Acquiring native-like intonation in Dutch and Spanish: Comparing the L1 and L2 of native speakers and second language learners

ACQUIRING NATIVE-LIKE INTONATION IN DUTCH AND SPANISH Comparing the L1 and L2 of native speakers and second language learners Introduction Learning more about the interaction between the native language (L1) and the target language (L2) has been the aim of many studies on second language acquisition (SLA). Where the first studies mostly focussed on the influence of the L1 on the L2, later studies showed that the opposite is also possible, making linguistic transfer a bi-directional phenomenon (Pavlenko & Jarvis 2002). However, many of these studies concentrated on the acquisition of native-like phonology and grammar, while mostly ignoring intonation (and other suprasegmental features) as a possible distinguishing factor between L2 learners and native speakers. The studies that did look into prosodic features either did not perform an objective acoustic analysis (Rasier & Hiligsmann 2009) or were on a small scale and more concerned with analyses of formal properties of intonation and not with its functional meaning (Mennen 2004). Furthermore, former studies hardly ever took into account another factor that could be of importance in bi-directional prosodic interference, which is the proficiency level of the L2 learners. Prior work by Swerts & Zerbian (2010) on Zulu and English suggests that interference is reduced when the proficiency level of the L2 learner increases, yet their findings were based on non-acoustic analyses and did not incorporate bi-directionality. In short, although prior work has looked into several aspects of the L1-L2 relationship that seem to be relevant to SLA, no study has ever bi-directionally and systematically compared the competing prosodic systems of an L1 and an L2, while taking into account language proficiency. Moreover, most previous studies in this area are somewhat limited in scale, and have not been concerned with functional aspects of intonation, e.g. in the way it is used to mark information status in typologically different languages. In comparison with earlier work, this is the first study that we know of using an experimental design in which native and non-native Spanish and Dutch are directly compared by means of an acoustic analysis of the intonation production data of L1 speakers and early and late L2 learners. Dutch and Spanish are especially fitting for a contrastive analysis since they are known to differ with respect to the way they use intonation to mark information status. In Dutch, contextually important or new information gets accented and less important or given information is deaccented (Swerts, Krahmer & Avesani 2002). In Spanish, however, both new and given information receive a falling tone (Ramírez Verdugo 2002), and deaccentuation is not applied very often. The data collected in this study might also tell us more about the typological intonation distance between languages and consequently help us answer the question whether it is equally difficult for speakers of Dutch to acquire the Spanish intonation pattern as it is for speakers of Spanish to adapt to the Dutch intonation system. It is expected that it is easier for L1 speakers of Dutch to acquire a native-like intonation in Spanish than it is for L1 speakers of Spanish to do so for Dutch. This can be explained by the fact that Spanish has a fixed rule for using intonation in information status marking, while in Dutch the use of a certain intonation pattern is context-dependent (Rasier & Hiligsmann 2009). Method For our study 123 participants performed a speech production task designed to elicit intonation patterns in different contexts. The participants were chosen as belonging to one of the following groups: 1) native speakers of Dutch, without knowledge of Spanish (NSD), 2) native speakers of Spanish, without knowledge of Dutch (NSS), 3) less proficient Dutch learners of Spanish (DLS-), 4) proficient Dutch learners of Spanish (DLS+), 5) less proficient Spanish learners of Dutch (SLD-) and 6) proficient Spanish learners of Dutch (SLD+). During the experiment they were asked to describe short series of images depicting objects in varying colours. The fourth picture of this sequence was the target image (e.g. a blue balloon) that could have one of the following four types of information status: 1) New/New (NN), in which both the first and the second word of the NP were new in the list of pictures (e.g. green donkey, followed by blue balloon), 2) Given/New (GN), in which the second word of the NP differed from the preceding NP, but the first one was the same (e.g. blue donkey followed by blue balloon), 3) New/Given (NG), in which