So the danger is over-interpreting their output. Was this sentence intended or just random babbling? Does the dog understand the word differently from what it means in English? E.g. “bye” seems to become a verb meaning “leaving”, “love you” is used for affection but obviously doesn’t reflect deep understanding of the human concept of love.
Interestingly, there’s an argument that human infants also learn language by their parents over-interpreting their input, with the infants then adopting those interpretations as true. So one could argue that even if over-interpretation happens with dogs, that only makes it a process similar to human language learning, with the parent/child and owner/dog creating a shared language game.
Our first type of example comes from our own data concerning Zulu infants of between three and four months of age interacting with their mothers, and suggests an answer to this question. [...]
As noted above, there are times when a caregiver will want an infant to fall silent, or in isiZulu to ‘thula’. Zulu children are traditionally expected to be less socially active than contemporary Western children, to initiate fewer interactions, and, crucially, to show a respectful attitude towards adults. An early manifestation of this is in behaviours where a mother attempts to make an infant keep quiet, sometimes saying ‘thula’ (‘quiet’), ‘njega’ (‘no’), while simultaneously gesturing, moving towards or away from the infant, and reacting to details of the infants own behaviour (see Cowley et al., in press).
At these times the mother regularly leans forward, so that more of the infants visual field is taken up by her face and palms. New vocalisations, and movements or re-orientations of gaze by the infant, are often ‘nipped in the bud’ by dominating vocalisations (sometimes showing prosodic properties indicative of disapproval, comforting, attention and/or arousal towards the mother herself) from the mother, sometimes accompanied by increasingly emphatic hand-waving, and even closer crowding of the infants visual field. [...]
With high regularity, and within relatively little time, the particular infant often does ‘thula’, at which point it is generally rewarded with smiling, gentle touching, and other comforting.
At this stage there is no reason to believe that the infant knows what ‘thula’ or ‘njega’ means, or even that it could reliably re-identify the words, let alone produce or contemplate them, so it is extremely unlikely that the word-based aspects of maternal utterance-activity provide labels for the infant. We are considering infants before the stage linguists call ‘babbling’, let alone recognisable speech production. It is not even necessary to suppose that it ‘knows’ that it is supposed to be quiet when behaved at in the ways we have just described. We know that the mother wants the child to be quiet, that this expresses itself in behaviour by the mother, and that the infant comes to be quiet.
If we examine the mothers behaviour, though, we can make sense of it. She ensures that it is difficult for the infant to attend to anything else by crowding its visual field. She rejects active or new behaviours on its part by cutting off its vocalisations and movements with dominating signals of her own. She largely restricts approval signals, including relaxing the crowding, and reducing the magnitude of her gesturing, as well as expressing comfort through vocalisation, facial signalling and touch, to moments when the infant begins to quieten down. Its not particularly surprising, then, that it does quieten down.
The mothers behaviour includes salient, repeated, features which are apt for learning. Her patterns of hand gesturing, for example, could at the outset be iconic of the whole episode including her behaviour and the infants becoming quiet, but, when repetition allows the gesture to be individuated and recognised in its own right, go on to become an indexical cue that quietness should follow. The infants responses then become indexical for the mother of the degree to which the child is co-operative, well-behaved, or, more plainly, ‘good’. Caregiver descriptions of infant behaviour at these times, manifest either in their explicit vocalisations to the child, including references to being ‘good’, or references to possible disciplinary sanctions such as ‘kuza baba manje’ (‘wheres your father now?‘) or, in interviews following the videotaping, show that infant behaviour even at this early age is being classified in line with culturally specific expectations of good and bad behaviour. And a crucial part of what makes for a ‘good’ child is responding in ways sensitive to what caregiver behaviour is actually about, strikingly in controlling episodes such as the one just described, which make possible the earliest ascriptions of ‘obedience’, ‘cooperativeness’ and so forth.
These ascriptions are over-interpretations. They are, though, necessary overinterpretations, in so far as they motivate caregivers to imbue their own behaviour with regularities manifest regularities in their own behaviour which are then available as structure in the interactional environment for (learning by) the infant. A further episode from our data, in this case concerning a child of around four months, illustrates this point about over-interpretation. In it an infant repeatedly vocalises in ways which to its mother, at least, are suggestive of its saying ‘up’. Each time she says ‘up’?, or ‘you want to go up’? and after a few repetitions she lifts the child. Prior to the lifting, there is little evidence that the child actually wants to be lifted, or that it has its attention focussed on anything in particular, except perhaps its own experiments in vocal control. When it is lifted, though, it beams widely. Whatever it did want, if anything, it is now, we suggest, one step closer to figuring out how to behave in ways that lead to its being lifted up.
Still on the subject of lifting, consider the common gesture made around the eighth month by infants who want to be picked up (that is, who subsequently smile or otherwise show approval when they are picked up following such a gesture): a simultaneous raising, or flapping, of both arms (see Lock, 1991). This gesture is not simply copied from common adult behaviours. In the terms we are using here it is partly iconic, in virtue of being a common posture of infants while they are in fact being held up, and partly indexical, in virtue of being able to stand on its own as an indicator of ‘being up’, as well as being symbolically interpretable as an invitation to lift, or a request to be lifted. Such gestures are, importantly, serviceable label candidates, in virtue of being amenable to disembedding from behaviour, and eventually coming under deliberate control. An infant need not want to be lifted the first few times it makes such a gesture, it has only to be able to notice that the gesture tends to be followed by liftings.
If and when such learning takes place, it does so in the affectively charged environment we have briefly described. We want to bring discussion of the current example to a close by suggesting a way in which these interactions should be regarded as a further example of how minds can be extended through action. Clark and Chalmers suggestion is that paradigmatically mental states and processes can be realised by structures and resources external to the brain. The world beyond the skull of any individual includes, of course, the skulls and brains of others. If active externalism motivates the recognition of a cognitive prosthesis such as a filofax as part of what realises a mind, then the embodied brain of another can also play that role. Here, then, is our suggestion: that at times interacting caregiver-infant dyads are neither one individual nor two, but somewhere in between. At the risk of sounding sensational and un-PC at the same time, infant brains can be temporarily colonised by caregivers so as to accelerate learning processes. [...]
The instances of indexical learning we describe also permit the beginning of a kind of ‘semiotic arms race’ between infants and caregivers. Once an infant has learned, for example, that the arms-up gesture can lead to being lifted, it is possible for ‘requests’ (that is, behaviours taken as requests by others, no matter how they are to the infant) to be lifted to be acted on, or to be refused. Prior to the construction and learning of the indexical relationship, this was impossible––a parent would lift a child when the parent wanted to, or thought it would serve some end. Once it has been learned, ‘requests’ can be differentially responded to, depending on their situation in patterns of interaction extending through time. Personal and cultural contingencies about infants and parents will codetermine what patterns are formed, and whether, for example, requested lifting is more likely after relatively quick acquiescence to silencing behaviour, or less likely in the period following failure to attend to objects or events in which a caregiver attempted to arouse interest.
Interestingly, there’s an argument that human infants also learn language by their parents over-interpreting their input, with the infants then adopting those interpretations as true. So one could argue that even if over-interpretation happens with dogs, that only makes it a process similar to human language learning, with the parent/child and owner/dog creating a shared language game.