Neural Correlates for the Acquisition of Natural Language Syntax Marco Tettamanti,* , † ,1 Hatem Alkadhi,† Andrea Moro,‡ Daniela Perani,‡ , § Spyros Kollias,† and Dorothea Weniger† *Scientific Institute San Raffaele, Milan, Italy; †University of Zurich, Zurich, Switzerland; ‡Vita-Salute University San Raffaele, Milan, Italy; and §Institute of Neuroscience and Bioimaging CNR, Milan, Italy Received September 19, 2001 Some types of simple and logically possible syntactic rule never occur in human language grammars, leading to a distinction between grammatical and nongrammati- cal syntactic rules. Comparison of the neuroanatomical correlates underlying the acquisition of grammatical and nongrammatical rules can provide relevant evi- dence on the neural processes dedicated to language acquisition in a given developmental stage. Until present no direct evidence on the neural mechanisms subserving language acquisition at any developmental stage has been supplied. We used fMRI in investigating the acquisition of grammatical and nongrammatical rules in the specified sense in 14 healthy adults. Gram- matical rules compared with nongrammatical rules spe- cifically activated a left hemispheric network including Broca’s area, as shown by direct comparisons between the two rule types. The selective role of Broca’s area was further confirmed by time  condition interactions and by proficiency effects, in that higher proficiency in grammatical rule usage, but not in usage of non- grammatical rules, led to higher levels of activation in this area. These findings provide evidence for the neu- ral mechanisms underlying language acquisition in adults. © 2002 Elsevier Science (USA) INTRODUCTION The two major determinants of language acquisition are the characteristics of the linguistic input, such as the structural properties of phrases and the distribu- tion of each linguistic item, and the learner’s encounter with such an input (Kuhl, 2000; Saffran et al., 1996). The latter involves primarily the degree of exposure to a language and the age of acquisition. Information on the developmental time course of language acquisition in infancy has been gained from normal and neurolog- ically or developmentally impaired children (Fletcher and MacWhinney, 1995). Clinical studies, together with neuroimaging investigations on bi- and multilin- guals, have contributed to the identification of the neu- roanatomical correlates of language production and comprehension in languages acquired before and after puberty (Abutalebi et al., 2001). However, no direct evidence concerning the neural mechanisms subserv- ing language acquisition at any developmental stage has been forwarded. An unresolved issue pertains to how the brain regions involved in the acquisition of a second language in adulthood are related to the regions involved in processing the mother language. In adults, specific patterns of neural activity associated with the different language components as defined by linguistic theories have been found. In particular, there is now converging evidence as to the brain areas subserving syntactic processing in the mother language. Many non- fluent aphasic patients with a lesion in left inferior fron- tal cortical areas (BA 44 and 45) display the clinical syndrome of agrammatism. Agrammatism is character- ized by the omission of bound and free-standing gram- matical morphemes (Kean, 1985). Furthermore, some agrammatic patients are unable to understand the mean- ing of a sentence, when functional arguments such as subject and direct object have to be identified on the basis of syntactic structure (Grodzinsky, 2000). In agreement with such clinical findings, neuroimaging studies have revealed activations within and around Broca’s area, when healthy adult subjects were required to perform tasks calling for syntactic processing. Some studies var- ied the level of syntactic complexity (Caplan et al., 1998; Stromswold et al., 1996), resulting in differential de- mands on verbal working memory resources. Other stud- ies used materials consisting either of syntactically anomalous sentences (Embick et al., 2000; Meyer et al., 2000; Ni et al., 2000) or of sentence pairs with different syntactic structure, which could be same or different in meaning (Dapretto and Bookheimer, 1999). A drawback of the latter studies is that processing of the pertinent syntactic features is tied up with sentential semantic processing. To eliminate such confounding semantic effects, Moro and colleagues (2001) also used a syntactic vio- 1 To whom correspondence should be addressed at Neuroscience and Bioimaging Institute, LITA, Via Fratelli Cervi 93, I-20090 Segrate (MI), Italy. Fax: +39-02-21717558. E-mail: tettamanti.marco@hsr.it. NeuroImage 17, 700 –709 (2002) doi:10.1006/nimg.2002.1201 700 1053-8119/02 $35.00 © 2002 Elsevier Science (USA) All rights reserved.