Time pressure and the limitations of memory compel you and your listener to engage in a fascinating linguistic trade-off
It seems obvious that the shape of a bird has everything to do with its struggle against gravity. Its entire physiology bears witness to the triumph of flight despite the tug of Earth’s forces. Less obvious, perhaps, is the fact that human language is shaped by an equally powerful force acting upon it: the stricture of time.
People perceive language, like everything else, through a tiny pinhole known as the present. What lies in the future remains out of view, then very briefly passes through that pinhole before it vanishes into the past. The language used in everyday conversations – whether it’s spoken or sign language – is by nature fleeting. The information it contains is different at every slice of time. So, for any of us to use language, we must rely on our memories of a no-longer-existent past and our imagination of a not-yet-existent future. Think of someone who is midway through a sentence: ‘The co-worker who my boss promoted the other day ac— …’ A stream of linguistic elements has been tossed out, and it’s already quickly receding into the past. Meanwhile, the sentence could fork out into any one of a number of different trajectories, and the likelihood of each depends on what has already been uttered.
Our ability to speak and understand each other, then, hinges on the spaciousness of our memory and our accuracy at prediction. The properties of human language are determined largely by the limits of these capacities.
When you are speaking, the dilemma is that your short-term memory isn’t capacious enough to hold the details of a full sentence. Its form would dissipate in your mind in the time between uttering the first syllable and the last. You are, in this sense, working against time when you speak. And so, you begin talking with only a vague sense of how the sentence will unfold, taking a leap of faith that you can work out the details of what comes next by the time the earlier part of the sentence has scrolled into the past.
Within that pinhole of the present, you simultaneously map out the sentence’s structure, rummage for upcoming words in your long-term memory, and draw up a plan for the movements of the lips, tongue and mouth that form the word you’re on the verge of uttering, all while actually speaking. The future comes bearing down mercilessly on the present, and occasionally you find that you are stalled at an uhm, not ready to utter the word that comes next, or that you’ve rushed things and lined up the wrong word or sound. What we call a slip of the tongue is really a slip of the mind under time pressure.
Language exists between the cognitive demands of the disappearing past and the nebulous future
Time also exerts pressure on the receiving end of language. A listener’s memory for the bare form of language, untethered from meaning, is fleeting. (Notice how much harder it is to remember a nonsensical list of syllables than a meaningful phrase of the same length: ‘lep blintasp lorset ap lep howd’, versus ‘the smartest person in the room’.) To avoid the accumulation of free-floating phonetic bits, the listener races to extract meaning from the first sounds that drop from a speaker’s mouth. If you hear ‘the cap—’, you are automatically rifling through your vocabulary to decide whether this snippet of speech refers to a hat or whether it will continue as ‘the captain’, ‘the capital’, or ‘the cappuccino’.
As words unfurl into phrases, you scramble to give them structure, well before you have clear evidence for how they fit together. ‘I told the teacher that the school …’ might continue with ‘… needed better programmes for disabled kids’, or it might take an unexpected syntactic turn, and continue as ‘… had just hired about my child’s disability.’ Sometimes, you’re able to leap into the future and predict the shape of the sentence to come, drawing on your stored memories of how language tends to pattern and your assessment of the current context. But if a sentence is complex – or you’ve predicted wrongly – you risk drowning in the flow of incoming speech as you struggle to recover meaning from the sounds that have vanished into the past.
Language exists in this tenuous space between the cognitive demands of the disappearing past and the nebulous future. These demands become apparent to language scientists when cracks in performance appear – as disfluencies, slips of the tongue, wrinkles in comprehension. Language is a compromise between the limitations of speakers and perceivers who are perpetually under time pressure. It’s an imperfect solution, riddled with ambiguity and indeterminacy of meaning. In fact, given its imprisonment in linear time, language could not exist at all, if not for the fact that we are surprisingly good at coping with linguistic uncertainty.
Some ambiguity is the inevitable result of language being laid out in time. The indeterminacy of the syllable ‘cap—’, or of the structure of a phrase, comes from the fact that we comprehend language from within the pinhole of the present, assigning meaning to what has been uttered even before its full shape is revealed. But ambiguity is endemic to language to a degree that may seem perversely unnecessary. There are times when it seems akin to a choice.
Both speaking and listening are complicated juggling acts but, of the two, speaking seems to be especially prone to logjams or outright collapse under the pressures of time. Hence, the speaker aims to offload as much work as possible on to the hearer, leaning on the latter’s fluency at toggling between memory and prediction, and resolving ambiguities. Hearers are rarely aware that they regularly navigate a minefield of ambiguities, and yet it is everywhere.
