A Reconstruction of Early Human (and Neanderthal) Grammars
Ljiljana PROGOVAC, Wayne State University
In order to hypothesize about the evolutionary origins of grammar, it is essential to rely on a theory or model of human grammars. Interestingly, scholars engaged in theoretical study of syntax, particularly those working within the influential framework associated with linguist N. Chomsky, have been reluctant to consider a gradualist, selection-based approach to grammar/language (e.g. R. Berwick and N. Chomsky, 2016, Why Only Us, MIT Press; see L. Progovac for a review of this book, to appear in Language). Nonetheless, this theoretical framework has in fact been used to reconstruct the stages of the earliest grammars, and to even identify the constructions in present-day languages which resemble/approximate these early proto-grammars (L. Progovac, 2015, Evolutionary Syntax, Oxford University Press). These constructions can be considered Òliving fossilsÓ of early grammars, in the sense of R. Jackendoff (2002, Foundations of Language, Oxford University Press). This reconstruction is at the right level of granularity to be able to exclude some hypotheses regarding the hominin timeline, and to support others, as well as to engage the fields of neuroscience and genetics.
According to the Minimalist Program (N. Chomsky, 1995, MIT Press), as well as the predecessors to this framework, modern sentences are analyzed as hierarchical constructs, consisting of several layers of structure, as illustrated in (1):
(1) [TP [vP [SC/VP]]]
Here TP is a Tense Phrase layer (sentence), vP is a transitive (higher) verb Phrase, VP is the basic Verb Phrase, and SC is a Small Clause. Thus, to derive a sentence such as Deer will eat fish, one first assembles the inner layer, SC/VP (eat fish). The transitivity layer (deer) and the tense layer (will) are added only later on top of this SC foundation. In other words, while VP/SC can be constructed without any higher layers, vP or TP can only be constructed upon the foundation of VP/SC. Importantly, this type of hierarchy offers the following precise and straightforward method of reconstructing previous syntactic stages in evolution:
(2) Structure X is considered to be evolutionarily primary relative to Structure Y if X can be composed independently of Y, but Y can only be built upon the foundation of X.
The layer upon which the whole sentence rests is the inner, foundational (eat fish) layer (VP/SC), which I therefore reconstruct as the initial stage of grammar: a flat, tenseless, intransitive two-slot mold, in which transitivity, and with it the subject/object distinction, cannot be expressed grammatically. The unspecified role of the noun in this layer can be characterized as the absolutive role, as such roles are not directly sensitive to the subject/object distinction. Absolutive-like roles are found not only in languages classified as ergative-absolutive, but probably in all languages, in some guise or another. Human languages in fact differ profoundly in how they express transitivity, and this reconstructed absolutive-like basis, as will be shown, provides the common denominator, the foundation from which all the variation can arise. Given this approach, variation in the expression of transitivity can shed light on the hominin timeline, as well as the timing of the emergence of hierarchical stages of grammar.
To take just one example, absolutive-like fossils are found among verb-noun compounds, as in English: cry-baby, kill-joy, tattle-tale, turn-coat, scatter-brain, tumble-dung (insect); Serbian cepi-dlaka (split-hair; hair-splitter), ispi-_utura (drink-up flask; drunkard), vrti-guz (spin-butt; fidget), jebi-vetar (screw-wind; charlatan); and Twi (spoken in Ghana) kukru-bin (roll-feces; beetle). Comparing compounds such as turn-table and turn-coat, we observe that the first describes a table that turns (table is subject-like), and the second describes somebody who turns his/her coat, metaphorically speaking (coat is object-like). But these two compounds are assembled by exactly the same grammar: the two-slot verb-noun mold, not marking any subject/object distinctions.[1] In addition to illustrating this basic grammar, such verb-noun compounds in various languages specialize for derogatory reference and insult, with many crude, obscene representatives, suggesting a possible role of sexual selection in the emergence of early grammars (L. Progovac and J. L. Locke, 2009, ÒThe urge to merge,Ó Biolinguistics).
Interestingly, in their article ÒOn the antiquity of language,Ó D. Dediu and S. Levinson (2013, Frontiers in Psychology) express their hope that some combinations of structural features will prove so conservative that they will allow deep reconstruction, which would shed light on the possible language abilities of e.g. Neanderthals. I propose that the earliest stages of syntax/grammar as reconstructed here provide just such a conservative platform, which could have been commanded also by our cousins and the common ancestor. I will also provide a sample of such flat proto-grammar, together with the concrete vocabulary that can be deduced from fossil structures.
Here I also briefly report on some specific findings of the fMRI experiment in which we tested the predictions of this proposal (Progovac, Crabtree, Rakhlin, Angell, Liddane, Tang, and Ofen, ÒNeural correlates of syntax and proto-syntax,Ó in submission). In contrast to their modern hierarchical counterparts, the flat (fossil) structures were hypothesized to show less focused activation in the networks specialized for syntax, in particular BrocaÕs-basal ganglia network, but more activation in the areas not narrowly specialized for abstract syntactic processing, expecting that such proto-syntactic structures can still recruit more general and more diffuse, ancestral patterns of brain activation. Importantly, comparable processing patterns have also been reported for KE family members with a FOXP2 gene mutation, implicated in SLI (specific language impairment) (F. LiŽgeois et al., 2003, ÒLanguage fMRI abnormalities associated with FOXP2 gene mutation,Ó Nature Neuroscience). This approach thus reveals some intriguing new possibilities for testing and cross-fertilizing the specific hypotheses on the evolution of grammar/language with the findings in genetics, language disorders, and neuroscience.
[1] What is more, this kind of two-slot grammar seems to be within reach to non-humans, revealing where continuity should be sought. To take one example, the bonobo Kanzi has reportedly mastered such grammar in his use of lexigrams and gestures (P. Greenfield and S. Savage-Rumbaugh, 1990, ÒLanguage and intelligence in monkeys and apesÓ).