When a person learns a second language (L2), achieving a native-like accent involves far more than memorizing vocabulary and grammar. It requires a fundamental restructuring of both how they perceive auditory signals and how they physically manipulate their vocal apparatus. This process is not a simple linear progression of mimicry; rather, it is a proficiency-dependent convergence where a learner's internal phonological maps, prosodic timing, and motor habits gradually realign with the target language's acoustic realities.
The Perceptual Foundation: Re-mapping the Auditory Space
Before a learner can accurately produce a sound, they must be able to perceive it. In the initial stages of language acquisition, the brain filters L2 sounds through the sieve of the native language (L1) phonological system. As proficiency develops, this filter is systematically rebuilt.
Paola Escudero et al. in Formal modelling of L1 and L2 perceptual learning: computational linguistics versus machine learning demonstrated that advanced learners' perceptual systems actively converge toward native norms, carving out entirely new, distinct sound categories. This isn't just a passive byproduct of exposure; it is a refinement of acoustic processing. For instance, Wenfu Bao et al. (Perception of Mandarin tones by native tibetan speakers) found that higher proficiency allows learners to map complex phonological features, such as Mandarin tones, with greater consistency and predictability, aligning their perception directly with physical acoustic reality.
This perceptual sharpening is especially evident in how learners handle dynamic, time-varying acoustic information. Linda Bakkouche et al. (Influence of Proficiency and L2 Experience on Dynamic Spectral Cue Utilization in L2 Vowel Perception and Production) showed that advanced proficiency significantly enhances a learner's ability to utilize dynamic spectral cues to distinguish challenging vowel contrasts, which directly feeds into better production.
However, this perceptual learning can encounter a plateau. As observed by Kimiko Tsukada et al. in both Cross-language perception of Japanese singleton and geminate consonants: preliminary data from non-native learners of Japanese and native speakers of Italian and australian English and Non-native Perception of Japanese Singleton/Geminate Contrasts: Comparison of Mandarin and Mongolian Speakers Differing in Japanese Experience, while increased instruction and experience improve the perception of novel phonemic contrasts like consonant length, advanced learners often face a "ceiling effect" where their performance approaches, but does not completely match, native benchmarks.
Interestingly, while proficiency alters acoustic mapping, it does not always change the underlying structural processing strategies. Tamami Katayama (Effects of L1 Phonotactic Constraints on L2 Word Segmentation Strategies) found that second language proficiency influences overall speech segmentation performance, but both high- and low-proficiency learners rely on the same core syllable-based segmentation strategies, showing that some universal processing mechanisms persist regardless of language mastery.
Articulatory Refinement and the Suppression of L1 Interference
As the perceptual map becomes clearer, the motor control system adapts to minimize production errors. This manifests as a shift from clumsy L1-based approximations to precise, target-like articulatory gestures.
In the early stages, beginners rely on phonetic strategies imported from their native language. For example, Frank Zimmerer et al. (Productions of /h/ in German: French vs. German speakers) found that French beginners learning German often substitute glottal stops for the unfamiliar /h/ sound. As proficiency grows, advanced learners successfully suppress these L1 habits, achieving a significantly higher rate of native-like production.
A similar pattern of error reduction and articulatory refinement was documented by Dan Du et al. (The Production of Chinese Affricates /ts/ and /tsh/ by Native Urdu Speakers), who noted that as proficiency increases, both the frequency and variety of pronunciation errors drop, with remaining errors falling closer to the target phonetic space. In vowel production, Frank Zimmerer et al. (Perception of French speakers' German vowels) demonstrated that advanced learners achieve substantially higher accuracy than beginners, who frequently make lengthening and shortening errors due to L1 interference.
This convergence is particularly striking when mastering complex phonological structures like geminate (doubled) consonants. Work by Sonia d'Apolito et al. in L2 Pronunciation Accuracy and Context: A Pilot Study on the Realization of Geminates in Italian as L2 by French Learners and Context and Co-Text Influence on the Accuracy Production of Italian L2 Non-Native Sounds shows that while beginners' production of geminates is heavily disrupted by L1 constraints and communicative context, advanced learners consistently internalize and execute these distinct consonant lengths. This represents a true suppression of L1 interference, a phenomenon also observed by Anne Bonneau et al. (German non-native realizations of French voiced fricatives in final position of a group of words) regarding the voicing of final fricatives.
