A Conversation with Steven Pinker

Oct 29 / Rebecca Hope

On 10 October 2022, Steven Pinker visited Amsterdam to speak about Human Nature and the Future of War. For any linguist that adheres to the thinking of Western logic, meeting such a prolific linguist would be a dream come true. 

What I appreciate about Mr Pinker is his choice of topics and ability to condense masses of data into digestible narratives which attempts to steer as far away from bias as possible. He gently takes his audience on a rollercoaster of data while voicing his opinions and "takes" on the data (quite bold for someone emerging out of deep academia - at least by European standards). 

I can go on about Mr Pinker's positive attributes and contributions to society. But there was one particular subject which we disagreed on. Language acquisition. How the brain acquires a language. 

The mid-1900s was a great time of the height of mathematics, computation and new specialised fields emerging out of academia. Heck, I wish I were born in the 1950s. Maybe then my linguistic thesis would actually have some value. 

Noam Chomsky is one of the kings of linguistics. He has spoken and written a lot about second language acquisition. I like his work and respect his thought leadership in the field. But I have one problem with Mr Chomsky. He only speaks 3 languages. 

Where the compartmentalisation of languages into separate entities is largely rejected by the Global South and replaced by the idea that languages overlap and are impossible to count (translanguaging), my voice does not stand in isolation. 

Empty space, drag to resize

Accepting the Mystery of Universal Grammar

by Rebecca Hope

Empty space, drag to resize

The burning inquisition for understanding how exactly humans acquire a language has been a hot topic from the mid-1900s until today where a consensus on the validity of Noam Chomsky’s foundational language acquisition network of theories of Universal Grammar, the Poverty of Stimulus and the Language Acquisition Device has still not been attained. Following the development of these concepts of the mid- to late-1900s which support the idea that humans are born with the inherent framework to acquire a language, empirical linguistic evidence in the late 1900s and 2000s began to counter the abstract nativist theories of the innate language capacity. Here, two strong yet simple opposing bodies of recent empirical evidence demonstrated in case studies of infants and children using visual and sound prompts to elicit responses are explained and used to conclude the validity of the language acquisition theories of Chomsky in relation to the Poverty of Stimulus and Universal Grammar in particular. 

Although Chomsky did not discredit the importance of language input “to inform learners about the properties of a language” (Thomas, 2002, pp. 58), it was suggested that “poverty but not degeneracy is the essential obstacle for an input-driven theory of language acquisition” (Thomas, 2002, pp. 58) which became Chomsky’s Poverty of Stimulus theory and tied in with his theory of Universal Grammar suggesting that an innate ability to acquire a language is possessed by humans. Anomalies such as fragments and slips of the tongue, the finite nature of language input and the lack of an explanation for the relationships of synonymy, ambiguity, and ungrammaticality (Hornstein & Lightfoot 1981: 9-10) were all supporting elements which suggested that input was not enough to explain how language is acquired and which contributed to the strengthening of the concept of Poverty of Stimulus in the 1980s. In more recent times, “a range of facts about the interpretation of anaphors and pronouns, one-pronominalization, and that-trace effects”  (Thomas, 2002, pp. 58) have kept the support of the justification of Universal Grammar using the concept of Poverty of Stimulus alive. 

In the article Acquiring wanna: Beyond Universal Grammar written in 2016, Heidi R. Getz challenges the use of the principle of Universal Grammar to explain “children’s reported knowledge of the ‘wanna fact’” (Crain & Thornton, 1998) where the traditional Universal Grammar analysis of “wanna” suggests that children would not be able to pick up on subtleties that they can not hear and therefore the knowledge must be innate. Getz argues that an innate Universal Grammar can not be used to explain the use of “wanna” in children as there is little correlation between adults and children for the usage of “wanna”. If adults and children do not use “wanna” in the same way, it would imply that a Universal Grammar does not exist. 

Two experiments were conducted to explore what we do and do not know about the “wanna” facts. The participants were two groups of native American English-speaking children: twelve in the younger group aged three to four and fourteen in an older group aged five to seven. 

In the first experiment, children were shown images and the narrator prompted the children to ask the characters in the picture questions. The questions would use prompts with a WH-question like “What do you want to build?” (Getz, 2019, pp.123) or a “WHO” question like “Who do you want to talk to the alien?” (Getz, 2019, pp. 123). The aim was to elicit a response using the word “wanna” where the use of the word could either be grammatically incorrect (indicated as *wanna) or grammatically correct (indicated as ✓wanna). The average use of *wanna was then calculated.  Children use *wanna nearly half the time (forty seven percent) which is far more than adults where only two percent of the use of “wanna” is incorrect. Children do not use “wanna” like parents do: “wanna” is used much more by children than by parents and “wanna” is frequently used ungrammatically by children when with embedded verbs.

