Training Effects on the Perception of Final Melodic Contours in L2 French Questions
Abstract
The perception and production of several prosodic phenomena (stress, intonation, and rhythm) often pose a
challenge for adult learners of a second language (Mennen, 2015; Jongman & Tremblay, 2020). Research has
shown that while explicit instruction improves oral fluency and native-like rhythm in Anglophone L2 French
learners (Drouillet et al, 2024), implicit approaches, such as the Verbo-Tonal Method, enhance prosodic
mastery in reading and spontaneous speech (Alazard et al., 2010; Alazard, 2013; Saito & Plonsky, 2019). To
our knowledge, no prior studies, however, have compared explicit (otherwise known as Form-Based, FB)
versus implicit (Meaning-Based, MB) instruction effects on the perception of difficult prosodic patterns in
ecologically valid classroom settings.This study addresses two questions: (1) To what extent are beginner
Anglophone learners sensitive to the prosodic cues distinguishing yes/no and wh-questions in French, and how
does this sensitivity change with training? (2) Which type of training, Form-Based or Meaning-Based, yields
greater improvements in the perception of final melodic contours in L2 French questions?
Thirteen English-speaking undergraduates enrolled in beginner L2 French courses in the U.S. (ages
18–21; M = 19.9 y/o; 8 women, 5 men; A1–A2 CEFR proficiency after one semester ~ 45–50 hours)
participated in our study. Participants were randomly assigned to FB (n = 7) or MB (n = 6) conditions,
completing 7.5 hours of standard curriculum instruction over two weeks.
In FB classes, the instructor provided explicit metalinguistic corrections, prompted self-correction,
reformulated learner output, and used paralinguistic cues (gestures, exaggerated intonation). MB classes
emphasized recasts, clarification requests, and implicit noticing of prosodic forms through communicative
tasks. To assess participants’ sensitivity to intonation, 20 questions (10 yes/no and 10 wh-) were recorded,
each produced in three manipulated versions using PRAAT: canonical, non-native 1, and non-native 2. For
yes/no questions, the final f0 contour on the last syllable was modified to create (1) a native-like late rise
(L)H H% (6–8 ST), (2) an over-amplified rise (L)HHH% (+2–4 ST, total 8–12 ST), and (3) a falling
contour L L%. The same three manipulations were applied to open wh- questions: (1) a native-like fall L
L%, (2) a rise H% (+6–8 ST), and (3) an over-amplified rise HH% (+2–4 ST).
Results show that question type significantly influenced performance: wh-questions consistently proved
more challenging than yes/no questions. MB training produced slightly stronger gains for interrogative
intonation recognition, particularly for distinguishing canonical falls from exaggerated rises in wh-questions.
Error patterns revealed persistent L1 influence, with learners favoring exaggerated rises (HH%), which aligns
with previous studies on L2 learner preferences (Santiago et al., 2014).
Our results suggest that ecologically valid classroom prosody training supports more consistent perceptual
patterns for L2 French interrogatives, with Meaning-Based instruction showing modest advantages over
Form-Based approaches for melodic contour recognition. These findings highlight the feasibility and value of
early prosodic instruction within standard beginner curricula, addressing a critical gap in L2 acquisition.
