Arguments from the Poverty of the Stimulus

Children do not make errors they would be expected to make if they generalized from positive evidence alone (Poverty of Stimulus Arguments; Chomsky 1975), e.g.,

         i. Yes/no-question formation (Chomsky 1975:32; c.f. Ingram p. 66)

            The man who is tall is in the room.

            Is the man who is tall t in the room?

             * Is the man who t tall is in the room?

         ii. dative acquisition (Baker 1979, originally proposed in Braine 1971), p. 28

            I gave the book to Mary 

            I gave Mary the book 

            I said some things to Mary 

            * I said Mary some things

iii. wanna-contraction (Guasti, p. 9, but see Pullum 1997, O’Grady 2005; wanna is a morphological derivation, not syntactic, and therefore children can use lexical learning, not UG. Want is a ‘subject-to-object’ raising verb like consider, so the difference is between who as the object of invite and who as the object of want. Warren, Spear & Schafer (2003) report that speakers consistently lengthen want in contexts where the contraction is blocked. They are also more likely to pause after want in these contexts.)

            Who do you wanna invite?

            When do you wanna come?

            * Who do you wanna come?

         iv. Negative Polarity Items (Crain & Pietroski 2002)

            No linguist has any brains.

            Every linguist with any brains admires Chomsky.

            * Every linguist has any brains.

         v. “Inclusive or” interpretations (Crain & Pietroski 2002)

            Eat your veggies or (exclusive) you won’t get any dessert.

            No one with any brains admires linguists or (inclusive) philosophers.

            Every linguist or (inclusive) philosopher with any brains admires Chomsky.

            Everyone admires a linguist or (exclusive and inclusive) philosopher.

         vi. Pronoun interpretation; Principle B (Crain & Pietroski 2002)

            He_i said the Ninja Turtle_*i/j has the best smile. (pronoun c-commands NP)

            As he_i was leaving, the Ninja Turtle_i/j smiled. (pronoun does not c-command NP)

            Who did he say has the best smile?

                  Who_i did he_j say has the best smile? (Deictic Pronoun Reading)

                  *Who_i did he_i say has the best smile? (Bound Pronoun Reading)

            Who said he has the best smile?

                  Who_i said he_j has the best smile? (Deictic Pronoun Reading)

                  Who_i said he_i has the best smile? (Bound Pronoun Reading)

         vii. that-trace effect (Haegeman 1994)

            Who did he think was available?

            *Who did he think that was available?

            He thought that Sally was available.

Pullum & Scholz (2002:9-50) analyzed the Poverty of Stimulus (POS) arguments and distill the following points about the acquisition process and children’s language environment that are used to support POS arguments:

(1) Properties of the child’s accomplishment

   a. SPEED: Children learn language so fast.

   b. RELIABILITY: Children always succeed at language learning.

c. PRODUCTIVITY: Children acquire an ability to produce or understand any of an essentially unbounded number of sentences.

d. SELECTIVITY: Children pick their grammar from among an enormous number of seductive but incorrect alternatives.

e. UNDERDETERMINATION: Children arrive at theories (grammars) that are highly underdetermined by the data.

f. CONVERGENCE: Children end up with systems that are so similar to those of others in the same speech community.

g. UNIVERSALITY: Children acquire systems that display unexplained universal similarities that link all human languages.

(2) Properties of the child’s environment

a. INGRATITUDE: Children are not specifically or directly rewarded for their advances in language learning.

   b. FINITENESS: Children’s data-exposure histories are purely finite.

   c. IDIOSYNCRASY: Children’s data-exposure histories are highly diverse.

d. INCOMPLETENESS: Children’s data-exposure histories are incomplete (there are many sentences they never hear).

e. POSITIVITY: Children’s data-exposure histories are solely positive (they are not given negative data, i.e. details of what is ungrammatical).

f. DEGENERACY: Children’s data-exposure histories include numerous errors (slips of the tongue, false starts, etc.).

P&S (19) discern five points that must be proved for a poverty of stimulus argument to be true:

   i. ACQUIRENDUM CHARACTERIZATION: describe in detail what is alleged to be

      known.

   ii. LACUNA SPECIFICATION: identify a set of sentences such that if the learner had access

      to them, the claim of data-driven learning of the acquirendum would be supported.

   iii. INDISPENSABILITY ARGUMENT: give reason to think that if learning were

data-driven then the acquirendum could not be learned without access to sentences in the lacuna.

   iv. INACCESSIBILITY EVIDENCE: support the claim that tokens of sentences in the lacuna

      were not available to the learner during the acquisition process.

   v. ACQUISITION EVIDENCE: give reason to believe that the the acquirendum does in fact

      become known to learners during childhood.

Pullum & Scholz claim that no one has produced a poverty of stimulus argument that successfully addresses all five of these points. They begin with a logical argument against the POS argument made by Sampson (1989, 1999a) that goes:

Consider the position of a linguist - let us call her Angela - who claims that some grammatical fact F about a language L has been learned by some speaker S who was provided with no evidence for the truth of F. Sampson raises this question: How does Angela know that F is a fact about L? If the answer to this involves giving evidence from expressions of L, then Angela has conceded that such evidence is available, which means that S could in principle have learned F from that evidence. That contradicts what was to be shown. If, on the other hand, Angela knows L natively, and claims to know F in virtue of having come to know it during her own first-language acquisition period with the aid of innate linguistically-specific information, then Angela has presupposed nativism in an argument for nativism (15).