The same groups as before were used in the second experiment (except for four children who did not produce utterances and were therefore excluded from the experiment). More high-frequency verbs like “do, have, go, see” were embedded in the questions used to elicit responses from the children. The ungrammatical use of the word “wanna” reduced significantly but was still much higher than that of adults. With the added highest-frequency verbs used in eliciting questions from children, there is still little correlation between adults and children for the usage of “wanna”. “The wanna facts are a famous Poverty of Stimulus puzzle” (Getz, 2019, pp. 137).

The traditional Universal Grammar analysis of “wanna” suggests that children would not be able to pick up on subtleties that they could not hear and therefore the knowledge must be innate. However, here we see that this study provides evidence for no innate ability stepping in to help children acquire the correct use of “wanna”. Children neither pick up on the subtleties of the correct use of “wanna” based on input-learning nor does an innate ability guide them to its correct usage. 

Getz’ research poses a somewhat catch-22 situation. On the one hand, this experiment supports the idea of Poverty of Stimulus because clearly there is some sort of inaudible element which children are picking up in order to learn the word “wanna” which they receive little language input of. On the other hand, this experiment shows that children’s use of “wanna” is not adultlike which implies that there is no Universal Grammar aiding the children to acquire the word in the way that adults use it. If not the input, it would be expected that Chomsky’s proposed Language Acquisition Device would help the children to use “wanna” like adults do. However, this is not the case. Therefore, the Poverty of Stimulus concept can not in itself support the nativist theory of Universal Grammar which provides a theory for language acquisition. This theory is not completely invalid but cannot be used solely to explain language acquisition. 

A more inclusive alternative is proposed by Getz which does not completely discredit Universal Grammar but also makes it very clear that there are other mechanisms at play in language acquisition. The language learning process or language acquisition requires various mechanisms such as input, innate knowledge and other elements including “cue-based acquisition” (Lightfoot 1991, 2017; Fodor 1998). 

The “wanna” case study is merely one of many types of empirical evidence studies presented to disprove the idea of Poverty of Stimulus and to demand that attention be given to the implicit messages through processes like cue prompting which children have the cognitive ability to grasp from language exposure. This leads one to question how such an idea postulated by Chomsky gained so much credibility as “generativism’s signature warrant”  (Thomas, 2002, pp. 55) to start with. Perhaps the language input itself carries the internal structure or Universal Grammar that Chomsky is so fervently searching for.  

Contrasting Getz’ conclusions, in a 2003 article What infants know about syntax but couldn't have learned: experimental evidence for syntactic structure at 18 months, Jeffrey Lidz collected empirical evidence in support of the Poverty of Stimulus which targeted a small group of infants with prompts. The article was written after the rise of empirical evidence which countered the Poverty of Stimulus theory.  It explores whether the group of infants know that one is anaphoric in nature and therefore that the noun phrase has a hierarchical structure and not a flat structure (Lidz, Waxman, & Freedman, 2003, B69). 

A group of twenty four English-speaking infants of which twelve were male and twelve were female between the ages of sixteen months and twenty three days and eighteen months and fifteen days were studied for their interpretation of sentences with one-substitution. Children of this age start uttering more than one word which is the first sign of the verbal use of language after a short time of exposure to language.

The infants listened to a video recording in an infant-friendly voice where visual colourful drawings of four objects were used: a bottle, car, shoe, and bear  (Lidz et al., 2003, B69). This experiment was conducted over four trials with a familiarization phase where the objects were named and placed in a simple sentence with a determiner, adjective and noun for example “Look! A yellow bottle”  (Lidz et al., 2003, B70). In the test phase, two objects were placed next to each other. These images were shown on a screen to the infants who were also being recorded by a camcorder attached to the screen. Two different conditions - a control condition and an anaphoric condition - in the language stimulus were presented to the children randomly. A control condition prompt would be a neutral phrase such as “Now look. What do you see?” and an anaphoric condition prompt would contain the anaphoric expression one in a question like “Now look. Do you see another one?” (Lidz et al., 2003, B70). The whole experiment lasted three minutes and forty six seconds while the test phase only lasted eight seconds.

The times that the children spent looking at the video recording were entered into an analysis of variance (Lidz et al., 2003, B71).  The analysis was categorized into two factors: direction of look (familiar vs. novel) and condition (anaphoric vs. control). “Subjects in the anaphoric condition were more likely to look at the familiar image (i.e. the one labeled by the N0 ) than were subjects in the control condition” (Lidz et al., 2003, B71). 

