(born Düsseldorf 1921; died White Plains, N.Y., 1975)
Biological research on language seems to be paradoxical to the extent that we have to acknowledge that languages consist of arbitrary cultural constraints.
—Lenneberg, Biological Foundations of Language
Ideology is science’s worst enemy. It creates expectations of data, makes the nonexistent seem real, and in the end leads straight to bitter disappointment. In Life and Destiny Vassily Grossman prepares the reader for the systematic and perverse annihilation of humanity by totalitarian regimes in his description of how the roads, railways, and power lines of the Nazi concentration camp imprison everything with their geometric rigor: “It was a space filled with straight lines; a space of rectangles and parallelograms that cut through the foggy autumn sky.” If we don’t take into account the rough-edged and irreducible facts, if we let ideology prevail over data, we risk paralysis. This was where linguistics was in the mid-twentieth century: through a desire to impose at all costs the notion that languages are “arbitrary cultural conventions,” work on the biology of language was effectively ruled out.
In his treatise on the biological foundations of language, Eric Lenneberg felt obliged to include this prefatory caveat. Was this really necessary? It wouldn’t be today, thanks to the fact that Lenneberg concluded his work by demonstrating that recovery from aphasia, if it was possible, had different outcomes depending on whether the lesion happened before or after puberty. If the lesion had been before puberty, then the chances of recovery were better, and, tellingly, recovering patients tended to follow the same developmental path as children acquiring language spontaneously. If the lesion was postpuberty, patients tended to recover less, and the recovery typically followed a chaotic path. Language acquisition is thus sensitive to a so-called “critical period” in which latent abilities are either stimulated or lost, as is well known to be the case with many other biological phenomena we understand much better because they can be studied in animals. Now it would certainly seem strange to say that the age of puberty was an “arbitrary cultural convention” (although of course it can vary as a function of complex variables to do with nutrition and social factors), so the idea that language in general is an arbitrary convention becomes somewhat implausible.
However, Lenneberg’s quotation cannot be put aside so easily. English usage sets a trap (which can be avoided in various other languages, including Italian), since we use a single word, “language,” for both specific tongues and language in general (where Italian, for example, uses lingua in the first sense and linguaggio in the second). In English the two senses are only distinguished in the plural, so we interpret Lenneberg to mean that research on language in general appears to be paradoxical because individual languages must be taken to be arbitrary cultural conventions. This is the fundamental point and the real novelty in relation to the nineteenth-century research tradition that continued into the latter part of the twentieth century (Bambini 2012): the new, central question is not the extent to which language in general, in the sense of the capacity to communicate, depends on the brain (this is considered to be firmly entrenched) but if the structure of human language does, that is, whether particular languages, and in particular their syntax, do (since syntax marks the watershed between the human communication system and those of other animals).
Despite that fact that ideological prejudices are never satisfied but rather tend to become more disingenuous, since they admit that science is something more than a method (the opium of the people, perhaps), the results are encouraging. Among the most solidly confirmed results are those obtained from neuroimaging techniques, as long as they don’t give way to a tendency toward a new phrenology. I mention this because it’s important to underline that this type of research doesn’t consist in finding neuropsychological correlates to specific activities in order to produce a functional map of the brain, but rather in understanding how it is possible for the obviously modular structure of the brain to give rise to the cognitive and behavioral activities we observe—a step analogous in certain respects to what happened earlier in linguistics in phonology and then in syntax with the abandonment of taxonomies in favor of decomposition of units into more primitive elements. The other risk, of course, consists in translating the formal generalizations of linguistics into neuronal mechanisms: this is a premature goal, and it’s not even clear that it can ever be reached without radically changing our ideas about neuronal mechanisms. On the other hand, what we can do and what is in fact being done is to check whether certain very general properties of the structure of language discovered by formal approaches are reflected in dedicated neural networks. In this area, as I said above, the results are encouraging. It has been shown that the brain is not at all neutral to the type of grammar it is exposed to. For example, if exposed to a syntax without recursive rules, the activity of the language circuits diminishes gradually in favor of other circuits that are typically implicated in solving nonlinguistic problems. But this tells us nothing about the neuropsychological algorithms working at the level of neuronal mechanisms that make this distinction possible (Moro et al. 2001; Tettamanti et al. 2002, 2008b; Musso et al. 2005; Friederici et al. 2006; Moro 2013, 2015; Dehaene et al. 2015). The old idea that the brain is hardware that runs various kinds of software—grammars, but also other cognitive capacities—is thus completely outmoded (Di Francesco 2002): software—i.e., the grammars of different languages—is really to be thought of as an expression of the hardware, determining its structural limits.
The limits to Babel thus exist and are inscribed in our bodies. Obviously, that doesn’t mean that we’ll manage to reduce language comprehension to neurological mechanisms (an idea that I personally do not even understand); but we might be able to see the extent to which neurological mechanisms are isomorphic to the structures of natural language (see Poepple 1996; Moro 2015; Dehaene et al. 2015 and references cited there), if at all, and consequently to define and circumscribe the limits to variations of these mechanisms, in order to be able to define, on neurobiological grounds, the class of possible languages. In other words, we could arrive at a “rational anatomy” of grammar. This is certainly not much, perhaps nothing at all, in relation to all that the totality of language means to us humans: poetry, curses, promises, prayers, sweet nothings, and jokes all fall outside of this project. But there are no reasonable alternatives.
It’s as if we were in the position of someone who, asked to define a caress, could only describe the structure of the hand and the angles of rotation of the bones, muscles, and tendons that constitute it: it wouldn’t be easy to distinguish it from a slap. But if nothing else, at least it wouldn’t be confused with a kick, and that’s a start.
