Many have criticized Chomsky’s theory of the Universal Grammar of language (e.g., Pinker as described in Sihombing 2022), but the most effective criticisms have come from Daniel Everett, given that Chomsky (according to Everett) has never addressed the criticisms. Everett has two issues with Chomsky’s theory of language: the evolutionary timeline of language for Homo sapiens[1] and the lack of universality of language structure for all languages. On the evolution of language, Chomsky has proposed that language began some 60,000 years ago (Chomsky 2012). Everett’s contrary explanation (Everett 2017) is that rudimentary language started 2.5 million years ago in the South Pacific amongst Homo erectus, who are estimated to have had 62 billion neurons (24 billion short of Homo sapiens, Herculano-Houzel 2012), and for whom there is archeological evidence that they were skilled sailors with territories throughout the south Pacific Ocean; navigation between territorial islands was done using the stars and sea currents, which would have required some form of communication between group members (Everett 2017)[2]. Also, at the time of Homo erectus, there is evidence of an asteroid strike in the South Pacific, which could have accelerated the evolutionary process (as it did from mammals 64 million years ago) by bringing about the introduction of large, big-brained primates.

On the generalizability of Chomsky’s theory to all languages (including primitive languages), Everett (2006, 2016) spent many years in the Amazon basin of Brazil studying the Pirahã people, who have no written language or number system. To transmit their history across generations (two at most) it is all done by word-of-mouth. The language has eight consonants, three vowels, and two tones. The sentences are very simplistic, with no embedded clauses such as, “John, who is a hunter, is an active individual.” Instead, the utterance would be: “John is a hunter. John is an active individual.” This language structure is apparent when children or adults begin to learn a language (thereby having no recursive structure). Also, the language has no pronouns. Furthermore, it has a proximate tense (e.g., for the present) and a remote tense (e.g., for the past) but no perfect tense, a tense with no time stamp, e.g., “I have prepared some food.” The language does not permit the establishment of a creation myth. The sense of time, e.g., historic time, is not well developed. Much is set in the present. Hunting and foraging are a daily affair for the Pirahã people. The children are taught the names of all the plants and animals in the jungle, which can number in the thousands.

Accordingly, Chomsky’s theory fails to account for the evolutionary history of language. As well, his theory accounts only for complex, recursive languages with little to say about the more primitive languages such as the one spoken by the Pirahã people of Brazil. It is noteworthy that if a Pirahã child is raised in Sao Paulo in the Portuguese language, the child will have no problem mastering all the complexities of Portuguese, which has way more verb tenses than English but a similar number system, as well as a comparable written script.

Neanderthals (Homo neanderthalensis), who occupied Northern Europe for much of their existence up until 40,000 years ago (Sansalone et al. 2023), were around when Homo sapiens were endowed with an ability to generate speech sounds and therefore able to express their cognition (Chomsky 2012). Doreen Kimura, who spent most of her life studying how the brain processes human language by examining brain-damaged patients (Kimura 1993), believed that human language does not represent some type of species exceptionalism, but instead represents a characteristic of the brain and the body that was shaped by evolution thereby leaving traces of its genetics (e.g., Chomsky’s universal grammar) in other species such as the electric fish, song-birds, bats, elephants, dolphins, whales, and non-Homo sapiens. She argued that communication of early Homo sapiens some 500,000 years ago was non-verbal and gesture-based, but changed to the vocal apparatus at this time (i.e., by the formation of a right-angled vocal tract, see Fig. 1.1 Kimura 1993) allowing for the utterance of vowels. This notion runs contrary to the idea that some 60,000 years ago humans just started producing language spontaneously (Chomsky 2012), with no clear link to evolution, brain, and behavior despite many challenges to this idea (Bizzi and Mussa-Ivaldi 1998; Changizi 2001b, 2003; Dawkins 1976; Dawkins and Dawkins 1976; Everett 2017; Fentress and Stilwell 1973; Gallistel 1980)[3].

As for Neanderthals, Sansalone et al. (2023) have recently opined that the Neanderthal neocortex was as sophisticated as the human neocortex exhibiting a high degree of interareal integration, which does not exist in other primates. The overlap between Neanderthals and Homo sapiens 40,000 years ago permitted the sharing of genes between the two groups. A common language would have facilitated their genetic intimacy, and there is evidence that Neanderthals and Homo sapiens exchanged genes, for over 6% of the genomes of Europeans are Neanderthal. Perhaps before their extinction, Neanderthals possessed Chomsky’s universal grammar. This could be verified by evolutionary biologists.