the first word was new in that list, but the second word corresponded (only) with the second word in the preceding picture (e.g. red balloon, followed by blue balloon) or 4) Given/Given (GG), in which both the first and the second word were used to describe the preceding picture, but not any other picture in the sequence (e.g. blue balloon, followed by blue balloon). We had four different target NPs, which resulted in 16 experimental items, which were randomized and alternated with filler items. It is expected that Dutch participants only accentuate the first word of the NP in the NG condition (e.g. red balloon, BLUE balloon), while the Spanish always accentuate the last word of the NP, irrespective of whether the first word is new or given. In addition, during the task it became clear that speakers of Spanish tended to produce the elicited NPs with a list intonation that uniformly ended in a very high boundary tone, whereas such boundary tones were less extreme in data of the speakers of Dutch. In the acoustic analysis, we measured the pitch maximum (F0) in Hertz of the two separate words of the NP. To account for the relative difference between them, the maximum F0 value of the first word was deducted from the maximum pitch value of the second word. This means that if the second word is stressed more than the first, the result is a positive difference score, and if the first word receives more stress than the second, the result is a negative difference score. It is expected that in native Spanish all the information status types receive more or less the same positive difference score, while in native Dutch at least those items in which the first word is new and the second is given (NG) receive a negative one since the first word is accentuated and the second one is not. Results & Discussion Our data confirm that Spanish and Dutch truly have competing prosodic systems concerning intonation as predicted. As figure 1 reveals, the prominence patterns produced by the Dutch speakers match the given/new distinctions in the elicited NPs (especially, when comparing the NG to the GN contexts, which result in markedly contrasting difference scores), whereas the Spanish speakers produce essentially the same intonation patterns across the different contexts. Furthermore, it appears that prosodic interference indeed takes place (see figures 2 and 3), most notably in the use of very high boundary tones by all native speakers of Spanish and the absence thereof in the speech of native speakers of Dutch. But even though the difference scores of the Spanish participants are much higher than those of the Dutch as a result of their use of higher boundary tones, the intonation pattern per information status type does seem to become more native-like when the proficiency level increases. In other words, the DLS produce less and less variation among the four contexts, and the SLD start to differentiate more between the information status types. As shown in figure 4, there is some variance between the production data of native speakers and the L1 data of SLD and DLS, which entails that prosodic interference might be bi-directional in the process of information status marking by means of intonation. Proficient SLD start to accent the new information in GN and NG contexts in Spanish as they do in their L2. Conversely, the DLS show less variation among information status types in their L1, which results in almost identical patterns in their L1 and L2, were in not for the fact that they produce higher boundary tones in their L2. The results of a more in-depth data analysis will be available at the conference. References Mennen, I. (2004). ‘Bi-directional interference in the intonation of Dutch speakers of Greek’, Journal of Phonetics, 32, 543-563. Pavlenko, A. & Jarvis, S. (2002). ‘Bidirectional transfer’, Applied Linguistics, 23, 2, 190-214. Ramirez Verdugo, D. (2002). ‘Non-native interlanguage intonation systems: a study based on a computerized corpus of Spanish learners of English’, ICAME Journal, 26, 115-132. Rasier, L. & Hiligsmann, P. (2009). ‘Exploring the L1-L2 Relationship in the L2 Acquisition of Prosody’, Online proceedings of the international conference First and second languages: exploring the relationship in pedagogy-related contexts. Consulted on 22 October 2012 on: www.education.ox.ac.uk/wordpress/wp-content/uploads /2010/08/RasierHiligsmann.doc. Swerts, M., Krahmer, E. & Avesani, C. (2002). ‘Prosodic marking of information status in Dutch and Italian: a comparative analysis’, Journal of Phonetics, 30, 629-654. Swerts, M. & Zerbian, S. (2010). ‘Intonational differences between L1 and L2 English in South Africa’ Phonetica: internationale Zeitschrift fuer Phonetik, 67, 3, 127-145.