Most language learners reach a point where they feel confident speaking. Grammar is mostly automatic, vocabulary is solid, and conversations flow without constant hesitation. Then a native speaker reacts with that polite, knowing pause, and the learner realises there is still a gap they were not aware of. That gap, almost always, is pronunciation.
What makes pronunciation uniquely difficult is not that learners struggle with it openly. It is that they mispronounce words they feel completely certain about. Commonly mispronounced words in a foreign language are not the difficult ones, they are the familiar ones, practiced daily with full confidence. That combination of habit and invisibility is what makes pronunciation the last frontier of genuine fluency, and it has a name in linguistics, fossilization.
What Is Fossilization in Language Learning?
Fossilization is the process by which incorrect language patterns become so deeply ingrained that they resist correction even with explicit instruction. The concept was introduced by linguist Larry Selinker in his foundational 1972 paper on interlanguage, where he estimated that approximately 95% of adult second language learners fail to achieve native-like competence, only around 5% cross that threshold across all linguistic domains.
Pronunciation is where fossilization is most audible and most persistent. Unlike grammar errors, which a learner can catch and self-correct in writing, fossilized pronunciation errors are neurological. The brain has mapped a foreign sound to a familiar motor pattern, and without targeted retraining, that pattern stays fixed regardless of how much time passes or how much exposure the learner accumulates.
A 10-year longitudinal study published in Language Awareness by Thomson, Derwing, and Munro, tracked adult ESL learners from Mandarin and Slavic language backgrounds, measuring accentedness, comprehensibility, and fluency at the 2-month, 1-year, 2-year, 7-year, and 10-year marks. The Slavic language speakers showed measurable improvement in fluency and comprehensibility during the first two years. The Mandarin speakers showed no significant change at any point, not even after a decade of naturalistic immersion. Neither group reached native-like accentedness.
The conclusion is direct, exposure alone does not reverse fossilization. Time in a language environment improves fluency up to a point, but it does not retrain phonological patterns that have already solidified.
Why Commonly Mispronounced Words Feel Correct
Fossilization is particularly deceptive because it does not feel like a problem from the inside. A learner who consistently mispronounces a word has heard their own version hundreds or thousands of times, and it sounds right to them. Native speakers detect it immediately.
The examples span every major language. Russian stress patterns trip up virtually every anglophone learner "babushka" is stressed on the first syllable, not the second, but English speakers conditioned by penultimate stress in Latinate borrowings almost universally shift it. In Mandarin, the third tone changes when two third-tone syllables appear in sequence, a sandhi rule that affects the most basic greeting in the language. Learners who have studied Mandarin for years often produce it incorrectly because they learned tones in isolation, not in connected speech.
Italian borrowings in English follow the same pattern. "Bruschetta" is brusketta, not brushetta. "Gnocchi" is nyokki. "Porsche" has two syllables with a final vowel, a fact that surprises a significant number of English speakers who have used the word confidently for decades. These are not beginner errors. They are advanced errors, practiced into fluency.
The scale of the problem is visible in search data. Individual pronunciation queries, "gyro pronunciation," "porsche pronunciation," "acai pronunciation", each attract between 7,000 and 15,000 monthly searches in the US alone. People know these errors exist and actively seek to correct them, but only after an external trigger. Without one, most speakers never audit their pronunciation at all.
Why Some Sounds Are Cognitively Invisible
Part of what makes fossilization so stubborn is that certain phonemes simply do not exist in a learner's native language. When a sound has no equivalent, the brain substitutes the nearest available match automatically and unconsciously, before the learner is even aware it has happened.
Native English speakers learning Hindi face this with voiced aspirated consonants like the "bh" that opens the country's own name in Hindi. The aspiration exists in English, but not combined with voicing in that configuration, making the sound nearly imperceptible until it is explicitly demonstrated. Finnish double consonants carry semantic meaning, "tuli" means fire, "tulli" means customs, but foreign learners frequently cannot hear the distinction, let alone reproduce it. The problem runs the other way too, Japanese phonology does not permit consonant clusters, so Japanese speakers learning English automatically insert vowels between consonants, not from ignorance of English rules but because their phonological system processes those sounds before conscious awareness can intervene.
What This Means for Professional Language Services
For translation companies, language service providers, and organisations using multilingual communication, fossilization in pronunciation is not a peripheral concern. It sits at the centre of interpretation quality, voiceover accuracy, and the credibility of multilingual corporate communication.
An interpreter who subtly mispronounces a delegate's name or a key technical term during a high-stakes negotiation introduces friction. A voiceover artist delivering correct words with wrong stress patterns reduces the clarity of pharmaceutical or legal audio content. Research by Derwing and Munro, whose work on second language speech comprehensibility spans three decades, demonstrates that even a heavily accented speaker can be intelligible in controlled settings, but professional contexts do not offer controlled settings. In diplomatic interpreting, courtroom translation, and medical communication, the listener is not always trained, the stakes are high, and the margin for misunderstanding is close to zero.
This is why rigorous language service providers vet for native-speaker competency in the target language, not merely advanced proficiency. The difference between the two is invisible in written output and immediately audible in speech.
What Actually Fixes Fossilized Pronunciation
Simple immersion does not fix fossilization, the Thomson, Derwing, and Munro study makes that unambiguous. What works is explicit, targeted instruction with calibrated native-speaker feedback focused on specific sounds.
A 2023 study in Frontiers in Education, found that perception-based instruction, training learners to hear the correct sound before attempting to produce it, was more effective than production-based instruction alone for improving pronunciation in adult EFL learners. The implication is practical, self-directed practice, however disciplined, rarely corrects fossilized errors without an external, native-speaker ear providing precise feedback.
So, commonly mispronounced words are not a problem of effort or intelligence. They are a predictable outcome of how adult brains acquire phonology, compounded by the false confidence that comes with genuine communicative fluency. Fossilization in language learning is well-documented, affects the vast majority of adult learners, and is resistant to correction without targeted intervention.
For language learners, the answer is explicit feedback from native speakers, not more exposure. For organisations that depend on professional language services, the answer is identical, native-speaker expertise is the baseline that ensures pronunciation fossilization does not become a communication failure at the moments that matter most.