Exploring the production and perception of Dutch word stress by Francophone learners of Dutch: Pilot study

Dutch is taught as a foreign language in French-speaking Belgium, as it is one of the three official languages. Although an official part of DFL (Dutch as a foreign language) teaching programmes (cf. the Common European Framework of Reference for Languages (CEFR)), prosody is still underrepresented in Belgian DFL classes where it is very often even completely left aside. However its crucial role for the success of communication with L2-speakers has been demonstrated in studies on lexical stress and sentence stress (Hahn 2004 on L2 English; Caspers, 2009, 2010 on L2 Dutch), and on prosody as a whole (e.g. Derwing & Munro 2005 on L2 English). A correct prosody seems to enhance intelligibility and to yield more positive native listener judgements about the speaker. Research has dealt with the production and perception of English and Polish stress by Francophone L2 learners (Altmann 2006, Kijak 2009). Yet very little is known about the native and non-native perception of DFL prosody by Francophone speakers, which is the topic of my current studies. Given the on-going struggle between the Francophone and the Flemish communities in Belgium, such a study would enable us to understand the potential role of prosody on cross-language misunderstandings. As far as prosody is concerned, Dutch and French are very different: Dutch has an unpredictable lexical stress, whereas French does not. Instead, French has a fixed final phrase prominence (Di Cristo 1999, 2000; Mertens, 1992). Word stress enables Dutch listeners to enhance word recognition (e.g. Cutler & van Donselaar, 2001; Tyler & Cutler 2009), whereas Francophones are sometimes claimed to be ‘stress-deaf’ (Peperkamp & Dupoux 2002). The pilot experiments carried out consisted of production and perception tasks. First, six Francophone DFL learners pronounced 27 three-syllabe high-frequency words in order to identify DFL spontaneous stress position. The stimuli were equally distributed over three conditions: initial stress, median stress, final stress. The expectation was that the final syllable would be preferred by the Francophone speakers due to transfer from French. The informants were first-year students at the Université catholique de Louvain (Belgium) who had had four to six years of Dutch at secondary school. In a perception experiment 26 native listeners listened to the same stimuli pronounced by a native and a Francophone. Word stress had been partially manipulated and as a result, three conditions were created per lexical stress condition: a first correct one (PAgina, Engl. “page”), and two manipulated ones (*alGEbra instead of ALgebra, *handeLEN instead of HANdelen, Engl. “to act”). 13 words were covered with babble noise at three SNRs (+5dB, +3 dB, 0dB) as it has been shown that Dutch natives are quite familiar with Francophones’ segmental deviations (Caspers & Horloza (to appear)). A reaction time (RT) task was carried out, the aim being to examine if words with a non-native pronunciation and a (non-) native-like stress are less intelligible to natives than native stimuli across the same word stress and noise conditions. Adding noise can indeed show us if T1 stimuli with correct and wrong stress are more robust in intelligibility tests than non-native equivalents. These pilot tests will be enhanced and replicated in larger experiments which will hopefully enable us to compare the effect of stress shifts on intelligibility (see also Van Heuven 1988) and to link these results to the word stress Francophones actually produce spontaneously. In the final experiment, Francophone listeners will also participate in the RT experiment. As they might not use stress for word recognition, it can be interesting to compare their results with native ones. In a second phase, Francophone DFL stress will also be tested for acceptability judgments thanks to rating scales by native listeners (cf. Hahn 2004, Derwing & Munro 2005).