To talk about breathing more precisely, we first need to determine which function of the respiratory system is actually under focus here. Broadly speaking, breathing has two main functions:
- Breathing as a gas exchange system connected to the overall functioning of the body, controlled by the brainstem.
- Speech breathing – controlling the airflow to make sounds and to speak. This system, as well as other speech and language functions, is controlled on a higher organisatory level of the brain than the main gas exchange system.
For conversations, it is most important how air is used to speak and influence the course of a conversation.
The basic patterns of speech breathing can be divided into two as well. Silent breathing, usually the characteristic breathing pattern of a quiet listener, consists of inhalation and exhalation phases more or less similar in terms of duration and amplitude. Breathing during ongoing speech, on the other hand, is characterised by very short inhalation periods and rather long exhalation periods.
While speaking, we have to consider the amount of air in our lungs. If there’s too little air, we can’t finish what we intended to say and have to inhale again to continue. It is known that when reading aloud, one always inhales just before the beginning of a new paragraph. It also very likely happens before a new sentence, and sometimes between the clauses of more complicated composite sentences, i.e. before conjunctions or a comma.
Thus, based on the structure and compositions of a text, it is fairly well predictable when a new breath will be taken. To determine when it would be most convenient to inhale, we use information about what we are about to say and where we need or want to end up with our speech. When reading aloud, we get that information from the text in front of us.
The situation gets a lot messier when we don’t have a pre-written or memorised text to read aloud. That’s the case with ordinary everyday conversation, the course of which can’t be predicted.
Although the process of speaking requires the speaker to be more or less aware of what he or she is about to say, speech planning in real time puts a greater toll on the brain than just reading aloud something that’s in front of you. Therefore, spontaneous speech is characterised by many thinking pauses, unfinished sentences, abrupt changes of topic, jumping back and forth between topics, etc. There are more inbreaths in grammatically less appropriate places as well. Despite that, during spontaneous speech one still tries to take a breath at structurally logical places, i.e. between larger constituents of the sentence. This makes it easier to plan one’s speech, chunk by chunk, as well as making the talk more easily comprehensible for the listeners.
Speak before you breathe, otherwise you might be too late!
The situation is made even more complicated by the fact that every conversation has two or more participants, and everyone wants to say something sooner or later. It is clear that speaking at the same time is not a good idea, because then no one would understand anything. So, the participants somehow need to signal to others that they wish to start speaking themselves (take a turn), that they have not yet finished their speech (hold the turn), or that their talk is over now and somebody else can continue instead (pass the turn).
To get ahead in this imaginary speaking queue, all kinds of sneaky tricks can be used. For example, one might interrupt just when the other speaker is out of breath and needs to inhale for a moment before the thought can be completed.
This trick is part of a larger system that includes a lot more of such tools used for organising conversations, even though we might not be aware of using them. For example, in many languages a falling tone sometimes accompanied by a creaky voice in the speaker’s intonation shows that he or she is about to stop speaking. Why? It shows that the air in their lungs is running out and the subglottal pressure necessary to sustain a steady voice is decreasing. Thus, we perceive the falling tone and creakiness as a signal that the speaker must soon stop speaking to inhale, even if just for a brief moment.
During conversations we are looking for signs like this to know what the other participants might want to do or say and how our own conversational intentions fit in.
Speech breathing is an important tool for regulating the fluctuation of turns as well, although we tend to perceive it subconsciously. For example, inhalation has been thought to be a signal showing that an utterance is about to begin. This signal is perceived more strongly when the speaker’s inhalation is loud and/or physically noticeable. Holding one’s breath might show that the turn is not over and the speaker is just having a pause to think. Exhalation has been thought to be a sign that the speaker has finished and somebody else can take over the turn.
If the number of participants is greater than two, tools for passing the turn to a specific other participant can be used. Sometimes it is achieved with eye contact, but it might involve direct addressing as well.
So far, the breathing patterns accompanying spontaneous speech have been studied less than the patterns of reading speech, due to some technical difficulties. However, the Phonetics Lab at Stockholm University has recently begun experimental research on conversations and breathing. In spring 2015, spontaneous multiparty conversations held in Estonian were recorded in the lab to study the connection between the depth of inbreaths and how it could influence the organisation of turn-taking. More precisely, it was studied whether the inbreath that takes place immediately before the start of a turn is deeper than inbreaths occurring during the same turn to see if and/or how speaking turns are planned.
In addition to the conversation, each participant’s breathing signals were recorded with special elastic belts. One belt was placed at the level of the armpits and another at the level of the bellybutton. The belts measured changes in thoracic circumference. The signals collected from the belts were combined into a weighed sum, which basically looks like the wave of inhalations and exhalations depicted in the first image.
The results of statistical analysis showed that the inhalation before a turn is really significantly deeper than the following breaths taken during the same turn. Additionally, it turned out that despite remarkably different amplitudes, in both positions the breaths started from a similar amount of air in the lungs. This means the difference between the amplitude of breaths taken before a turn and during it is caused by the different amount of air in the lungs at the end of the inhalation.
Thus, the signal used for indicating the start of speech might not only be a deeper breath but a physically more noticeable one too. The more air one inhales into the lungs, the more the thoracic cage expands, making the inhalation and the speech starting right after it easily noticeable to other participants in the conversation.
Kätlin Aare is a PhD student at the University of Tartu. The research initiated in her master’s thesis, “Respiratory patterns and turn-taking in spontaneous Estonian: Inhalation amplitude in multiparty conversations“, defended at Stockholm University in the spring, will be continued in a collaborative effort between the phonetics labs of the universities of Stockholm and Tartu. The goal is to figure out even more about what breathing patterns of spontaneous speech can tell us about conversational intentions.
The Estonian version of the story first appeared in ERR Novaator.