based answer system (i.e. languages where the truth of a negative alternative is confirmed by uttering Yes, and languages where it is confirmed by uttering No) is argued to be a consequence of differences in syntactic structure and, specifically, differences in the syntax of negation: truth-based answers are utilized in languages and constructions that have low negation, polarity-based answers in languages and constructions lacking low negation. Polarity and negation are viewed as two distinct properties, and the ‘wrong’ type of answer to a negative...