We almost always mean somewhat more than what we say, leaving the hearer to infer the full meaning
For instance, languages like to reuse the same sound combinations for multiple meanings. The prevalence of these ambiguous words across languages may be the result of speakers exploiting the strengths of hearers as a way to shave off some of the effort of speaking. In a study of English, German and Dutch, researchers found that the words most likely to be ambiguous were short, frequent and contained sound patterns that were common to the language. Notice, for example, how many different concepts can be expressed by the words ‘run’ and ‘see’ (as opposed to ‘quench’ or ‘eczema’). The researchers suggest that speakers are especially likely to repurpose such words because these are among the easiest syllables to produce. The time saved in mentally assembling their sounds can be allotted for some of the other demanding tasks of planning a sentence. But this need for efficiency in speaking leads to greater ambiguity for the hearer.
Another way to save time when speaking is to leave unsaid what the hearer can reasonably reconstruct. We rarely truly mean what we say. We almost always mean somewhat more than what we say, leaving the hearer to infer the full meaning. Our sentences are not so much blueprints for meaning as they are decipherable clues to what we intend to convey.
This opens the door to elegant compression. It allows you to say: ‘The bridge collapsed. The wood was rotten,’ rather than lumbering your way through the more precise statement: ‘The bridge collapsed because the wood out of which it was built was rotten.’ You can be reasonably sure your listener will fill in the blanks, that they won’t entertain the possibility that the bridge was knocked down by a nearby tree whose wood was rotten, or that you happen to be describing two entirely unrelated events.
A subtle compact exists between speaker and hearer: the speaker, eager to avoid the time-consuming steps of planning a complex sentence, can afford to jettison linguistic information that is readily inferred while making the effort to utter that which is not. No one is likely to say: ‘Ben moved the car’ if he did so by carrying it down the block with the help of five of his burly friends. In reporting that someone was stabbed, the speaker might leave the weapon unmentioned if it was a knife, but take pains to mention if it was an icepick.
Every utterance reflects a split-second decision about what to explicitly say and what to leave implicit, a balancing of the time it would take to plan and utter a phrase against the hearer’s ability – or need – to quickly recover its meaning if it were left unspoken. Languages reflect the accumulation of such decisions over historical time, congealed into grammar. The solutions take a variety of forms. In Turkish, any statement must include a tag on the verb to show whether the speaker witnessed the event directly (‘di’) or knows about it secondhand (‘miş’), whereas English offers some clunkier options (‘Apparently, John died’ or ‘I heard that John died’), which may be expressed or omitted as the speaker sees fit. The Turkish solution forces the speaker to encode this information, but then offers quick and easy syllables to minimise the time and effort of uttering it.
All languages must contend with the fact that time presses upon both the speaker and hearer
Similar contrasts abound: English forces speakers to commit to verb forms that express the past, present, or future; in Mandarin, the same form is used for all three and the hearer is left to infer tense through other means; Yagua, a language spoken in the Amazon, requires speakers to choose one of eight tenses, five of which slice the past into fine-grained categories.
The outcomes for different languages likely reflect the specific balance struck between the burden of speaking and the burden of inference. Subtle trade-offs occur within each language as to what is said and what is left unsaid, but all languages must contend with the fact that time presses upon both the speaker and hearer. As a result, there don’t seem to be some grammars that overwhelmingly require information to be explicitly encoded or others that leave airy gaps everywhere for the hearer to fill in.
All languages use the strategy of reducing information where it is easiest to infer – witness the ubiquity of pronouns, which tend to be very short and contain almost no information at all. The linguistic content of the pronoun ‘he’ tells us only that the referent is male. But because the use of this anaemic word is usually clear from the context, its lack of content poses little problem. Pronouns save the speaker from having to utter the person’s entire name repeatedly (‘Bernardo arrived. Bernardo brought a cake …’) Some languages use pronouns with even less content than English – Farsi pronouns, for example, do not commit to gender. And in Spanish, the pronoun can be entirely silent. To convey ‘He died’, one simply says: ‘Murió.’
All of this, of course, means that human language has uncertainty – even fragility – baked into it. Failures of communication can and do occur. Perhaps this is why, when much is at stake, people still seek out face-to-face dialogue, even though time pressures are most acute in spontaneous conversation. Along with our linguistic tools, we humans have developed the social skills to negotiate understanding – the nod to confirm that we’re following, the furrowed brow when we’re not, the patience with a speaker’s disfluencies and backtracking, the ability to instantaneously repair misunderstanding – so that our language is not grounded by its imperfections. If studying language patterns is like examining a bird’s anatomy, witnessing a conversation is like seeing it in full flight.