P&S claim that Haegeman’s argument from the that-trace effect is based on a false claim about the nature of the ACQUIRENDUM. Rather than learning to rule out ‘V + that + finite VP’ sequences such as *Who did they think that was available? children instead learn ‘N + that + finite VP’ sequences such as things that were available. Children never hear ‘V + that + finite VP’ constructions so they do not produce them. P&S (16) suggest children may not be learning the type of movement rule that is the basis for Haegeman’s argument.

P&S suggest that an argument made by Lightfoot (1998:585) has a similar problem. Lightfoot claims that children might be misled into overextending contraction to sentences like *Kim’s taller than Jim’s. P&S (17) observe that if children learn that contraction only applies in stressless contexts they will not apply contraction to cases where the VP head has weak stress.

Gordon (1986) and Pinker (1994) also misanalyzed the ACQUIENDUM in claiming that noun compounds can only have irregular plural modifiers (mice-catcher, teeth eater), but P&S cite many counter-examples (forms-reader, securities-dealer, drinks trolley, rules committee, complaints department).

   P&S (27) cite an argument by Kimball that ‘English-speaking children will acquire the full auxiliary system . . . without having heard sentences directly illustrating each of the rules’ (1973: 74). Kimball claims that ‘sentences in which the auxiliary is fully represented by a modal, perfect, and progressive are vanishingly rare’ illustrating the INACCESSIBILITY EVIDENCE part of the argument. P&S (28-29) cite many examples in which the full auxiliary is present to suggest that Kimball’s claim lacks credibility.

Baker (1978:413-425) claimed that one serves as an anaphor for a constituent smaller than a noun phrase:

         This box is bigger than the other one.

Baker also claims evidence for this possibility is unavailable to children. P&S question the accuracy of Baker’s linguistic claim (ACQUIRENDUM) as well as the claim that children lack evidence for one’s usage (INACCESSIBILITY) citing (36) the following dialogue:

         A: "Do you think you will ever remarry again? I don't."

B: "Maybe I will, someday. But he'd have to be somebody very special. Sensitive and supportive, giving. Hey, wait a minute, where do they make guys like this?"

         A: "I don't know. I've never seen one up close."

P&S (39-40) claim that Chomsky’s yes/no-question example rests on a claim about the absence of relevant examples in the input (INACCESSIBILITY). They cite many examples to the contrary (40-44). Chomsky’s argument also ignores the semantics of yes/no questions and the discourse function of relative clauses (Hausser 2004:921). Relative clauses provide background information while questions focus on elements in the foreground:

      * What did Robin give a pencil to the woman who found ___ ?

      * Are the people who ___ on the bus look tired?

      * How did the car that drove ___ was parked?

Core versus Periphery

One problem P&S do not address is the distinction between core and periphery. UG only specifies the core properties of human languages; language-specific features constitute the marked periphery. POS arguments assume the innate properties are part of core grammar, but do not show that they hold in other languages.

Conclusion

We have reviewed the evidence for two key arguments for Universal Grammar: 1. parameters and 2. the argument from the poverty of the stimulus. Current evidence does not support either of these arguments. We conclude that there is no evidence to support UG.

Bibliography on the Poverty of Stimulus Argument

Baker, C. L. (1979). Syntactic theory and the projection problem. Linguistic Inquiry 10: 533-581.

Braine, Martin D. S. (1963). The ontogeny of English phrase structure: the first phase. Language 39:1-13.

   Chomsky, Noam (1975). Reflections on Language. New York, New York: Pantheon

Crain, Stephen and Paul Pietroski. (2002) Why Language Acquisition is a Snap. The Linguistic Review 2002.

Haegeman, Liliane (1994). Introduction to Government and Binding Theory. 2nd edition. Oxford, U.K. and Cambridge, MA: Basil Blackwell.

Lasnik, Howard and Juan Uriagereka (2002). On the poverty of the challenge. The Linguistic Review 19: 147–150.

Levine, Robert D. (2002). Review of Juan Uriagareka, Rhyme and Reason. Language 78:325-330.

Legate, Julie Anne and Yang, Charles D. (2002). Empirical re-assessments of stimulus poverty arguments. The Linguistic Review 19:151-162.

MacWhinney, Brian. (2004). A multiple process solution to the logical problem of language

       acquisition. JCL 31:883-914.

O’Grady, William (2005). Syntactic Carpentry: An Emergentist Approach to Syntax. Mahwah: NJ: Lawrence Erlbaum.

   Pinker, Steven (1994). The Language Instinct. New York: William Morrow.

Pullum, Geoffrey (1996). Learnability, hyperlearning, and the poverty of the stimulus. Paper pre-sented at the Parasession on Learnability, 22nd Annual Meeting of the Berkeley Linguistics Society, Berkeley, CA.

   Pullum, Geoffrey (1997). The morpholexical nature of to-contraction. Language 73: 79-102.

Pullum, Geoffrey and Barbara Scholz (2002). Empirical assessment of stimulus poverty arguments. The Linguistic Review 19: 8–50

Sampson, Geoffrey (1989). Language acquisition: growth or learning? Philosophical Papers 18:203–240.

   -----. (1999a). Educating Eve. London: Cassell.

-----. (2002). Exploring the richness of the stimulus. The Linguistic Review 19: 73–104.

Warren, Paul, Shari Speer & Amy Schafer. 2003. Wanna-contraction and prosodic disambiguation in US and NZ English. Wellington Working Papers in Linguistics 15:31-49.

4. Assessment: we need to combine the methodological rigor of Child Language research with the theoretical orientation of Language Acquisition, i.e. CL needs more theory, LA needs more data.

What is required is a real theory of child language acquisition, one that predicts language development

   Current theories describe rather than explain language acquisition, c.f. Brainard 1978

      They are primarily limited to English or a few European languages

      They lack independent evidence to support them

      They do not make interesting predictions