Barbara King points to a very interesting article in press at Animal Behaviour. In their essay "The language void: the need for multimodality in primate communication research" Katie Slocombe, Bridget Waller and Katja Liebal analyse more than 550 studies on primate communication from 1960 to 2008 and argue that research in one modality (e.g. gesture) often differs so strongly in its methodology from research on another modality (e.g. alarm calls) that the results can hardly be reliably compared. Here's their abstract:
Theories of language evolution often draw heavily on comparative evidence of the communicative abilities of extant nonhuman primates (primates). Many theories have argued exclusively for a unimodal origin of language, usually gestural or vocal. Theories are often strengthened by research on primates that indicates the absence of certain linguistic precursors in the opposing communicative modality. However, a systematic review of the primate communication literature reveals that vocal, gestural and facial signals have attracted differing theoretical and methodological approaches, rendering cross-modal comparisons problematic. The validity of the theories based on such comparisons can therefore be questioned. We propose that these a priori biases, inherent in unimodal research, highlight the need for integrated multimodal research. By examining communicative signals in concert we can both avoid methodological discontinuities as well as better understand the phylogenetic precursors to human language as part of a multimodal system.
Barbara King's discussion of the article is also very illuminating.