Suprasegmental and Prosodic Realignment
Phonological convergence extends beyond individual vowels and consonants into the global architecture of speech: rhythm, intensity, and pitch (suprasegmentals).
Languages differ fundamentally in their rhythmic timing—some are stress-timed (like English), while others are syllable-timed (like Korean or French). Jong-Pyo Lee et al. (A comparative study on the production of inter-stress intervals of English speech by English native speakers and Korean speakers) discovered that proficient non-native speakers alter the duration of their inter-stress intervals to match the rhythmic patterns of the target language, whereas less proficient speakers fail to do so. This indicates that rhythm acquisition is highly dependent on language learning progression, a concept further supported by Hongwei Ding et al. (L2 English Rhythm in Read Speech by Chinese Students).
Similarly, prosodic intensity and pitch profiles shift with proficiency. Tomoko Nariai et al. (Comparative analysis of intensity between native speakers and Japanese speakers of English) demonstrated that lower-proficiency speakers exhibit pronounced deviations from native-like intensity patterns (such as over-emphasizing function words), but these deviations diminish as proficiency rises.
Yet, some suprasegmental features are incredibly resistant to change. Yike Yang et al. (F0 Patterns in Mandarin Statements of Mandarin and Cantonese Speakers) revealed that even highly fluent bilinguals often retain the "pitch fingerprint" (average pitch and range) of their native tongue, raising fundamental questions about whether certain global prosodic traits become permanently fossilized or remain open to long-term immersion.
Phonatory Adaptation and Vocal Effort
Perhaps the most subtle layer of convergence occurs at the level of the larynx: phonatory adaptation. Speakers must learn to adjust their vocal fold vibration patterns to match the characteristic voice quality of the target language.
Learners demonstrate a remarkable, though sometimes asymmetrical, ability to adapt these habits. For instance, Agathe Benoist-lucy et al. (The influence of language and speech task upon creaky voice use among six young American women learning French) observed that American women learning French dynamically modulate their use of creaky voice depending on the language they are speaking and the communicative context, showing sensitivity to target phonatory norms.
Intriguingly, this phonatory adjustment often requires significant physical effort as proficiency increases. Manfred Pützer et al. (Evaluation of Phonatory Behavior of German and French Speakers in Native and Non-Native Speech) discovered that advanced L2 learners, regardless of their native language, frequently adopt a more "strained" vocal production in their non-native speech compared to beginners.
Why does this happen? Rather than a sign of failure, this increased vocal tension reflects an active, effortful motor strategy to approximate native-like phonatory characteristics and overcome deep-seated L1 muscular habits. Beginners often sound more relaxed simply because they are not yet attempting to alter their baseline laryngeal settings.
Why Proficiency Unlocks Interactive Alignment
If phonological convergence is the structural realignment of a speaker's vocal habits, interactive alignment is its real-time manifestation during conversation. In dialogue, speakers naturally mirror each other's linguistic choices. Crucially, this capacity to align is itself a function of proficiency.
Intuitively, one might think that beginners would mirror their conversational partners more heavily as a crutch to get by. However, Arabella Sinclair et al. (Does Ability Affect Alignment in Second Language Tutorial Dialogue?) demonstrated the exact opposite: the ability to linguistically align with an interlocutor increases as a learner's proficiency grows.
The mechanism behind this is cognitive bandwidth. Lower-proficiency learners expend nearly all of their cognitive resources on basic lexical retrieval and grammatical formulation, leaving no mental real estate for tracking and mirroring the subtle phonetic or lexical choices of their partner. Higher proficiency automates these foundational processes, freeing up the cognitive resources necessary to actively perceive, adapt, and converge toward the interlocutor's speech patterns in real time.
Key Foundational Work
A comparative study on the production of inter-stress intervals of English speech by English native speakers and Korean speakers (2004) — contributed 1 concept elements to this direction.
Formal modelling of L1 and L2 perceptual learning: computational linguistics versus machine learning (2007) — contributed 1 concept elements to this direction.