This study investigates the ongoing /l/–/n/ merger in Southwestern Mandarin through acoustic analysis of eight speakers (seven male), yielding a total of 1884 disyllabic words. Impressionistic transcription reveals a pronounced asymmetry: while 82% of /l/-initial tokens are realized faithfully as l, 54% of /n/-initial tokens are also realized as l, substantially reducing the contrast. Speakers cluster into two groups: an L-group (n = 5) who nearly categorically merge /n/ into l, and an N-group (n = 3) who partially reverse this pattern, often producing /l/ as n. Acoustic measurements focused on the onset consonant midpoint and the first 10% of the following vowel, targeting canonical lateral–nasal cues—F2–F1 spacing, F3 frequency, bandwidth of the first formant (BW1), and relative RMS amplitude—alongside vowel-initial spectral measures A1–P0 and ΔA1. While formant-based cues failed to show consistent group-level differences, A1–P0 and ΔA1 reliably distinguished l from n (0.01), underscoring the diagnostic strength of early-vowel energy cues in this regard. The merger is shown to be gradient, speaker-dependent, and sensitive to phonological context: tokens followed by nasal codas or nasal-initial syllables exhibit more resistance to neutralization. These findings refine our understanding of /l/-/n/ merger across Sinitic languages and demonstrate that seemingly symmetrical mergers may unfold along asymmetrical and acoustically distinct trajectories. Corresponding articulatory data will be presented at the conference.
Chang et al. (Wed,) studied this question.