Much has been made of the idea that humans are genetically programmed to learn languages at an early age, suggesting that learning plays a minor role in this process (Chomsky 1959). But we have argued that a large part of being able to speak at an information transfer rate exceeding 40 bits per second (i.e., over a trillion possibilities per second, Coupé et al. 2019; Reed and Durlach 1998) is due to having a one-decade-long formal education in one’s native and secondary languages (Tehovnik, Hasanbegović, Chen 2024). For example, Joseph Conrad, whose native language was Polish and who became a world renowned writer, in his 20’s learned to write in English (Wikipedia/Joseph Conrad/July 11, 2024). In what is now Poland, Conrad was mentored by his father, Apollo Korzeniouwski, who was a writer and later a convicted political activist by the Russian Empire. To escape the political turmoil of eastern Europe, Conrad (to the dislike of his father) exiled himself to England, which marked the start of his writing career. And the rest we know about: ‘Heart of Darkness’, ‘Lord Jim’, ‘Nostromo’, and so on.
The study of second language learning by 20 year olds was investigated by Hosoda et al. (2013). They recruited twenty-four Japanese university students who were serially bilingual with the earliest age of learning English at seven years of age. The students completed a 4-month training course in intensive English study to enhance their vocabulary. They learned 60 words per week for 16 weeks yielding a total of 960 words, which translates into an information transfer rate of 0.0006 bits per second (see Footnote 1), which is appreciably lower than the transfer rate of ~ 40 bits per second for producing speech (Coupé et al. 2019; Reed and Durlach 1998).
Furthermore, there is this belief that learning a language is accelerated in children as compared to adults (Chomsky 1959). By the age of eighteen, one can have memorized some 60,000 words in the English language (Bloom and Markson 1998; Miller 1996), which represents an information consolidation rate of 0.0006 bits per second (see Footnote 2), which is the same as the rate experienced by the Japanese students learning English as a second language as adults (Hosoda et al. 2013).
Two conclusions can be drawn: First, consolidating a language is many orders of magnitude slower than delivering a speech (i.e., 0.0006 bits per second vs. 40 bits per second). Second, the idea that children learn languages at an accelerated rate may not be true. This needs to be properly investigated, however, whereby the rate of language learning (in bits per second) is measured yearly starting neonatally and ending in adulthood. Also, there is more to language than just memorizing words, so linguists will need to design experiments covering all the major parameters of language and express these parameters in terms of bits per unit time. It is time that linguistics (like neuroscience) becomes a quantitative discipline.
Footnote 1: Bit-rate calculation: if each word is made up of 4 letters (on average) then the bit rate of learning (using values by Reed and Durlach 1998) = 1.5 bits per letter x 4 letters/word x 960 words/16 weeks = 360 bits per week = 0.0006 bits/sec. The learning period includes not only the time spend memorizing the words, but also the time required to consolidate the information in the brain, which occurs during sleep and during moments of immobility (Dickey et al. 2022; Marr 1971; Wilson and McNaughton 1994). After the learning there was an increase in the grey matter volume of Broca’s area, the head of the caudate nucleus, and the anterior cingulate cortex; as well, there was an increase in the white matter volume of the inferior frontal-caudate pathway and of connections between Broca’s and Wernicke’s areas (Hosoda et al. 2013). The grey and white matter enhancement correlated with the extent of word memorization.
Footnote 2: Bit-rate calculation: Memorizing 60,000 words in 18 years translates into 360,000 bits of information [i.e., 60,000 words x 4 letters per word x 1.5 bits per letters, Reed and Durlach 1998] or a word consolidation rate of 55 bits per day (or 9 words per day) over eighteen years of life. Therefore, the rate per second is 0.0006 bits per second. For other details see Footnote 1.
In fact I should say that children acquire and not learn languages much faster because learning goes conscious effort while acquisition is unconscious. The reason is that children from 0 to 18 years old have their Language Acquisition Device (LAD)active and for that acquiring any spoken language around them is unconscious while from 18 years old and above, the LAD is no more active, and for that matter, one need learn a language consciously.
Thank you for your comment, Dr. Sibite. After writing the piece in which I suggest that the information transfer rate of consolidation of children and adults is similar based on my back-of-the-envelope calculation for the consolidation of English words (at about 0.0006 bits per second in both children and adults), I spent several hours reviewing past lectures by Noam Chomsky. I listened to his 1.5-hour lecture on language as posted on YouTube and delivered at MIT in 2019 before a group of students and colleagues. Once his lecture was over, he was asked a question by a student about how neurons carry out the “genetically endowed” language computations [by the Language Acquisition Device] in children/adults (since such details were largely absent from his lecture). He went on to suggest that the neurons of the brain are much too slow to execute the computation (called Merge) that he is talking about and that computations must be carried out at the subcellular/molecular level as suggested by Roger Penrose (and studied by Eric Kandel 2006).
On the point of speed of processing, we understand that the transmission duration of a chemical synapse is about one millisecond and that it can take tens of milliseconds for signals to be transmitted from caudal to rostral parts of the neocortex (Schiller and Tehovnik 2015; Yeomans and Tehovnik 1988). Therefore, Chomsky must be thinking in the micro-second range. Whether this microseconds duration of computation time would render the linguistic signal unconscious is unclear; note that the duration of consciousness/ thinking is estimated to occur in the millisecond to second range (Varela 1999ab; Dwarakanath, Logothetis 2023).
