The intelligibility principle, why accent reduction is the wrong goal for your students.

Murray Munro and Tracey Derwing ran a study in 1995 that should have ended the accent-reduction industry. It did not. The study is the empirical foundation for what pronunciation teaching has called the intelligibility principle for thirty years.

Three things that are not the same

Munro and Derwing's 1995 paper (Foreign Accent, Comprehensibility, and Intelligibility in the Speech of Second Language Learners, Language Learning) made a distinction the field has tried not to forget. There are three separate constructs in any utterance from an L2 speaker, and they vary independently.

Intelligibility

Did the listener actually understand the message? Measured as the proportion of the utterance correctly transcribed word-for-word. This is the only one that affects the speaker's ability to function in the language.

Comprehensibility

How much effort did the listener have to expend to understand? Measured by listener self-rating on a scale. Two speakers can be equally intelligible but differ in how hard the listener worked.

Accentedness

How different did the speaker sound from a reference variety? Measured by listener rating of perceived difference. This is the metric the accent-reduction industry sells. It is the least useful of the three.

The finding that should have settled it

Munro and Derwing recorded ten Mandarin-accented English speakers reading the same passages and played the recordings for native English listeners. They asked the listeners to transcribe (intelligibility), rate the effort (comprehensibility), and rate the accent (accentedness).

The three measures did not move together. Speakers could have heavy perceived accents and still be one hundred percent intelligible. Speakers with mild perceived accents could still cause comprehension breakdowns when their suprasegmentals (stress, intonation, rhythm) were off. Accent was real but not the variable that predicted whether communication succeeded.

John Levis built the pedagogical argument on this foundation (Changing Contexts and Shifting Paradigms in Pronunciation Teaching, TESOL Quarterly, 2005). The intelligibility principle says the goal of pronunciation teaching is for the speaker to be understood, not for the speaker to sound native. The nativeness principle (the traditional goal) sets a target most adult learners cannot reach and would not need to reach if they could.

The intelligibility principle holds that learners simply need to be understandable. There is no reason to expect learners to acquire a native accent, and unsuccessful attempts to do so are likely to discourage learners.

What features actually matter for intelligibility

Tracey Derwing and Murray Munro spent the next two decades identifying which pronunciation features actually predict intelligibility breakdowns. Their 2015 book (Pronunciation Fundamentals, John Benjamins) is the standard summary. The ranked list looks roughly like this.

1. Word stress

Stressing the wrong syllable is the single most reliable way to make a Spanish word incomprehensible to a Spanish speaker. Estudio stressed on the wrong syllable becomes estudió. The whole tense changes. This is the most teachable feature with the most acquisition leverage.

2. Sentence-level rhythm and stress

Spanish has a syllable-timed rhythm. English has a stress-timed rhythm. American students applying English rhythm to Spanish produce sentences that listeners cannot chunk. The fix is repeated exposure to native-paced audio with shadow-reading.

3. Intonation contour for questions and statements

Spanish yes-or-no questions rise sharply at the end. Information questions fall. Statements have a relatively flat contour. Students who keep English intonation patterns produce utterances that are intelligible word-by-word but confuse listeners about whether a question was asked.

4. Vowel quality on the five core vowels

Spanish has five clean vowel positions (a, e, i, o, u). English speakers reduce unstressed vowels to schwa, which Spanish does not have. The schwa is the most reliable accent marker for English-speaking learners of Spanish, and it does affect intelligibility on words with multiple unstressed syllables.

5. Specific consonant contrasts

The /r/-/rr/ distinction. The /b/-/v/ collapse that English speakers add. The /d/ between vowels that should weaken to the dental fricative. These are real targets but lower in priority than the suprasegmentals.

The features that do not matter much

American students who roll their /r/ imperfectly. American students who keep the aspirated /t/. American students who use the English /a/ instead of the Spanish /a/. These show up as accent. They almost never cause an intelligibility breakdown. Time spent drilling them is mostly wasted.

Levis's claim is not that segmentals do not matter at all. It is that pronunciation teaching has spent decades prioritizing the segmentals that are most teachable and least intelligibility-relevant, while ignoring the suprasegmentals that drive most breakdowns.

What this means for heritage learners

Heritage students often have native or near-native pronunciation in conversational Spanish and dialect-specific features that are not the textbook's reference variety. The intelligibility principle is the right defense against the trap of correcting their natural pronunciation to match a Castilian or Mexican-standard reference.

A Salvadoran-American student who drops final /s/ is using a feature of their family's dialect. It is intelligible to other Salvadoran Spanish speakers and to most Latin American Spanish speakers. Correcting it as a pronunciation error is a sociolinguistic violation. The intelligibility principle says: the student is already intelligible, and the variety they speak is legitimate. Move on.

How to teach for intelligibility

Front-load stress and rhythm

In novice and intermediate Spanish, build five-minute weekly rhythm and stress activities. Choral reading of marked text. Shadow-reading of native-speaker audio. Listen and clap on the stressed syllable. The activities are boring on paper and effective in measurable ways.

Use minimal pair work for stress contrasts

Hablo versus habló. Estudio versus estudió. Practico versus practicó. The same letters, different stress, two completely different meanings. Build the listening discrimination first, then the production.

Stop correcting accents

If a student's utterance is intelligible, do not comment on the accent. The most common pronunciation intervention I see in Spanish classes is the teacher repeating a perfectly intelligible student utterance with a more native-sounding accent. The student understood. The listener understood. The correction is cosmetic. It costs the student confidence with no acquisition gain.

What this means for your Monday

Audit your last week of class. Count the times you corrected pronunciation. For each correction, ask: was the original utterance intelligible to me? If yes, the correction was probably unnecessary. If no, was the issue a suprasegmental (stress, rhythm, intonation) or a segmental (specific sound)?

For the next two weeks, only correct pronunciation when the utterance was unintelligible. Within those corrections, prioritize stress and rhythm. Notice what happens to student willingness to speak. The intelligibility principle pays off twice: in acquisition gains and in classroom climate.