Another issue is that Chomsky’s theory emphasizes the rapid acquisition of language during childhood, which Chomsky attributes to a universal grammar programmed genetically in all humans (Chomsky 1965). A child does not need to spend time in school to master the verbal aspects of a language, which is acquired automatically between birth and adolescence, but reading and writing necessitates schooling. FOXP2 gene expression occurs in new-born humans and in new-born and adult songbirds for the accelerated acquisition of language and songs, respectively (Rochefort et al. 2007). This acquisition is mediated by neurogenesis in the telencephalon (neocortex of humans) and the hippocampus (Goldman and Nottebohm 1983; Rochefort et al. 2007); neurogenesis ceases by the age of twelve in humans (Charvet and Finlay 2018; Sanai et al. 2011; Sorrells et al. 2018). Neurogenesis may accelerate language learning in children, whereas it promotes the learning of songs for mate selection in adult songbirds.

One might expect that the number of new words learned as a child should be much greater than the number of new words learned after the age of 10 to 12 when neurogenesis begins to subside (Charvet and Finlay 2018; Sanai et al. 2011; Sorrells et al. 2018). According to Bloom and Markson (1998) by the age of ten, children learn an average of 23,651 words to yield an acquisition rate of 2,365 words per year, and from the age of ten to eighteen children learn an average of 36,350 additional words to yield an acquisition rate of 4,544 words per year (this is based on children who attend school). Now some of this increase in acquisition rate after the age of ten may be related to a child having more methods by which to consolidate information; on this point, the ability to speak, read, and write tends to accelerate after the age of ten, which should contribute to the efficiency of word consolidation and retrieval. Nonetheless, no one would argue that language acquisition (through speaking and hearing) up to the age of 10 or 12 is relatively effortless and word and phrase utterances are free of any accent (other than the parents’/teachers’ accent) even when learning multiple languages. These points are emphasized by Chomsky (1959) and used effectively to challenge Skinner’s Verbal Behavior Theory of language (Skinner 1957).

Lastly, an analysis of 19 different languages including European and Asian languages revealed that the information transmission rate is comparable for all the languages at about 39 bits per second (Coupé et al. 2019). This means that the brain sets the same limits on language irrespective of language type, which bolsters Chomsky’s notion that there is a neuro-genetic structure in humans that controls the universal acquisition of language (Chomsky 1965).

Summary:

1. That language was acquired by Homo sapiens as late at 60,000 years ago may not be correct, since there is evidence that Homo erectus (an ancestor of Homo sapiens) some 2.5 million years ago may have required this capability to organize communications to navigate between territories in the south Pacific Ocean.

2. The theory of Universal Grammar does not account for all languages, particularly languages that have no recursive structure, such as the language of the Pirahã people of the Amazon. Nevertheless, advanced languages have a comparable information transfer rate, and Pirahã children can learn a recursive language, which suggests that all humans are genetically endowed with a common neural mechanism for the acquisition of language.

3. A universal grammar may be represented in non-human species. There is evidence that Neanderthals had a brain as advanced as that of Homo sapiens and therefore this species could have supported human-like language.

4. More English words are learned after the age of ten than before the age of ten, even though neurogenesis stops by or shortly after this age in humans. This, however, does not take away from the fact that before the age of ten children learn to speak effortlessly and without an accent, a point emphasized by Chomsky.

Footnotes:

[1] Chomsky is not sure whether language is affected by natural selection, since when asked questions about this he never gives a clear yes or no on the topic (Chomsky 2020-2023/Youtube).

[2] Soccer robots have both proprioception to note the position of their bodies as well as a visual sense to detect the ball, the goals, and the position of the other robots (Behnke and Strucker 2008). To communicate the location of the ball and other items with other robots, an allocentric coordinate system is used, much like that utilized by a group of electric fish (who use electricity to communicate), a pack of wolves (who use gestures and sounds to communicate), or a pod of killer whales (who use sounds to communicate) in pursuit of prey. Language may have evolved to enhance allocentric communication, as is required by soccer robots.

[3] For example, when an estimate is made for the value of ‘d’ (a word-syllable quotient) between the number of words (E) and number of syllables (C) using the formula ‘E = Cd’ (derived from Changizi 2001b), the value for ‘d’ turns out to be ~ 1.046 for humans [i.e., there are approximately 170,000 words of common usage in the English language and there are approximately 100,000 corresponding syllables, which yields ‘d’ = 1.046 (the ‘E’ and ‘C’ values are based on the full, 20 volume, Oxford English Dictionary)]. This means that for the English language words and syllables have a combinatorial relationship, namely, for every one word there is an average of 1.046 syllables. Now what about birdsong? Much like for human language, the number of birdsongs (E) varies as a function of the number of syllables (C), such that the number of syllables per song, ‘d’, is estimated to be 1.23 (Changizi 2001b), which is even greater than the combinatorial estimate for human language of words to syllables.