Warbling starlings startle linguists
Starlings startle linguists
Warblers wow language pundits
Could starlings learn to speak?
Work on the language ability of starlings, at the University of California at La Jolla has questioned an assumption that humans are a ‘superior’ species, says
Scientists have found that starlings are able to learn to distinguish grammatical forms in sequences of birdsong, a capability not suspected in animals.
The common thread in the grammars of all the languages of the world had forced the hypothesis that humans have an in-born a ‘language acquiring’ ability. “When we study human language, we are approaching what some might call the "human essence," the distinctive
qualities of mind that are, so far as we know, unique to man”, says Noam Chomsky, American Linguist and pioneer in the field.
Grammar and context
The pre-Chomsky view of language was that it consisted of a collection of phrases and sentences, which users would memorise and use when relevant. The more communicative would be the ones with the greater store of phrases and sentences, while the less endowed
would be limited in their ability to talk or to understand. This, in fact, is very basic communication, the kind of language used when soldiers drill, to assign knobs to start or stop a machine or even to train animals. This kind of communication is said to
use a ‘finite state’ grammar.
But when it comes to programming computers, for instance, there is need for a method of composing finite commands into complex procedures. The same set of symbols, again need to fit into other sets of symbols, so that a phrase could occur in more than one
‘context’. These rules are called ‘generative grammars' or ‘context-free grammars’. These are many times more complex than the ‘finite state’ grammars, which is found in almost all non-human communication, of cats, insects chimpanzees, whales, elephants.
It is this quality, of units of meaning being embedded within another unit of meaning, referred to as ‘recursive’ structure, that was found to characterize human speech. An example would be: “This is
the house that Jack built.” The portion in italics is the complement of ‘This is’. Now this (italics) part can be transferred to play the same role in another sentence: ”This is the malt”, to give this sentence: “This is the malt that lay in
the house that Jack built” Now the portion in italics is saying not ‘what this is’ but is telling us ‘where the malt lay’.
This quality of parts of a sentence being able to refer to the whole sentence enables uses of such languages construct sentences that they have never heard before. And it is this quality that enables human babies to ‘work out’ the rules of grammar and become
fully communicative long before they could have heard more than a fraction of the constructions that they are able to use.
The field of linguistics studies the rules and relations between words and how languages evolve. Linguists began to see that patterns could rapidly be seen to exist for all languages. A trained linguist could, once she had checked out the pattern in a few
sentences of an unknown language, even detect an error of grammar in that foreign tongue. The way these patterns exist for all languages and the way a baby could rapidly pick even two or three different languages at the same time led to the belief that a ‘universal
grammar’ was ‘wired’ into humans. And that this did not happen in non-humans, which is why no other species exhibits speech.
Starlings are different
Scientists in University of California have now experimentally found that starlings, and perhaps other songbirds, can recognize recursive grammars. The experiment used phrases with just 2 ‘words’, the rattle (R) and the warble (W), which occur is starling
birdsong. A typical ‘finite state structure would be : RWRWRW. A longer phrase than that would then be: RWRWRWRW, one ‘RW’ has been attached to the end of earlier phrase.
But in the ‘context-free’ form, a phrase would be: RRRWWW or WWWRRR. Here, if a longer phrase is to be created, like: RRRRWWWW, it is really a case of inserting the ‘RW”, shown in bold, into the middle of the phrase. The question was whether starlings could
tell the difference between fixed-state compositions and context-free ones.
The starlings were trained, through operand conditioning, which is how we train a dog, to peck when they heard one form, but not the other. The snatches used were of varying length, but always with one or the other structure. It took a long time, but finally
the birds were able to tell clearly which form had been played.
The conclusion is that even if there is a ‘universal grammar’ wired in, it is not humans alone have the ability to use complexity in sounds to compose sentences!
[The writer can be contacted at: firstname.lastname@example.org]