doi 10.4436/JASS.91012 JASs forum Journal of Anthropological Sciences Vol. 91 (2013), pp. 239-244 Paleogenomics, hominin interbreeding and language evolution Antonio Benítez-Burraco1 & Lluís Barceló-Coblijn2 1) Departamento de Filología Española y sus didácticas, Universidad de Huelva, Campus de “El Carmen”, Avda. de las Fuerzas Armadas s/n, 21071 Huelva, Spain e-mail: [email protected] 2) Universidad de Murcia, Departamento de Filosofía, Campus de Espinardo, E-30100 Murcia, Spain Limited admixture between other hominins (Enard et al., 2002), which seemingly predate to and anatomically-modern human (henceforth, our common ancestor (who is also the ancestor of AMH) populations has recently been con- Denisovans), about ≈ 300-400 kya. FOXP2 is the firmed. Non-African human genomes contain ‘language gene’ par excellence (Vargha-Khadem ≈ 1-4% of Neanderthal DNA (Green et al., et al., 2005; Fisher & Scharff, 2009; Benítez- 2010). Melanesian genomes additionally con- Burraco, 2012). Not surprisingly, this find was tain ≈ 4–6% of Denisovan DNA (Reich et al., regarded as the definitive piece of evidence for 2010). Approximate-likelihood analyses of some the Neanderthal language case (Trinkaus, 2007; Sub-Saharan African genomes are suggestive of d’Errico & Vanhaeren, 2009, p. 38; Frayer et the presence of ≈ 2% of archaic DNA, plausi- al., 2010, p. 113). Can we be confident that the bly introgressed from an extint hominin popula- analysis of the DNA introgressed from archaic tion that split from our ancestors about 700 kya populations into AMHs actually sheds light on (Hammer et al., 2011). It has been argued that language evolution and the linguistic abilities of AMH cognition could have been partially mod- late hominins? We think that at this stage, cau- elled by this kind of introgression events (Hawks tion is in order. et al., 2008). Linguistic abilities seem a natural To begin with, all human languages appear to target, in view of the fact that language has been share some basic structural properties (Chomsky, traditionally regarded as a species-specific trait. A 1980; Baker, 2001; Boeckx, 2009; among many long-standing controversy divides the field con- others). Moreover, language acquisition is a pro- cerning the possibility that language is a synapo- cess universal to the species (Lust, 2006; Slobin, morphic trait in late hominins, and particularly, 2006). Taken together, this suggests that all that Neanderthals already had it (Mellars, 1996; human beings are endowed with the same capac- d’Errico et al., 2003; Mithen, 2006). Genetic ity for language. Given the human biological pre- evidence has fuelled this debate. Different ‘lan- disposition to acquire a language, the ‘linguistic guage-related’ genes (i.e. genes that give rise to genotype’ must be similar in all human beings language disorders in our species when they are as well. Consequently, since each AHM popu- mutated) have been positively selected in our lation incorporated different species-specific clade (e.g. Taipale et al., 2003, and Hannula- DNA portions, we should expect that this ‘lin- Jouppi et al., 2005 for some candidate genes for guistic genotype’ is not part of the introgressed dyslexia). Crucially, Krause et al. (2007) found DNAs. Moreover, this would imply that these in Neanderthals the derived alleles of the two species were endowed with it, and, ultimately, human substitutions in the FOXP2 protein had modern, human-like language. However, the JASs is published by the Istituto Italiano di Antropologia www.isita-org.com 240 JASs forum: Paleogenomics, Hominins and language this is controversial, even for Neanderthals (see and MCPH1, two genes related to brain size evo- above). The opposite possibility is also appeal- lution (Zhang 2003; Evans et al., 2004). In the ing: extinct pre-modern hominins would have same vein, the Denisovan CNTNAP2 shows a gained (or improved) their linguistic abilities fixed single nucleotide change compared to that when they interbred with AMHs. Although cur- of AMHs (Meyer et al., 2012). CNTNAP2 is one rent analyses have only detected a gene flow from of FOXP2 targets (Vernes et al., 2008). It has the former into the latter, but not vice versa, a been linked as well to specific language impair- reciprocal gene flow is certainly expected from ment (SLI) (Vernes et al., 2008), autism (Alarcón the colonizing population to the resident popu- et al., 2008), and diverse clinical conditions in lation (see Green et al.,2010 for a discussion). which language is disordered (Petrin et al., 2010; Nonetheless, that possibility is also problematic. Sehested et al., 2010). Finally, the derived vari- Firstly, evidence of modern language in other ants of all regulatory mechanisms of gene expres- hominins is, once again, controversial. Secondly, sion (and in fact, of the whole interactomes) the introgression events do not coincide with should have been introgressed as well. The state significant cultural changes. Allegedly it is mod- (derived or ancestral) in other hominins of the ern language that fuels constant and feedback, regulatory networks of genes that are relevant AMH-specific cultural changes, because it allows for language is currently unknown. However, we to virtually explore new options and also to trans- know, for instance, that some of the physiologi- mit the results in a efficient, quick, and instan- cal targets of FOXP2 (whose mutation gives also taneous fashion (Dennett, 1995, 1996). As for rise to language disorders) have been positively Neanderthals, the admixture plausibly predates selected in our clade (Spiteri et al., 2007; Vernes the emergence of modern-like cultures among et al., 2008; see above). Moreover, some differ- them. Châtelperronian and related technocom- ences have been attested as well concerning cis- plexes only emerged ≈ 40 kya (d’Errico et al., regulatory regions of gene expression. For exam- 1998; Langley et al., 2008). However, according ple, Maricic et al., (2012) have recently found to Green et al. (2010) the admixture took place an AMH-specific substitution within a regula- ≈ 50-100 kya. In fact, they found that the exem- tory region of FOXP2 which was known to have plar from Mezmaiskaya, who lived ≈ 60-70 kya been affected by a selective sweep; this substitu- (Golovanova et al.,1999), is genetically similar to tion is likely to alter FOXP2 expression via the the specimens from Vindija and El Sidrón (we transcription factor POU3F2. Eventually, as the are not considering here the possibility that these analysis of the Neanderthal genome suggests (see ‘modern’ assemblages are not genuine innovations Green et al., 2010 supplementary material for [Coolidge & Wynn, 2004; Mellars, 2005], or discussion), we expect hundreds of amino-acid cannot be attributable to Neanderthals [Bar-Yosef sequence changes to be fixed in the AMH line- & Bordes, 2010; Higham et al., 2010]). Thirdly, age after the divergence from Neanderthals and the introgressed DNA should have contained Denisovans, a greater number of potential regu- most (if not all) derived variants of the ‘language latory substitutions, and also some fixed changes genes’ that were fixed after the split of our lineage in human accelerated regions. It could even be from the line that gave rise to Neanderthals and the case that a differential activity of transpos- Denisovans. The problem is that these genes are able elements or of viruses has modelled their scattered throughout the whole genome (Smith, respective genomes in dissimilar ways (Agoni et 2007; Benítez-Burraco, 2012). More impor- al., 2012). We cannot rule out the possibility that tantly, for some of them we have direct evidence these specifically-human innovations have also (see Green et al., 2010 supplementary material) played an important role in the development of that both species exhibited the ancestral alleles. neural devices involved in language processing. For instance, Neanderthals were endowed with Further, even if we could eventually prove that the ancestral alleles in some positions of ASPM pre-modern hominins had the derived version JASs forum: Paleogenomics, Hominins and language 241 of the ‘linguistic genotype’ (partly due to these system’ and a similar ‘speech’. However, the introgression events), a Homo sapiens-like linguis- available data suggests that the ‘languages’ they tic phenotype is not granted. There is not a direct plausibly spoke would have lacked some defining link between the genotype and the phenotype. properties of human languages, particularly, com- Development is always synergistically regulated plex syntax, which is strongly based in recursive by multiple factors, which are all equally neces- embedded structures, but exhibits as well other sary (Oyama, 2000; Robert, 2008). Most of distinctive features, such as cross-serial depend- them remain substantially unknown in extinct encies (see Mellars, 1996 and Mithen, 2006 for hominins. Different genotypes can give rise to similar conclusions regarding Neanderthal ‘lan- the same phenotype, but the other way around guage’). Other evolutionary mechanisms must also holds (Pigliucci et al., 1996; West-Eberhard, account for its presence in our species only. 2003; Balaban, 2006). Neanderthals, Denisovans, and HAMs evolved in different environments (Finlayson, 2005; Carrión et al., 2011). Hence, Acknowledgments the norms of reaction of their ‘linguistic geno- types’ could have been different as well. Notice Antonio Benítez-Burraco’s research was funded by also that genes are more important during the the Spanish Ministry