If a neocortical hemisphere is damaged in children, language ability can be assumed by the undamaged hemisphere (Olulade et al. 2020), substantiating that there is enhanced ‘linguistic’ neuroplasticity in children that is lost in adults (Kumura 1993; Penfield and Roberts 1966). Also, many faculties of the brain need to be tuned by the environment during a critical period and if missed a faculty will not develop (Fine et al. 2003; Hubel and Wiesel 1977). In the case of language, the first year of life is critical for the establishment of syntax, and other language attributes develop before puberty (Friedmann and Rusou 2015).
As to whether the development of language faculties is largely an unconscious affair is now addressed. In all mammals, the neocortex mediates the storage of declarative information, which depends on consciousness (or thinking) to guide the learning process (Hebb 1949, 1968). Every day while learning to ride a bike my son would ask: “Daddy, are we going to train today?” This question was repeated for months until one day my son’s vestibular system was finally programmed to race around the neighborhood track; years after learning to ride a bike my son had no recollection of how this occurred, which makes one assume that consciousness had little to do with the learning process.
The establishment of an automated state (or an unconscious execution) requires extensive training, even for language. Formal, global education (since the 1950’s) has been central in elevating humankind out of poverty (the proof: travel to Nigeria, Brazil, India, or China today and compare your social experience with what it was like many decades ago, e.g., the 1980’s). To automate a behavior requires daily training so that circuits between the neocortex and cerebellum can be programmed (which is what putatively happens while learning to ride a bike, Tehovnik, Hasanbegović, Chen 2024; also see Miles and Lisberger 1981). Unlike the neocortex, the cerebellum is responsible for developing the executable motor code for all behaviors, including language. Once the neocortex and cerebellum are tuned for a given behavior, a reduced amount of neural tissue is used to summon a correct response as when engaged in dialogue with a fellow interlocuter (Lehericy et al. 2005; Ojemann 1983). The exchange is so rapid that tens of milliseconds before the completion of a string of utterances, one is prepared to deliver a reply (Levinson and Torreira 2015). Interestingly, patient HM whose hippocampus was destroyed could still engage in rapid conversation, but in the absence of his hippocampus he was unable to update his declarative memories, i.e., he could not learn any new words or new facts and if asked to recall what his mother or father were like he could not narrate the history (Corkin 2002). In short, the unconscious execution of speech entails using a reduced amount of neocortical tissue with the remaining tissue activating cortico-cerebellar loops (Hasanbegović 2024) to maintain a particular state of automaticity. Any upgrades to the cortico-cerebellar loops, however, would require a modification of the efference-copy code at the level of the cerebellum (Bell et al. 1997; Chen 2019; Cullen 2015; De Zeeuw 2021; Fukutomi and Carlson 2020; Loyola et al. 2019; Miles and Lisberger 1981; Noda et al. 1991; Shadmehr 2020; Tehovnik, Patel, Tolias et al. 2021; Wang et al. 2023).
As for Chomsky’s ‘Merge’ to be expressed automatically in children (Chomsky 1965), it would require that the cortico-cerebellar loops be programmed genetically such that all the gains at the Purkinje neurons are able to anticipate a linguistic world with minimal adjustments of the gains once a child begins hearing and making sounds. That all this is finalized syntactically by the age of one, will need to be verified quantitatively using information theory (Tehovnik and Chen 2015).
I do not agree with idea that the children are unable to learn language faster. I total agree with Noam Chomsky (1959) that the children learn language faster than adults. The idea of Edward J Tehovnik that the children learn languages at an accelerated rate (Chomsky 1959 may not be true? It is not true. According to my observations children and women learn language faster than adult men. For instance, the children who are born in UK, U.S. Canada, Australia or New Zealand were found that they have learn language faster and master it than their parents. The children can memorize easily than adults.
From the age of two, humans can learn up to 10 words per day or up to 14,000 words by the age of six (Yang 2006). From age 6 to 8, the average child in school is learning 6–7 words per day, and from the age of 8 to 12, approximately 12 words per day (Bloom and Markson 1998). Therefore, the acquisition of words is not accelerated during early childhood. Syntax (through a process called merge, Chomsky 1965) is central to establishing a symbolic system—namely, memorizing words is necessary but not sufficient for learning a language, as a Japanese colleague of mine discovered after entering the United States armed with a vocabulary of twenty thousand words, but with no means by which to link the words for speech. This person remained verbally illiterate in the English language for the duration of his stay at MIT. The critical period for the development of syntax is believed to occur before the age of one (Friedmann and Rusou 2015), but how the merging of symbols (i.e., words) is done before and after the age of one by the brain is not well understood. It is possible that the lack of apparent effort to learn a language by infants has been conflated with the noticeable effort made by adults (Chomsky 1959). Therefore, information theory (expressing knowledge in terms of bits per second) will need to be used to quantify linguistics to better assess whether language aptitude is truly accelerated in infants.
Dr. Edward J Tehovnik : If I may add a suggestion, consider the emergence of a new sign language in Nicaragua in the late 1970's/early 1980's. There has been considerable discussion of what happened there in the literature. In a nutshell: a large group of deaf people were brought to school that had been organized for them for the first time in Nicaragua. These people didn't have any sign language whatsoever - they just used very simple home signs to communicate with their families. When they were at school, they wanted to comunicate with each other, so they created a new sign language from scratch, with no explicit instructions or teaching (there was no language to teach). Strikingly, this language shows language-specific properties that we know from other languages. How could this process be explained if humans were not genetically predisposed to acquire a language? Interestingly, the ones who developed this brand new language were all before puberty. The older people kept using home signs. Please check the relevant literature on the topic; it is a widely discussed case of spontaneous language emergence.
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity