![Sample our Language & Literature journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5941/TOC-Subject-Sample-2021-Language-Literature.jpg"})
Abstract
This study examines the relationship between German second language (L2) learners’ awareness of the German lexical stress assignment system and their ability to accurately assign stress to cognate words with predictable lexical stress. Participants were 31 adult L2 German learners from three groups: native English speakers with a range of German proficiency levels (N = 10), native French speakers with intermediate German proficiency (N = 10), and native French speakers with advanced German proficiency (N = 11). They produced target items in a carrier phrase and then indicated both which syllable they stressed and where stress is supposed to fall. Finally, they provided a rule for assigning stress to each word. Stress production accuracy was similar across the groups, regardless of L1 or L2 proficiency. Participants’ ability to verbalize where they had placed stress was a significant predictor of stress assignment accuracy. They produced relatively few rules overall, and the rules they produced were mostly inaccurate. The results point to the greater importance of self-assessment over the ability to produce discrete rules in accurate lexical stress assignment.
Acknowledgements
The author would like to thank Stephanie Chicoine, Ross Sundberg, and Helene Bramwell for their assistance in data collection and to the reviewers for their valuable feedback on earlier versions of the manuscript.
Disclosure statement
No potential conflict of interest was reported by the author.
This work was supported by a grant from the Social Sciences and Humanities Research Council of Canada.
Notes
1. It is possible to operationalize lexical stress assignment acoustically, for example, by measuring the duration, F0, or intensity of syllables relative to others in the word.
2. The penultimate, or next-to-last, syllable is considered the default location for lexical stress assignment in German. It is important to note that the penultimate syllable and second syllable are only identical in three-syllable words.
3. An underlying assumption of ANOVAS (and by extension, the Tukey HSD) is that all sample sizes are equal or nearly equal. In this study, the ns of 10, 10, and 11 are in line with this requirement. In addition, the analysis allows for “considerable deviations from normality” (Zar, Citation1998, p. 185). Thus, the underlying assumptions of the analysis are upheld.
4. Although read-aloud tasks may elicit more constrained, and thus potentially less valid, speech samples than spontaneous tasks do (e.g., Campbell, Citation1995), it would be impossible to elicit such a wide range of cognate words in a spontaneous task.
5. There were originally 65 target items in the current study, but one English stress cognate, <Professor>, was significantly more frequent than all of the others. For that reason, it was excluded from all analyses presented here. Production results are not significantly different with this item included.
6. When reporting the results of GEE analyses, researchers include the following: EMM (estimated marginal mean from the GEE model), Wald χ2 (a χ2 statistic that provides information about whether or not coefficients are equal to zero, where χ2 = (B/SE)2), B (a parameter estimate) and SE (standard error).
7. The results of an analysis group as a between-subjects factor and cognate type as a within-factor indicate no interactions between group and cognate type.
8. The proficiency groups patterned similarly in their performance. Participants in both proficiency groups were most accurate in their production of schwa cognates. Intermediate proficiency participants produced schwa cognates more accurately than both English stress cognates (B = 2.487, SE = .655, χ2 (1) = 14.42, p < .001) and French stress cognates (B = 2.758, SE = .502, χ2 (1) = 30.22, p < .001). Advanced proficiency participants also produced schwa cognates more accurately than English stress cognates (B = 2.094, SE = .751, χ2 (1) = 7.77, p = .005) and French stress cognates (B = 1.548, SE = .674, χ2 (1) = 5.28, p = .022).
9. The one statistically significant group difference is in the production English stress cognates. Native speakers of English produced this set of words significantly more accurately than did French L1 intermediate learners.
10. As pointed out by an anonymous reviewer, L1 effects may have been uncovered with more participants.
11. It is certainly possible that the results of the current study only apply to lexical stress assignment in cognate words. Nonetheless, given participants’ general inability to produce rules, it is quite likely that they have few rules for the rest of the system.
12. It is important to note that native speakers of a language are often unable to provide rules about their own productions (Sanz, Citation2013).
13. Because French L1 advanced learners produced all of the schwa cognates correctly, there was no variation in the data, and the GEE model treated the data as a constant. As such, the analyses for schwa cognates only compare the performance of native speakers of English and French L1 intermediate learners.
Additional information
Notes on contributors
Mary Grantham O’Brien
Mary Grantham O’Brien is a Professor of German at the University of Calgary. There she teaches courses on German language, linguistics, and second language acquisition. She conducts research on second language learners’ speech perception and production, and she is also interested in listener judgments of nonnative speech.