102 Book Reviews LANGUAGE AND THE LEARNING cilitates future performance. Thus, CURVE: A NEW THEORY OF SYN- when the cumulative number of verbs TACTIC DEVELOPMENT. Anat Ninio. in a particular construction is plotted Oxford University Press, Oxford, 2006. over time, a power-law speed-up pp. 206. Price: £26.00, US $59.50. ISBN function is seen (Chapter 2). Children 9780199299829. do not generalize a verb’s valency frame to semantically similar verbs How do children acquire the syntax of (Chapters 3 and 4): The age at which a their native language? One popular child will use a particular verb in a view is that children are born with particular construction is predicted by some knowledge of syntax and that the overall number of verbs—not by the acquisition consists largely of linking number of semantically similar verbs— these abstract rules to the particular that she has previously used in this language that the child is learning (e.g. construction. Pinker, 1989). The opposite view is that One problem for Ninio’s account children form abstract syntactic con- concerns the origin of the ‘Merge’ op- structions by abstracting across utter- eration. The central claim of this ac- ances in the input that instantiate them count is that ‘children do not form (e.g. Tomasello, 2003). Ninio rejects abstract rules’ (p. 4), but what is a ‘very both these accounts, instead arguing general principle’ (p .31) that can that ‘children learn a lexicalist syntax, combine items as diverse as the1man in which the syntactic structure and saw1it if not an abstract rule? of the sentence is projected from the Furthermore, Ninio assumes that lexicon’ (p. 6) and ‘do not form abstract ‘children deduce the principles of rules or schemas’ (p. 4). In this review Merge/Dependency from sentences I will argue—on the one hand—that adults say to them’. It is difficult to see this account fails to address certain how learning the ‘principle’ that gen- findings that are more easily explained erates (for example) V1O combina- by a construction-based account and— tions such as saw it is different from an on the other—that the two accounts account under which children abstract may be more similar than Ninio from their input a VO schema (e.g. suggests. Tomasello, 2003), a claim that Ninio Ninio’s account is as follows: The explicitly rejects. Furthermore, the fact adult grammar (Chapters 1 and 3) that children say the man as opposed to consists of the syntactic operation man the is not—as Ninio claims—evi- ‘Merge’ and a lexicon in which each dence for Merge; children could have entry (e.g. see) is annotated for its se- simply acquired the phrase as a whole mantic relations (e.g. seer, thing seen) from the input. and its syntactic valency (e.g. _see_). The main problem with Ninio’s ac- Speakers construct sentences by re- count, however, lies with her rejection cursively applying the Merge opera- of construction semantics. Construc- tion to pairs of words in accordance tion grammars posit the independent with these valency requirements (e.g. existence of constructions because they saw and it merge to form one con- seem to be associated with meanings stituent, the and man to form another, above and beyond the words that in- and these resultant constituents merge stantiate them. For example, Goldberg to form the sentence). (1995) argues that part of the meaning Children’s two-word combinations of He sneezed the napkin off the table (e.g. saw it) are Merge couples of this comes from the caused-motion mean- form. Producing a sentence using a ing of the construction. Ninio’s lex- valency-based frame (e.g. VO) is like icalist account would have to posit a any other cognitive skill: Practice fa- lexical entry for a caused-motion sense Copyright r 2009 John Wiley & Sons, Ltd. Inf. Child. Dev. 18: 95–104 (2009) Book Reviews 103 of sneeze with three arguments, which However, one way of conceptualizing is both implausible and circular. the ‘constructions’ posited by authors Ninio herself recognizes that chil- such as Tomasello (2003), Goldberg dren’s semantically based over- (1995), or Bybee and Hopper (2001) is as generalizations (e.g. They [grapes] just emergent generalizations over stored cough me—presumably on semantic instances of the pattern. It is not clear analogy with causative sentences such that there is any conceptual difference as The man broke the stick; Bowerman, between saying that speakers produce 1988) are potentially problematic for SVO utterances with novel verbs (1) on her theory. However, such errors are analogy with verbs that have been at- dismissed as ‘a late phenomenon’, the tested in the SVO pattern (Ninio) or (2) product of a semantically based gen- using an SVO construction that has eralization process that ‘does not serve been formed by analogizing across dif- children in the acquisition of the basic ferent instances of this pattern in the syntax of their language’ (p. 103). Al- input (‘abstraction’ accounts). though such errors may be rare in In the end, although many researchers production until later in development, will not agree with Ninio’s conclusions, evidence from preferential-looking this account makes an important con- studies suggests that children are tribution to the literature by emphasizing aware of the semantics associated with the need for constructivist theories to be particular argument structure con- more explicit about the representations structions from as young as age 2 (e.g. that they assume (are ‘constructions’ Kidd, Bavin, & Rhodes, 2001). simply a shorthand for analogies across Despite these differences, Ninio’s ac- stored exemplars?) and how these re- count and accounts that assume that presentations develop (what does it children are learning constructions in mean to say that early constructions are the form of semantics–syntax pairings partially lexically specific and partially (e.g. Tomasello, 2003) may be, in many abstract?). In conclusion, Ninio’s account respects, less different than Ninio as- forces the reader to examine his or her sumes. In Chapter 2, Ninio shows that own (perhaps implicit) views on the familiarity with a construction (e.g. nature of syntax and its acquisition, and SVO) facilitates subsequent performance is therefore required reading for all re- with this construction (hence an accel- searchers working in this area. erating learning curve). Ninio takes this as evidence against Tomasello’s (1992) verb-island hypothesis, under which REFERENCES children’s earliest utterances are pro- duced using independent slot-and- Bybee, J., & Hopper, P. (2001). Frequency and frame schemas (e.g. Cut X, Draw X)until the emergence of linguistic structure. Am- around age 2–6. However these data are sterdam: John Benjamins. compatible with more recent formula- Goldberg, A. E. (1995). Constructions: A tions of this hypothesis, under which construction grammar approach to argument structure. Chicago: University of Chicago children begin to link islands early in Press. development to form weak schemas that Kidd, E., Bavin, E. L., & Rhodes, B. (2001). strengthen as development proceeds Two-year olds’ knowledge of (e.g. Tomasello & Abbot Smith, 2002). verbs and argument structure. In M. In several places (e.g. pp. 35, 86), Almgren (Ed.), Research on child language Ninio contrasts the view that sentences acquisition. Somerville, MA: Cascadilla Press. with novel items (especially verbs) are Naigles, L., Fowler, A., & Helm, A. (1992). produced using abstract constructions Developmental shifts in the construction with her own, under which they are of verb meanings. Cognitive Development, produced on analogy with stored items. 7, 403–427. Copyright r 2009 John Wiley & Sons, Ltd. Inf. Child. Dev. 18: 95–104 (2009) 104 Book Reviews Pinker, S. (1989). Learnability and cognition: Ben Ambridge The acquisition of argument structure. Cam- Department of Psychology, University of bridge, MA: MIT Press. Liverpool, Liverpool, UK Tomasello, M. (1992). First verbs: A case study of early grammatical development. New York: Cambridge University Press. DOI: 10.1002/icd.589 Tomasello, M. (2003). Constructing a lan- guage: A usage-based theory of language acquisition. Cambridge, MA: Harvard University Press. Copyright r 2009 John Wiley & Sons, Ltd. Inf. Child. Dev. 18: 95–104 (2009).