Along with all the sophisticated issues that are part of speech origins, it is good to remember every once in a while that the primary fact was and is the rise of words. From one angle, they seem empty as a dead man’s glove. “Words, words, words,” sighs Hamlet as he laments hollowness itself. But move your head a notch and words look almost miraculous — sounds, voluntarily shaped, that signify. How did two people, or, on this blog, protopeople, ever come up with such a thing? A couple of recent articles, however, suggest that a word, or at least sharing words, is not so miraculous after all.
One of the favorite mysteries of words is the burst of them that enter toddler speech, starting at about 18 months. The size of a child’s vocabulary starts to balloon and continues growing for the next seventeen years, if growth ever stops. My teen years were long ago, but I still pick up a novelty or two. This spurt has inspired writers for a long time. I can recall decades ago, before I’d read Chomsky or George Miller, reading some scholar describe the great moment when infants suddenly realize that everything has a name. It was inspiring indeed, only slightly quashed many years later when a comedian named Rich Hall coined the word snigglet, as the name for words that didn’t exist but should. It served as a reminder that not everything does have a name. Even so, infants never heard of Rich Hall and their vocabulary does shoot up.
The August 3 issue of Science magazine, however, included an article by Bob McMurray titled, “Defusing the Childhood Vocabulary Explosion” (abstract here). He has performed a mathematical analysis of what is involved in learning words as they are spoken around you and concludes that the surge is much less mysterious than it seems. McMurray’s basic idea is that some words are easier to learn than others, as determined by features like length, familiarity of sound pattern, and frequency of use. As a few words are learned, others become easier to learn, and still more are learned, still more become easier to learn. So the nature of perception and words makes a sudden rise in vocabulary size inevitable once those first few words are mastered. Our general abilities can take us a much further than has been appreciated.
Mathematical arguments in psycholinguistics commonly provoke some suspicion in me, partly because I’m not confident of my ability to detect the flaws in a bit of mathematical reasoning and partly because they often rest on assumptions that seem too simplistic and tailor-made for the results they produce. But in this case, I’m inclined to take the argument seriously. As a practical matter, I like the way the result tosses out the need to evolve special word-learning mechanisms. It gives this blog one thing less to worry about. And it ties in with some empirical evidence from work with chimps and bonobos. If you can get them beyond a few hand signals, their vocabularies go way up too.
Another recent article takes a hard look at the importance of joint attention to learning words. There is a new on-line, peer-reviewed journal about language called Language and Linguistics Compass whose May issue included an article titled “Joint Attention and Vocabulary Development: A Critical Look” (abstract here) by Nameera Akhtar and Morton Ann Gernsbacher. They survey the literature and note that although there is ample data showing that the size of a vocabulary, in normally developing children, is positively co-related with the amount of joint attention they engage in, there are special cases in which this co-relation does not work, notably in certain autistic children and those with Down syndrome.
Autistic children are very poor at joint attention, especially weak at starting it themselves, yet they can pick up words. Akhtar and Greensbacher cite a study of an autistic child from age 20 to 32 months who “did not engage in co-ordinated joint attention until well after the vocabulary spurt.” Of course, now we suspect that the vocabulary spurt may not be all it is cracked up to be, but still this kind of clinical evidence is useful in reminding us that we have to be careful in assuming how mysterious and miraculous word-learning is.
Yet, it must take something to learn a word because, as the article goes on to point out, children with Downs syndrome “engage frequently in joint attention” but their vocabulary growth is poor. In fact, “frequency of joint attention episodes was negatively correlated with early language development for the children with Downs syndrome.” Just the reverse of normally developing children.
This article did not mention probably the most famous case of vocabulary learning without benefit of joint attention: Kanzi the infant bonobo. Kanzi’s mother was part of an attempt to teach sign language to apes. She was apparently a very poor student, requiring extensive training to learn anything at all. Her infant, Kanzi, was present for these lessons, apparently because the investigators found that easier than providing day care. It turned out that Kanzi had been picking up all the secrets that were evading mom. Once Kanzi’s knowledge of sign language was discovered, naturally, the investigators paid him close attention, but the experience established very sharply that “overheard” words can be learned without first engaging the pupil’s attention.
Let’s be clear about what Akhtar and Greensbacher argued. It wasn’t that attention is not necessary for word learning, or even that joint attention is not necessary. They are complaining about a strong definition of joint attention that requires awareness of mutual attention. Covert attention, overhearing language, eavesdropping if you will, is sufficient to permit word learning:
We are not arguing against the importance of joint focus in early word learning; indeed, we agree with Bruner and Tomasello that for a child to learn words he or she needs to attend to the speaker and to what the speaker is attending. But we are raising questions about the overt manifestation of mutual engagement, as well as the overt manifestation of attention. … joint focus (which may be established covertly by the child on his or her own) may be critical for learning early words, but mutual engagement may not.
What this work suggests for a blog focused on the origins of speech is that no special capacities had to be evolved to develop a species with a vocabulary of perhaps as much as several hundred words. Some other process had to develop, and then words were ready to emerge and spread through a population.
Yet, that preadaptation for using words was not enough to get us the vocabularies we all enjoy today. The July/August issue of Child Development has an article about the acquisition of verbs. “Action Speaks Louder than Words,” by Amanda Brandone, Khara Pence, Roberta Michnick Golinkoff, and Kathy Hirsh-Pasek (abstract here). The article attempts to resolve a well-known “paradox” of language development: verbs like “eat” and “go” are a regular part of the earliest vocabulary of young children, but verbs as a class of words are only learned relatively late (in the third year or older) in talking.
The words that are covered in the articles mentioned above and the article discussed in last week’s post (Moving Beyond the Interesting) were mostly nouns, the names of things that are naturally and unambiguously perceived. Most verbs, however, are more subtle than that. As the authors put it:
Verbs refer to relations within events, and any event can be conceptualized in terms of a multitude of different components, including, but not limited to, path (the trajectory of an action with respect to some reference, e.g., approach, enter), manner (how an action is carried out, e.g., walk, swagger, stroll), result (e.g., open, fill), and instrument (e.g., hammer, shovel).
Put briefly, if a child hears somebody say, “Mr. Jones is Xing,” even watching Mr. Jones’ action, the child might be unable to determine what the speaker meant if Xing was approaching, walking, strolling, or entering.
The authors performed a series of experiments to tease out how children managed to grasp what the verb refers to. It turns out that young children, children of the age of the vocabulary spurt, cannot learn ambiguous verbs on the basis of hearing them alone. They do better when they have some linguistic or social cues to help direct their attention where the verb refers.
The paradox of learning verbs may hold a clue to this blog’s focus on the origins of speech. There is a class of words that name obvious things and actions (“perceptually salient,” to use the technical term) which could be learned, given the proper social conditions, by any ape or australopithecine without any history of linguistic selection. Not surprisingly, the first words of children around the world tend to be for the same things, or at least the same sorts of things, that jump out at them.
Later, a second class of words appears, that works against the grain of natural perception. They refer to perceptions that are less salient or interesting than those identified by the first group and can only be learned in social settings where existing language is used to redirect attention from the more salient or interesting to the more subtle details.
For this blog, it is more evidence for a story line that has been gaining evidence for some time.
- Our ancestors, like apes today, were already smart enough to use some words, if only there were some social circumstance in which a listener cared to know what was on a fellow’s mind.
- Eventually something happened that did support such interest; words emerged naturally, requiring only minimal (covert) joint attention.
- Once the first words did emerge selection, in both biological and cultural forms, supported the evolution of speakers who could learn from the linguistic, social, and attentive cues that alerted them to topics not previously natural to their ancestors. Something truly new had appeared under the sun.