Lidz concludes that these results (of which pictures infants look at) prove that by eighteen months, infants can interpret one as anaphoric in the correct instances and thus that children understand the noun phrase (NP) nested structure. It seems a bit of a far-fetched conclusion to draw. Lids’ findings were used to demonstrate that learners who are just starting to “combine words productively already have a very rich representational system for assigning syntactic structure to the exposure language” (Lidz et al., 2003, B72). This evidence is somewhat unique in that it is a more recent contribution in support of the Poverty of Stimulus concept and which contains empirical evidence of children at the very start of their language acquisition journey. 

Upon reflecting on this study, I am not convinced that the evidence of the infants proves that a Universal Grammar exists. The experiment seems one-dimensional, failing to paint a full picture of the various elements that could have led to the children looking at the particular images reflecting the correct use of the anaphoric one. Such elements could include the tonality, stress, pitch and rhythm of the speaker’s voice. The experiment seems corrupted in that it omits this information about the phonetic aspects of the infant-friendly voice where the sound in the speaker’s voice could play a large role in what type of response is elicited. Also, although the act of infants looking at a particular image within an limited eight-second time frame could be interpreted as a form of normal verbal communication alluding to the understanding of the anaphoric one, it seems too far-fetched and with no substantial evidence to justify the connection between the two phenomena. This gives the impression that the experiment was conducted to shape the elicited information from the infants in order to prove the existence of a Universal Grammar. Although I do not discredit the possibility of an innate ability to acquire a language, it still seems to be a very mysterious abstract concept with problematic supporting empirical evidence. Within the broader domain, I would argue that modern linguistics should completely do away with the flawed logic which uses the fact that humans are able to produce and understand an infinite amount of utterances with only having “‘finite and accidental’ exposure to language” (Chomsky, 1957, pp. 15) to prove that Universal Grammar exists. An innate capacity could still exist but should use evidence other than the Poverty of Stimulus to validate its existence. 

Bold hypotheses were put forward about language acquisition by monolingual linguists (such as Chomsky) with Eurocentric views in a time when a computational approach was ripe in academia (Nofre, Priestley & Alberts, 2014) and the idea of a Universal Grammar fit right in with the academic culture of that era. Such a narrow linguistic approach lacks careful consideration of grammar structures that can be learnt through various hidden implications (such as cue prompting as in the case of acquiring “wanna”) by a child’s mind. That is not to say that the human mind does not have an innate capacity to learn a language. A painting cannot exist without both the paint and the canvas. In the same way, it could be suggested that language acquisition cannot exist without both Universal Grammar and language input. Given Getz’ conclusions which proves a weak link between the Poverty of Stimulus and Universal Grammar drawn from the “wanna” case study and Lidz’ one-dimensional somewhat far-fetched conclusions extracted from the anaphoric one research with infants, it appears that the concept of Universal Grammar is yet to be validated by concrete empirical evidence and remains somewhat of a mystery.    

Empty space, drag to resize
Empty space, drag to resize
Empty space, drag to resize


Chomsky, Noam. 1957. Syntactic Structures. The Hague: Mouton.
Crain, S. & Rosalind Thornton. 1998. Investigations in Universal Grammar: A guide to experiments on the acquisition of syntax and semantics. Cambridge, MA: The MIT Press.
Fodor, Janet Dean. 1998. Unambiguous triggers. Linguistic Inquiry 29(1). 1–36. 
Getz, Heidi R. 2019. Acquiring wanna: Beyond Universal Grammar. Language Acquisition 26:2, 119-143, DOI: 10.1080/10489223.2018.1470242.
Hornstein, Norbert and David Lightfoot (eds.). 1981. Explanation in Linguistics: The Logical Problem of Language Acquisition. London: Longman.
Lidz, Jeffrey, Sandra Waxman and Jennifer Freedman. 2003. What infants know about syntax but couldn’t have learned: experimental evidence for syntactic structure at 18 months. Cognition 89 (2003) B65–B73.
Lightfoot, David W. 1991. How to set parameters: Arguments from language change. Cambridge, MA: MIT Press.
Lightfoot, David W. 2017. Discovering new variable properties without parameters. Linguistic Analysis 41(3–4). 409–444.
Nofre, David, Mark Priestley and Gerard Alberts. 2014. When Technology Became Language: The Origins of the Linguistic Conception of Computer Programming, 1950–1960. Technology and Culture Vol. 55, No. 1 (January 2014), pp. 40-75. The Johns Hopkins University Press and the Society for the History of Technology. 
Thomas, Margaret. 2002. Development of the concept of “the poverty of the stimulus”. The Linguistic Review 19(1-2):51-71.

Created with