In their target article Ackermann et al. make a valiant attempt to assemble a comprehensive account of the origin and neural organization of human speech on the basis of arguments confined by and large to comparative primatology. The nature of their topic is ill-suited to such an approach, because at its core lies a behavioral adaptation and corresponding neural mechanism which we share with some species of cetaceans, pinnipeds, and birds, but not with any nonhuman primate. For such a situation, the comparative method offers analogy instead of homology as guiding concept (e.g., the elucidation of body form in cetaceans is better served by turning to distant fishes rather than to far closer relatives among extant mammals).

The capacity in question is the ability to learn to reproduce, by voice, patterns of sound first received by ear. This capacity is of singular biological uselessness except in special cases, one of which happens to be us humans, because every word and phrase we know how to pronounce has become ours by such means. Technically, the capacity is known as vocal production learning (Janik & Slater Reference Janik, Slater, Slater, Rosenblatt, Snowdon and Milinski1997; Reference Janik and Slater2000), and though the concept does occur in the target article, it is more by way of an afterthought than as a principal pivot of analysis.

Putting vocal production learning at center stage removes the mystery of the “speechlessness” of even our closest primate relatives rightly emphasized by Ackermann and colleagues. Lacking the vocal learning mechanism (Janik & Slater Reference Janik, Slater, Slater, Rosenblatt, Snowdon and Milinski1997), they naturally cannot do that which inherently is dependent upon it, namely, learn to pronounce words and phrases of rather arbitrary phonemic composition. That vocal learning is, in fact, the crux of the matter is demonstrated by the ease with which numerous species of parrots and other mimics among the birds do what no chimpanzee has ever done: acquire a substantial repertoire of human words and phrases pronounced with a fidelity that fools the human ear (Nottebohm Reference Nottebohm, Harnad, Steklis and Lancaster1976).

The diction of bird mimics tells us that the entire pronunciatory part of the speech equation is a matter of being a vocal learner. Step 1 on the path to speech is accordingly to come into possession of the capacity for vocal learning. This first step, moreover, provides a plausible evolutionary context for the first step invoked by Ackermann et al., namely, the addition of direct (monosynaptic) cortical efference to lower brainstem motor nuclei controlling larynx, pharynx, tongue, and lips.

The species distribution of such direct connections (to which can be added direct cortical innervation of the nucleus retroambiguus for respiratory control) suggests that they evolve specifically for cerebral fine control of respiration and vocalization and not (as the target article assumes) as a general concomitant of brain expansion (Arriaga & Jarvis Reference Arriaga and Jarvis2013; Fitch et al. Reference Fitch, Huber and Bugnyar2010; Iwatsubo et al. Reference Iwatsubo, Kuzuhara, Kanemitsu, Shimada and Toyokura1990; Jürgens Reference Jürgens2002a; Kuypers Reference Kuypers1958a; Reference Kuypers1958b; Merker Reference Merker2009; Okanoya & Merker Reference Okanoya, Merker, Lyon, Nehaniv and Cangelosi2007; Okanoya et al. Reference Okanoya, Hihara, Tokimoto, Tobari and Iriki2007; Wild Reference Wild1993; Reference Wild1997).

As suggested in a previous BBS commentary (Merker Reference Merker2009), it is even conceivable that the “simple” addition, in ancestral Homo, of a direct primary motor cortex efference to those medullary motor nuclei sufficed to recruit the already present cerebral territories centered on Wernicke's and Broca's areas (see Fig. 12.4 of Falk [Reference Falk, Henke and Tattersall2007] for putative homologs in Pan and Macaca; see also Neubert et al. Reference Neubert, Mars, Thomas, Sallet and Rushworth2014) to the practice-based acquisition of complex vocal output matching auditory models, thus making our ancestors vocal learners.

The most common use of vocal production learning in nature is as a means to impress potential mates and rivals by mastery of a complex song tradition (for the evolutionary logic, see Merker [2012] and review by Spencer & MacDougall-Shackleton [Reference Spencer and MacDougall-Shackleton2011]). Humans are a singing species (von Humboldt Reference von Humboldt, Buck and Raven1836/1971), so the default assumption would be that the vocal learning capacity of our ancestors was exercised for similar purposes. If so, they were maintaining traditions of intergenerationally transmitted and culturally learned vocal lore (song) long before that lore became verbal by being semanticized.

Such a situation brings the “learner bottleneck” principle of iterated (transgenerational) learning into play (Kirby Reference Kirby and Briscoe2002). In the non-human examples the narrow focus of song displays on potential mates and rivals limits the operation of that principle to effecting progressive refinement of the formal properties of songstrings (their purely formal syntax; see Kirby et al. [Reference Kirby, Cornish and Smith2008] and references therein). Should, however, circumstances spread singing to the full range of daily and seasonal activities, the same principle would ensure a gradual and progressive differentiation of song repertoire by behavioral context, amounting to an implicit assortative contextual semanticization of the songstrings repertoire as a whole (Merker Reference Merker and Bannan2012; Merker & Okanoya Reference Merker, Okanoya, Lyon, Nehaniv and Cangelosi2007).

Therein lies the point of departure for a gapless path to human language, details of which are presented in a recent publication of mine (Merker Reference Merker and Bannan2012). At the point at which that path is about to arrive at fully instrumental language, it reaches an impasse. Attempts to use semanticized songstrings in a “displacement” mode of reference (Hockett Reference Hockett1960) would undermine the very contextual basis underpinning the semantic meaning of strings. To overcome that hurdle, I have postulated a naturally (as opposed to sexually) selected enhancement (expansion) of the cerebral storage capacity hosting the vocal learning mechanism, a capacity increase driven by individual self-interest in reaping the benefits of instrumental uses of songstring semantics.

Such an expanded storage capacity would allow string (and gestural) markers for communicative intent to be appended to the songstring repertoire by learning, launching it upon fully instrumental language use. Perhaps the FOXP2 enhancement of cortico-basal ganglia function in the human line provided the required extra storage capacity. As far as is known, it affects relevant neurons at the microscopic and functional level, promoting lengthening of dendrites (potentially increasing synapse numbers) as well as affecting synaptic plasticity by enhancing long term depression (Reimers-Kipping et al. Reference Reimers-Kipping, Hevers and Pääbo2011). Since such changes are general for the basal ganglia as a whole (along with associated thalamic and cortical domains), they fit better with a general expansion of storage capacity than with a remodeling of lateral interactions among circuits serving components of articulate speech assumed in the target article.