Syntactic Ambiguity Resolution Wh ileReading in Se cond and Native LanguagesC. Frenck-Mestre and J. PynteCREPCO, University of Provence, National Centre for Scienti® c Resea rch,Aix-en-Prov ence, FranceBilinguals’ reading strategies were e xamined in their n ative and second language via therecording of eye movements. Experiment 1 examined the processing of sentences t hat con-tained local syntactic ambiguities. Results showed that bilinguals reading in their secondlanguage tended to resolve these ambiguities in a different way from native readers. Bilin-guals tended to prefer to attach incoming information to the most recently processed con-stituent. However, this global strategy was in¯ uenced by lexical information provided by theverb. Moreover, the combined analysis of both groups of readers revealed an i n¯ uence ofverb subcategorization information on syntactic ambiguity resolution. Experiment 2 alsoexamined syntactic ambiguity resolution in the native and second language, for sentencesthat were ambiguous in only one of the bilinguals’ two languages. Results showed thatbilinguals hesitated when reading in their second language at points in the sentence wheretheir native language presented con¯ icting lexical information. Following this localized effectof ``transfer’ ’, however, bilinguals performed in a manner similar to native speakers of thelanguage. In combination, these experiments demonstrate that bilinguals perform a completesyntactic parsing of sentences when reading in the second language, and they do so in amanner sim ilar to native speakers. Although lexical information can apparently in¯ uenceparsing in t he second language, our results do not provide strong evidence that it acts tooverride syntactic analysis based on structural principles.As most adults who have attempted to master a second language after early childhoodwould agree, reading in a second language is anything but an automatic process. Thedif® culty associated with foreign-language reading is r e¯ ected in the results of variousbilingual studies of sentence processing using a wide variety of measurements, fromsimpl e reaction time (Mack, 1986), to on-line as sig nment of grammatical roles (Kilborn,1989), to the recording of evoked potentials (Ardal, D onald, Meuter, Muldrew, & Luce,1990). Unanimously, these studies have shown that not only is second-language process-ing slower than nat ive -la ng uage proces sin g, but the pattern of results observed forTHE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 1997, 50A (1), 119± 148Requests for repr ints should be sent to C. Frenc k-M estre, CREPCO± U niv er site de Provence, 29 av. Robert-Schuman, 13621 Aix-en-Pro vence, Cedex, France. E-mail: [email protected] authors would like to thank Wayne Murray, Gerry Altmann, and an anonymous reviewer for constructivecomments on earl ier drafts o f this text. We also extend our than ks to Philipe Do for help in developing thestimulus materials.Ó1997 The Experimental Psychology Societybilinguals when analysing sentences in their second langu age is qualitatively differentfrom that obtained for native speakers of the langua ge.Secon d-l an guage processing apparentl y differs from native-language processing inseveral ways. For exam pl e, a study of text comprehension in the native (L1) and secon d(L2) language of advanced second-language learners (Kozminsky & Graetz, 1986)reveale d that when reading a text, these subjects underlined ke y words in the L2 text,as opposed to complete phrases in the L1 text, an d they annotated the L2 text rather thansimpl y mark passages, whereas they did the opposite in the L1. When summa ri zin g thesame text s, these subjects were less likely to paraphrase and followed paragraph ordermore for L2 than L1 texts. The re sul ts of this study led its authors to claim that text-summa ri zin g re¯ ect s ``word-f oc us orientation’’ in t he second language, as opposed toperformance in the native language, which re¯ ect s greater in¯ uence of the text’s macrostruct ur e. I t has also be en suggested that bilinguals may be less sensi tiv e to syntax in theirsecond langu ag e than in their native language (where syntactic processing is presumablymore automated) and that second-language processing is more conceptually than syn ta ct -ically guided (Ulijn, 1980).Wherea s the above studies suggest the developme nt o f proc ess in g strategies speci® c tothe second language, a considerable body of re sea rch has focused on the interactionsbetwee n the native and the second lan gu ag e. In particular, the question has been raisedas to whether adu lt, late bilinguals (those who learned their second language after pu-berty ) are in¯ uenced by strategies developed in their native language when analysingsenten ce s in their second l an gu ag e (see Durgunoglu & Hancin, 1992; Odlin, 1989, forextens ive reviews). Results from several off-line studies performed in the fr amework ofthe ``Competition model’’ (Bates & MacWhinney, 1982) indicate that this type of transferdoes occur. For example, when asked to assign th e role of subject to one of the nouns inNVN triplets presented in Eng li sh, Japanese-dominant beginning bilinguals make use ofanima cy cues, as they do in Japanese, and much more so than do native monolingualspeake rs of English (Harrington, 1987). Increased use of the second language is, ofcourse, in¯ uential, as shown by a cross-lon gi tudinal study with Eng li sh± Dutch andDutch ± English bilinguals (McDonald, 1987). The results of that study showed a clearpictur e of decreased use of ® rst-language cues in favour of second-language cues withincrea sed expos ur e to the second language (a s measured by y ear s of residency in thecount ry of the second language). Nonetheless, an in¯ uence of the native tongue was stillappare nt , even in the most highly skilled group (over 15 year s of reside nc y). In likemanne r, two ot he r studi es perf ormed in this framework revealed tha t even highly pro® -cient bilinguals do not process sentences in their second language like native speakers do;they are considerably in¯ uen ced by properties of their native language (Kilborn, 1989;Kilbo rn & Cooreman, 1987).Trans fe r of native-language syntactic knowledge has been demonstr at ed in studiesthat compared the performances of bilinguals and of native spe ake rs on grammaticalityjudgement t as ks. Results show that when making grammaticality