TL;DR
This paper investigates how native (L1) and non-native (L2) speakers of Estonian produce and perceive the language's complex three-way quantity system. The study reveals that while L2 speakers can distinguish the first two quantity degrees, they struggle significantly with the third, highlighting a key challenge in acquiring this phonological feature.
What Problem Does This Paper Address?
Estonian possesses a unique and challenging phonological feature: a three-way contrast in "quantity" within disyllabic feet. This isn't just about vowel length; it's a complex interplay of vowel and consonant durations, along with tonal patterns, that distinguishes three degrees of prominence: short (Q1), long (Q2), and overlong (Q3). For speakers whose native languages don't rely on such fine-grained durational contrasts (like Russian), acquiring this system is a significant hurdle. This paper aims to understand the extent of this difficulty by examining both how L1 and L2 speakers produce these quantity distinctions and how they perceive them, specifically focusing on the production and perception of these ternary oppositions.
Methods
The researchers employed two primary methods:
- Production Analysis: They analyzed speech recordings from native Estonian speakers (L1) and Russian-native speakers learning Estonian (L2). The speech material consisted of disyllabic words embedded in sentence contexts. Acoustic analysis, using the Praat software, measured vowel durations (V1 and V2) and calculated the V1/V2 duration ratio for words representing the three quantity degrees (Q1, Q2, Q3). Statistical analysis (ANOVA) was used to compare the groups and the effects of quantity.
- Perception Experiment: A carefully constructed set of synthesized CVCV stimuli was created, varying the durations of the first and second vowels (V1 and V2) and manipulating the fundamental frequency (FO) contour. These stimuli formed continua from Q1 to Q2 and from Q2 to Q3. Participants (both L1 and L2 speakers) listened to these stimuli and performed a forced-choice identification task, categorizing each sound as belonging to one of the two quantity degrees in the continuum. Categorization functions and category boundaries were then analyzed.
Datasets
- Estonian Foreign Accent Corpus: This corpus was used for the acoustic analysis of naturally produced speech. It contains recordings of sentences and prompts from both native and non-native speakers.
- Synthesized Stimulus Set: A novel stimulus set was created for the perception experiment, consisting of isolated CVCV words with manipulated vowel durations and FO contours. This set was designed to create continua for Q1-Q2 and Q2-Q3 distinctions.
Key Insights and Ideas
- L2 Speakers Excel at Q1 vs. Q2: Non-native speakers showed a remarkable ability to distinguish and produce the Q1 and Q2 quantity degrees, mirroring native speakers to a significant extent. This suggests that the basic durational cues differentiating these two categories are learnable.
- The Q2 vs. Q3 Divide: The real challenge for L2 speakers lies in differentiating Q2 from Q3. They struggled significantly in both production and perception of this distinction.
- V1 Duration is Key (for L2): For L2 speakers, the duration of the first vowel (V1) emerged as the primary cue for distinguishing Q1 from Q2. However, for the Q2 vs. Q3 distinction, even V1 duration wasn't enough to reliably differentiate the categories.
- V2 Duration and FO are Less Critical for L2: Interestingly, the duration of the second vowel (V2) and the FO contour, which play a role for native speakers, had little to no significant effect on L2 speakers' perception of Q2 vs. Q3. This suggests L2 learners simplify the complex acoustic cues.
- Feature Hypothesis Supported: The findings align with the "Feature Hypothesis," which posits that phonological features not present in a speaker's native language are difficult to acquire. The Estonian Q2-Q3 contrast, lacking a direct parallel in Russian, proves particularly problematic.
Key Results
- Production: L1 speakers showed significant differences in V1, V2, and V1/V2 ratios across all three quantity degrees. L2 speakers showed significant differences in V1 and V1/V2 ratios for Q1 vs. Q2, but not for Q2 vs. Q3.
- Perception:
- For Q1 vs. Q2, both L1 and L2 groups showed similar categorization functions, with V1 duration being the most significant factor.
- For Q2 vs. Q3, L1 speakers demonstrated clear category boundaries influenced by V1, V2, and FO. In contrast, L2 speakers' categorization was primarily driven by V1 duration, but the observed "boundary" was likely an artifact of the experimental design rather than a true perceptual distinction, as they failed to reliably differentiate Q2 and Q3.
Limitations and Future Work
- Limited Stimulus Set: The perception experiment used a simplified CVCV structure. Future work could explore more complex word structures.
- Focus on Russian L1: While Russian is a relevant comparison, future studies could investigate L2 learners from other language backgrounds to see if similar patterns emerge.
- "Apparent" L2 Boundary: The paper notes that the Q2-Q3 boundary observed for L2 speakers in perception might not be a "real" boundary but rather an induced response strategy. Further research could explore how to train L2 speakers to develop genuine Q2-Q3 distinctions.
- Production-Perception Link: While the paper compares production and perception, future work could more deeply investigate the precise relationship between production accuracy and perceptual sensitivity for L2 learners.
Summary updated