Skinner's (1957) behaviourist account of language in *Verbal Behavior* proposed that children acquire language through:

A: An innate language acquisition device that generates grammatical rules from minimal input

B: Operant conditioning — reinforcement of correct verbal behaviour and shaping of successive approximations

C: Statistical detection of transitional probabilities between syllables in the speech stream

D: Social imitation of communicative intentions within caregiver interaction

Correct: Operant conditioning — reinforcement of correct verbal behaviour and shaping of successive approximations

Skinner proposed that verbal behaviour is no different from other operant behaviour: caregivers reinforce correct utterances and correct errors, shaping language toward adult norms. This includes imitation (echoic behaviour), response to commands (mand), and description (tact). Chomsky's 1959 review demolished this account on multiple grounds: language is too creative, errors are too systematic, and input too impoverished to support a purely reinforcement-based account.

Chomsky's (1959) critique of Skinner identified three fundamental problems for a behaviourist account of language. Which is NOT one of them?

A: Linguistic creativity — speakers produce and understand sentences they have never heard

B: The poverty of the stimulus — grammatical knowledge exceeds what could be learned from the input

C: Imitation failure — children cannot repeat sentences verbatim when asked

D: The universality of grammar — all languages share deep structural properties inconsistent with environment-specific learning

Correct: Imitation failure — children cannot repeat sentences verbatim when asked

Chomsky's main arguments were: (1) productivity/creativity — reinforcement cannot account for the unbounded generation of novel sentences; (2) poverty of the stimulus — children acquire grammatical principles not reliably present in the input; (3) universality — cross-linguistic grammatical properties suggest a biologically determined linguistic faculty. Imitation ability is unrelated to Chomsky's critique and children can often repeat sentences.

The "poverty of the stimulus" argument holds that:

A: Children in impoverished environments fail to develop full language competence

B: Children hear too few examples of complex sentences to learn grammar by statistical induction alone

C: Children acquire grammatical knowledge — such as structure-dependence — that outstrips what they could learn from the input, requiring innate grammatical constraints

D: Caregivers provide impoverished language models compared to peer interaction

Correct: Children acquire grammatical knowledge — such as structure-dependence — that outstrips what they could learn from the input, requiring innate grammatical constraints

Chomsky's poverty of the stimulus argument uses structure-dependence as the key example: children never form yes/no questions by moving the first auxiliary (producing "Is the man who is tall is happy?"), despite never being corrected for this error and rarely hearing relevant examples. This suggests knowledge of syntactic structure is innate. The POS remains controversial: Pullum & Scholz (2002) dispute whether the stimulus is really as impoverished as Chomsky claims.

True or False: Chomsky's Language Acquisition Device (LAD) is a specific, identifiable neural structure whose location has been confirmed by neuroimaging.

Answer: False

FALSE. The LAD is a theoretical construct, not a localised brain structure. Chomsky proposed it to explain the child's ability to acquire any human language from impoverished input — it represents the innate linguistic knowledge (Universal Grammar) that constrains possible grammars. No single brain region has been identified as the LAD, though Broca's area and associated perisylvian cortex are often discussed as candidates for innate language circuitry.

Saffran, Aslin & Newport (1996) exposed 8-month-old infants to a continuous, unparsed stream of syllables for just 2 minutes. After exposure, infants could:

A: Reproduce individual words from memory when prompted

B: Distinguish words (high within-word transitional probability) from non-words (low probability), showing statistical learning

C: Identify the grammar underlying the sequence and apply it to novel sentences

D: Only recognise the first word in the stream due to primacy effects

Correct: Distinguish words (high within-word transitional probability) from non-words (low probability), showing statistical learning

Saffran et al. found that infants tracked the transitional probabilities between syllables — if "bi" was always followed by "da" (high TP = 1.0) but "da" was followed by "ku" only one-third of the time (TP = 0.33), infants treated "bida" as word-internal and "daku" as a word boundary. This demonstrated powerful statistical learning with no explicit instruction — foundational evidence that language acquisition can proceed via domain-general mechanisms.

Tomasello's (2003) usage-based theory of language acquisition emphasises that children acquire language through:

A: Triggering of innate parameters in Universal Grammar by exposure to any human language

B: Reinforcement of correct speech by caregivers according to operant conditioning principles

C: Joint attention, intention reading, and pattern-abstraction from item-specific constructions learned through social use

D: Exposure to statistical regularities in the speech stream without social scaffolding

Correct: Joint attention, intention reading, and pattern-abstraction from item-specific constructions learned through social use

Tomasello argues that two species-specific capacities enable human language acquisition: joint attention (sharing a focus with another person) and intention reading (understanding communicative goals). Children learn language item-by-item in social contexts and abstract more general constructions through analogy and schema formation. This usage-based approach does not require an innate grammar — general cognitive mechanisms and social cognition are sufficient.

Greenberg (1966) identified linguistic universals — properties shared by all or nearly all human languages. Chomsky interpreted such universals as evidence for:

A: Cultural diffusion — all languages descended from a common ancestral proto-language

B: Universal Grammar — innate biological constraints on the structure of possible human languages

C: Cognitive universals — shared mental architecture that produces similar languages without innate linguistic knowledge

D: Environmental universals — all speech communities face the same communicative pressures

Correct: Universal Grammar — innate biological constraints on the structure of possible human languages

Chomsky argued that the existence of substantive universals (e.g., all languages have nouns, verbs, and recursive structure; all use structure-dependent transformations) cannot be explained by shared environments or learning history. Instead, they reflect a biologically determined Universal Grammar that constrains what languages can look like. This is the core of the nativist position, though some universals are now attributed to domain-general cognitive pressures (Evans & Levinson, 2009).

Rumelhart & McClelland's (1986) connectionist model of past-tense acquisition generated controversy because it:

A: Required a large hand-coded lexicon and explicit grammatical rules

B: Could not learn irregular forms at all, demonstrating the need for innate grammar

C: Replicated U-shaped learning (initially correct → overregularisation → correct) without explicit rules, challenging the need for a rule-based component

D: Used reinforcement learning identical to Skinner's model, confirming behaviourist predictions

Correct: Replicated U-shaped learning (initially correct → overregularisation → correct) without explicit rules, challenging the need for a rule-based component

Rumelhart & McClelland's PDP model learned English past tenses from input patterns without explicit rules and reproduced U-shaped development — good performance on high-frequency irregulars early, then overregularisation as regulars increase, then recovery — purely through statistical associations. Pinker & Prince (1988) contested this, arguing the model required hand-crafted inputs and didn't generalise properly. The debate crystallised the nativist vs. connectionist dispute about domain-specific innate knowledge.