A case for systematic sound symbolism in pragmatics: Universals in wh-words

- Interlocutors need to rapidly identify the pragmatic content of turns in order to respond appropriately in real-time conversation.
- In particular, responding to content questions constitutes a harsh cognitive environment.
- Languages may adapt to this ecology in order to facilitate rapid turn taking such as providing a low-level phonetic cue to upcoming questions.
- Wh-words sound systematically similar within a language, controlling for some basic confounds.

This study investigates whether there is a universal tendency for content interrogative words (wh-words) within a language to sound similar in order to facilitate pragmatic inference in conversation. Gaps between turns in conversation are very short, meaning that listeners must begin planning their turn as soon as possible. While previous research has shown that paralinguistic features such as prosody and eye gaze provide cues to the pragmatic function of upcoming turns, we hypothesise that a systematic phonetic cue that marks interrogative words would also help early recognition of questions (allowing early preparation of answers), for instance wh-words sounding similar within a language. We analysed 226 languages from 66 different language families by means of permutation tests. We found that initial segments of wh-words were more similar within a language than between languages, also when controlling for language family, geographic area (stratified permutation) and analyzability (compound phrases excluded). Random samples tests revealed that initial segments of wh-words were more similar than initial segments of randomly selected word sets and conceptually related word sets (e.g., body parts, actions, pronouns). Finally, we hypothesised that this cue would be more useful at the beginning of a turn, so the similarity of the initial segment of wh-words should be greater in languages that place them at the beginning of a clause. We gathered typological data on 110 languages, and found the predicted trend, although statistical significance was not attained. While there may be several mechanisms that bring about this pattern (e.g., common derivation), we suggest that the ultimate explanation of the similarity of interrogative words is to facilitate early speech-act recognition. Importantly, this hypothesis can be tested empirically, and the current results provide a sound basis for future experimental tests.