James Hurford: Animals Do Not Have Syntax (Compositional Syntax, That Is)

After passing my final exams I feel that I can relax a bit and have the time to read a book again. So instead of reading a book that I need to read purely for ‘academic reasons’, I thought I’d pick one I’d thoroughly enjoy: James Hurford’s “The Origins of Grammar“, which clocks in at a whopping 808 pages.
I’m still reading the first chapter (which you can read for free here) but I thought I’d share some of his analyses of “Animal Syntax.”
Hurford’s general conclusion is that despite what you sometimes read in the popular press,

“No non-human has any semantically compositional syntax, where the form of the syntactic combination determines how the meanings of the parts combine to make the meaning of the whole.”

The crucial notion here is that of compositionality. Hurford argues that we can find animal calls and songs that are combinatorial, that is songs and calls in which elements are put together according to some kind of rule or pattern. But what we do not find, he argues, are the kinds of putting things together where the elements put together each have a specified meaning and the whole song, call or communicative assembly “means something which is a reflection of the meanings of the parts.”

To illustrate this, Hurford cites the call system of putty-nosed monkeys (Arnold and Zuberbühler 2006). These monkeys have only two different call signals in their repertoire, a ‘pyow’-sound that ‘means’, roughly, ‘LEOPARD’; and a ‘hack’ sound that ‘means’, roughly, ‘EAGLE’.

They can also combine both kinds of calls. So ‘pyow-hack’ seems to mean something like ‘LET’S GO’.
This then, is a (very small) combinatorial system in which two elements put together contain a new meaning. However, this system is not compositional because, the new meaning ‘let’s go’ is not a function of and cannot be derived from the two single meanings of ‘leopard’, and ‘eagle’ put together.

Wild Campbell’s monkeys, in comparison, seem to have a much richer call type repertoire (see also this post by Hannah on Diana Monkeys). Apart from specific alarm calls for leopards and eagles, they also sometimes produce a pair of ‘boom’ vocalizations about 25 seconds prior to an alarm call. What is special about this call is that on one interpretation, the ‘boom’ can be said to have the function of cancelling the following alarm call out or more precisely, of modifying it in a way similar to ‘maybe’ or ‘not-to-worry-about’:

“Boom-introduced alarm call series are given to a number of disturbances, such as a falling tree or large breaking branch, the far-away alarm calls of a neighbouring group, or a distant predator. Common to these contexts is the lack of direct threat in each, unlike when callers are surprised by a close predator’ (Zuberbühler 2002: 294).

So on one interpretation, this call sequence can be interpreted as compositional in the sense of (Boom =) DISTANT + (Alarm Call=) THREAT = DISTANT THREAT. However, Hurford argues that the 25s delay between the elements makes is problematic to think of this as a ‘composed’ message. In addition, other monkeys seem to show little or no alarm at all in reaction to this call sequence. This makes quite impossible to get at whether other monkeys interpret these calls as semantically compositional, or whether, for example, the ‘boom’ simply produces a state of ‘Okay, whatever follows next, I don’t really need to worry about or pay attention to’, and that’s it.

But the communication system of Campbell’s monkeys is even more elaborate than that. Beside the ‘boom’ vocalization, they also have five different types of ‘hack’-sound. Outtara et al. (2009)found that these different types of ‘hack’-sounds can be combined into meaningful sequences, and that the combination of calls into sequences was governed by a number ob principles.

One example, was a combination of a pair of ‘booms’ (B), which according to Outtara et al. when used “alone instigate group movements toward the calling male” and a series of ‘krak-oo’-calls (K+), which on their own function as general and unspecific alarm calls. However, according to Outarra and colleauges B-K+ is used for, or ‘functionally refers to’ falling wood.

What is more, when this sequence is followed by a series of ‘hok-oo’-calls, which by themselves do not mean anything, the meaning of the call sequence changes again: The sequence B-K+-H+ is used by callers when they detect a neighbouring group.

There are also regularities regarding the order of elements within the different call sequences these monkeys produce. “Boom calls, indicative of a nonpredation context, always preceded any other call types. H and K calls, indicators of crowned eagles or leopards, were always produced early in the sequence and were relatively more numerous if the level of threat was high” (Outtara et al: 22029).

Hurford concludes that

These monkeys do produce systematically formed call-sequences, so, like birds, they have some combinatorial syntax. The sequences are meaningful, apparently referential, but the meanings of the sequences are not functions of the meanings of the parts, so the syntax is not semantically compositional.

He comes to the same conclusion regarding the complex communicative behaviour of songbirds and whales, which is also combinatorial, but not compositional. It is important to note however, that phonology is also combinatorial and that is quite likely that in the evolution of language phonological or ‘musical’ combinatorial evolved first (see also this talk by Keelin Murrary that Sean linked to). Syntactic compositionality might then have evolved out of this musical compositionality. This is another reason why it is very worthwhile to study the complex compositional systems of non-human animals.