Today’s topic in class was the history and the metaphors behind the genealogical classification of languages. Comparing two family trees – one of the Niger-Congo language phylum and one of Sunjata Keita brought to the fore what underlies both these conceptualisations of these lineages, and what information is not considered relevant in them. Have a look. Do you have an idea about what is missing?

Both family trees trace genealogical relatedness – descent from a common ancestor – through time, the Niger-Congo tree from left to right, the Keita lineage from top to bottom (incidentally, family trees tend to show the trees uprooted and upside down, or lying down, but almost never growing from the roots, despite using the image of a tree… And both trees assume monoparental descent – languages have one ancestor, and lineage in Sunjata’s case is determined through the male line, until the very bottom of the tree, when some women come in, because of their importance for the plot development of the Sunjata Epos.

In class, a vivid discussion ensued: on the word lists and features used to establish the trees, on selections of words and features resulting in conflicting trees, and, crucially, on the metaphor underlying genealogical classification itself, on how it is intended to exclude contact-induced language change and on the historical background in which it was created, in a sociopolitical context focusing on racial and linguistic purity. We evoked the problems that keep appearing for Niger-Congo – the place of Mande languages, the integrity of Atlantic languages, the name for the Gur language family, the belonging of Dogon – and we discussed how the view that language is passed akin to asexual reproduction in biology conflicts with the way in which children learn languages, focussing on West African language socialisation in large families which in their majority unite speakers of several languages, creating a multilingual input from the outset. I stop here, because each of these topics deserves and hopefully will get its separate blog post.

But this is what I will end tonight’s story with: One course participant is Dogon, and he reports, very upset, that in the most recent classifications, the Dogon cluster of languages is seen as unclassified – there is not enough evidence to give it a place in any established language family or stock. But, he says, we have relationships – we belong into the Mande world! It pains me how a model developed in the nationalist and racist context of the late 19th century can continue to cause that much harm by denying people’s relationships to spheres with which they feel much social and also linguistic belonging – just not the one of the kind measured by the comparative method. But lets’s remind ourselves that this method calculates relatedness by computing the lexical similarity of 100 to 400 words from the most basic vocabulary only, sometimes complemented by features selected to support particular classifications, explicitly and by design discarding the wealth of culturally meaningful layered tapestries that emerge when we look at other areas of language. Perhaps we want to keep using the family tree metaphor. But we need to be clear about one thing: it tells a very limited story, leaving out some of the main protagonists, and we need to give more room to other stories.