Universals of conversation

A word like 'Huh?' —used when one has not caught what someone just said—appears to be universal: it is found to have very similar form and function in languages across the globe. This is one of the findings of a major cross-linguistic study by researchers Mark Dingemanse, Francisco Torreira and Nick Enfield, at the Max Planck Institute for Psycholinguistics in Nijmegen, the Netherlands. The study was published in the journal PLOS ONE.

It might seem frivolous to carry out scientific research on a word like 'Huh?' But in fact this little word is an indispensable tool in human communication. Without words like this we would be unable to signal when we have problems with hearing or understanding what was said, and our conversations would be constantly derailed by communicative mishaps. The research is part of a larger investigation of language and social interaction funded by the European Research Council.

Dingemanse and colleagues studied languages from around the world and found that all of them have a word with a near-identical sound and function as English 'Huh?' This is remarkable because usually, words in unrelated languages sound completely different. Compare, for example, these very different-sounding words for 'dog': inu in Japanese, chien in French, dog in English. One might object that this suggests that 'Huh?' is not a word at all. But in a careful phonetic comparison, Dingemanse and colleagues find that it is. Although 'Huh?' is much more similar across languages than words normally should be, it does differ across languages in systematic ways. 'Huh?' is not like those human sounds that happen to be universal because they are innate, such as sneezing or crying. It is a word that has to be learned in subtly different forms in each language.

Why is 'Huh?' so similar across languages? To understand this, Dingemanse and colleagues studied the specific context in which this word occurs. In human communication, when we are somehow unable to respond appropriately, we need an escape hatch: a way to quickly signal the problem. This signal has to be easy to produce in situations when you're literally at a loss to say something; and it has to be a questioning word to make clear that the first speaker must now speak again. Since these functional requirements are fundamentally the same across languages, they may cause spoken languages to converge on the same solution: a simple, minimal, quick-to-produce questioning syllable like English 'Huh?', Mandarin Chinese 'A?', Spanish 'E?', Lao 'A?', or Dutch 'He?'.

The basic principle is well-known from evolutionary biology: when different species live in similar conditions they can independently evolve similar traits, a phenomenon known as convergent evolution. For example, sharks and dolphins have different evolutionary origins but similar body plans, because they live in the same aquatic environment. In the same way, Dingemanse and colleagues propose that words may converge on similar forms when they occur in strongly similar conversational 'environments'. A clear effect of this conversational ecology on the specific shape of linguistic expressions has not been observed before.

Although 'Huh?' may almost seem primitive in its simplicity, a word with this function is not found in our closest evolutionary cousins. Only humans have communication systems in which complex thoughts can be expressed and communicative mishaps can be solved on the spot. Even a humble word like 'Huh?' can teach us a lot about our nature as ultrasocial animals.

Original publication

Dingemanse, Mark, Francisco Torreira, and N.J. Enfield. 2013. "Is 'Huh?' a universal word? Conversational infrastructure and the convergent evolution of linguistic items." PLOS ONE. doi:10.1371/journal.pone.0078273 — http://dx.plos.org/10.1371/journal.pone.0078273

More information

Research question: Is "Huh?" a universal word? And what is the explanation for its striking similarity across ten languages from five continents?

Key findings

1. Huh? is not trivial. It might seem frivolous or even trivial to carry out scientific research on a word like Huh? But in fact this little word, along with others that function in similar ways (e.g., 'Sorry?' 'What?') is an indispensable tool in human communication. Without such words we would be unable to signal when we have problems with hearing or understanding what was said. Because conversation moves along so quickly, if we did not have reliable ways of signalling trouble, we would constantly fail to stay 'on the same page' in social interaction. While Huh? may seem an unlikely topic of scientific research, in fact human communication, and thus common understanding in social life, relies heavily on the use of such linguistic devices.

2. Huh? is universal. We sampled languages from around the world in this study, and we found that all of them have a word with a near-identical sound and function as English Huh? This is an exception to the normal situation, namely that when words in different languages mean the same thing, they will usually sound completely different: compare, for example, these very different-sounding words for 'dog': inu in Japanese, chien in French, dog in English. Why do these differences between the sounds of words across languages occur? Because language does not impose any necessary connection between sound and meaning in words (a principle that linguists call 'the arbitrariness of the sign'). This study shows that 'Huh?' is a rare exception to this otherwise strong rule.

3. Huh? is a word. An objection to our first finding might be that 'Huh?' is not a word after all. But our study finds that it is. Although the expression 'Huh?' is much more similar across languages than words normally should be, when we zoomed in and looked at the finer details, we discovered that this expression does differ across languages in subtle but systematic ways. These differences give us evidence that 'Huh?' is integrated into each linguistic system, thus supporting the view that it is, in fact, a word. Here are some of the subtle differences: In Spanish it's e?. In Dutch it often starts with /h/, as in hé. In Cha'palaa (an indigenous language of Ecuador) it has a falling tone and often starts with a glottal stop, as in ?a! Therefore: Huh? is not like those human sounds that happen to be universal because they are innate, such as sneezing or crying. It is a word that has to be learned in subtly different forms in each language.

