Conversation is not like a game of ping pong


You might have noticed that when you’re having a conversation with another person, you take turns with them – you switch between producing your own sentence and listening to your partner producing theirs. This process seems fairly effortless. In fact, we often take turns with our partner without leaving long gaps between the end of their turn and the start of our own. Research suggests that there is some variability in turn-taking across languages. In Japanese the average gap between turns is around 7 milliseconds, while in Danish it is around 470 milliseconds. However, this gap is most commonly around 200 milliseconds, regardless of the language we speak.

Although turn-taking may be fast, speaking on its own is slow. Naming a picture takes us at least 600 milliseconds from seeing a picture to actually producing its name, while it takes us even longer to produce a sentence describing the picture (around 1500 milliseconds). How, then, are we able to quickly respond to our partner during conversation?

What makes conversation so easy?
Studies have tried to answer this question using highly scripted tasks, such as question-answering. These studies have shown that we are highly pro-active – we listen to our partner and try to guess what they are likely to say so we can plan our own response. For example, if you hear a speaker start to ask ‘Are dogs your favourite …’, then you may guess that the last word is likely to be animal or pet. Based on this guess, you can begin planning the response–yes or no–before you actually hear the word “animal”. This pro-active planning removes the timing burden from speaking, and is thought to enable us to take turns with only a 200 milliseconds gap.

But does this pro-active planning work outside of laboratory studies, where highly scripted conversations are less common? To answer this question, we looked at transcripts of two-person conversations between speakers of Dutch, German, and English. These speakers were recorded while talking about whatever they liked with friends, family, and strangers, and so they did not have any strong expectations about what they were each likely to say. We found each speaker’s turn often consisted of only a few words, and was shorter than the minimum 600 milliseconds needed to produce a single picture name. As a result, the pro-active planning we see in laboratory studies cannot explain how we manage effortless conversations in natural speech – a lot of the time, the speaker’s turn was too short for the next speaker to have enough time to plan their own utterance and speak within the 200 milliseconds gap.

If pro-active planning isn’t the answer, then what is?
We also considered what the speakers actually said to each other. Often, we think of a conversation as involving an exchange of ideas, where one speaker directly responds to something said by the previous speaker. But surprisingly, we saw that speakers often did not respond to each other at all, and instead developed their turns in parallel with their partner. For example, consider this transcript from the English corpus:

Pamela: The food is like all unique
Darryl: Hey
Pamela: and wonderful
Darryl: it’s major league yin and yang
Pamela: and heavenly

In this example, Darryl and Pamela are having a conversation about food at a restaurant. Importantly, even though they are both contributing to the conversation, Pamela does not directly respond to anything that Darryl has said previously. For example, her second utterance (and wonderful) is not a direct response to Darryl’s first utterance, nor is her third utterance (and heavenly) a direct response to his second utterance. Instead, Pamela is continuing a sentence that she had already produced.

We found that this type of parallel talk occurred around half the time in the Dutch and German conversations, and around a third of the time in the English conversations. These findings suggest that the way we think about conversation in laboratory studies is very different from the way conversation actually occurs in natural speech. Laboratory studies tend to think of conversation as a game of ping pong, with speakers batting ideas backwards and forwards and directly responding to each other. But our findings suggest that although speakers may be sharing ideas and talking about the same theme, they often do not directly respond to each other. Instead, they often respond to something they said previously.
How, then, do we even manage to have a conversation with another person? It is likely that we don’t mind so much if our partner doesn’t directly respond to what we are saying, as long as they respond in an appropriate way. In fact, responding in this way likely allows the conversation to develop. If speakers always directly responded to each other, then their interactions would be very limited and would be almost formulaic in nature.
In sum, although we may have the impression that conversation is a joint action with another person, we often just interact with ourselves. Next time you’re having a conversation, pay attention to what you’re each saying and think about whether you are directly responding to what your partner has said, or whether you’re simply responding to yourself. This behaviour may explain why we sometimes feel as though we haven’t been listened to properly.

Further reading:

  • Clark, H. H. (1996). Using language. Cambridge: Cambridge University Press.
  • Corps, R. E., Gambi, C., & Pickering, M. J. (2018). Coordinating utterances during turn-taking: The role of prediction, response preparation, and articulation. Discourse Processes, 55, 230-240.
  • Corps, R. E., Knudsen, B., & Meyer, A. S. (2022). Overrated gaps: Inter-speaker gaps provide limited information about the timing of turns in conversation. Cognition, 223,
  • Levinson, S. C. (2016). Turn-taking in human communication – origins and implications for language processing. Trends in Cognitive Sciences, 20, 6-14.