4. Huh? is not innate. 'Huh?' may seem almost primitive in its simplicity, but in fact nothing like it is found in our closest evolutionary cousins. It's not an involuntary response like a sneeze or a cry of pain. Indeed, to have such a word, specialized for clarifying matters of understanding, only makes sense when a fully functioning cooperative system of communication (i.e., human language) is already in place — babies don't use it, infants don't use it perfectly, but children from about 5 have mastered it perfectly, along with the main structures of their grammar. If there is a plausible explanation that doesn't assume it's innate, we prefer that, on the standard scientific principle that it is best to keep to the simplest possible assumptions and explanations.

5. Huh? is likely shaped by convergent evolution. In conversation, we are under pressure to respond appropriately and timely to what was just said; when we are somehow unable to do this—for example, when we didn't quite catch what the other person just said—we need an escape hatch. This particular context places constraints on, and functional motivations for, the form of the word. The signal has to be something maximally simple and quick to produce in situations when we're literally at a loss to say something; and it has to be a questioning word to signal that the first speaker must now speak again. In language after language, we find a word like 'Huh?' that fits the bill perfectly: it is a simple, minimal, quick-to-produce questioning syllable. We propose this is a form of convergent evolution in language. Convergent evolution is a phenomenon well-known from evolutionary biology. When different species live in similar conditions, they can independently evolve similar traits. In a similar way, the similarity of huh? across a set of widely divergent languages may be due to the fact that the constraints from its environment are the same everywhere.

What did we do?

We examined 196 instances of huh? extracted from recordings of informal conversation in 10 languages around the world: Siwu (a minority language spoken in Ghana), Italian, Mandarin Chinese, Spanish, Cha'palaa (a minority language spoken in Ecuador), Icelandic, Lao (spoken in Laos, Thailand, and Cambodia), Dutch, and Murriny Patha (an Australian Aboriginal language). In all of the languages, we restricted our focus to cases that occurred in exactly the same context in conversation. The technical term for this context is "other-initiated repair": a sequence in which one person says something, the other then 'initiates repair' by saying Huh?, and the first provides a 'repair solution', usually by repeating the thing they said before (though often with slight revisions).

We studied these audio recordings in two ways. First, we did an auditory phonetic analysis in which three linguistically-trained analysts independently scored every one of the 196 recordings on five phonetic dimensions. The instances were presented in random order and no information was provided about the language. One set of findings presented in the paper was derived from the combined results of this auditory phonetic coding. Second, we did an instrumental phonetic analysis of a subset of the data in order to verify the quality of the auditory analysis. For this we focused in on two of the languages—Spanish and Chapalaa (an indigenous minority language spoken in Ecuador)—and we compared readings of pitch, and of the first and second 'vowel formants' of each example (these provide an objective measure of the differences in sound between vowels such as 'a' versus 'i' versus 'u'). To make sure that the similarities we saw were not just due to our particular selection of ten languages (some of which are related), we also located examples of the interjection in as many additional languages as we could find — adding 21 languages to our original sample. In all of the languages we found an interjection that showed the same tight fit of form and function.

What did we NOT do?

• We did NOT study huh? in dictionaries — many dictionaries don't list it as a word and even if they do we can't be sure of its precise phonetic form.
• We did NOT look at uses of huh? in contexts other than 'other-initiated repair'. We know that huh?-like words are also used in other contexts (e.g. to pursue a response, or as a tag question at the end of an utterance). We suspect that these uses are related to the function we studied, but that would be a matter for another study.
• We did NOT compare written-down versions of huh? across languages, nor did we rely on second-hand reports about the languages we studied. We collected almost 200 audio recordings of huh? (averaging about 20 per language) in our own field recordings, and we compared those auditorily and instrumentally. This was to make sure that our analysis was based on directly observed facts about what happens in real, everyday language use.
• We did NOT study only huh?. This research is part of a larger investigation of language and social interaction in the ERC-funded project "Human Sociality and Systems of Language Use" led by Enfield. 'Huh?' belongs to a larger set of expressions used to signal and solve communicative mishaps. Most of these expressions vary quite widely across languages. Only 'Huh?', which is the most common generic signal of trouble we found, happens to be strongly similar everywhere — a finding which we report in this paper. If you are interested in the larger systems, we can recommend another comparative article, to appear in the scholarly journal Studies in Language: Dingemanse, Mark, Joe Blythe, and Tyko Dirksmeyer. in press. "Formats for other-initiation of repair across languages: An exercise in pragmatic typology." Studies in Language.

---

---