The 7 Sounds That Break Intelligibility (And How to Fix Them)

You've been told to "work on your pronunciation." But which pronunciation? Every language has dozens of sounds, and you can't fix them all at once.

Here's what most pronunciation advice gets wrong: it treats all errors as equal. Mispronounce the "th" in "think"? That's treated as seriously as mixing up "ship" and "sheep." But research tells a very different story.

Some sound errors make listeners pause, rewind mentally, and ask you to repeat yourself. Others barely register. The difference comes down to a concept linguists call functional load—and understanding it can completely change how you prioritize your pronunciation practice.

What Is Functional Load (And Why Should You Care)?

Functional load is the amount of "work" a sound contrast does in a language. If two sounds distinguish hundreds of common word pairs, that contrast has a high functional load. If they distinguish only a handful of rare words, the functional load is low.

The concept dates back to the Prague School linguists in the 1930s, but it was applied to language teaching by Catford (1987) and Brown (1991), who ranked English consonant and vowel contrasts by how many minimal pairs they distinguish.

The landmark study by Munro and Derwing (2006), published in System, tested this empirically. They had native English listeners judge Cantonese-accented sentences containing various pronunciation errors. The result was clear: high functional load errors had large effects on perceived comprehensibility, while low functional load errors had minimal impact.

In plain terms: some mistakes break communication. Others don't. Here are the seven sound categories that research consistently identifies as the biggest intelligibility breakers across major world languages.

1. The /l/ vs. /r/ Distinction (English, and Beyond)

This is the most-studied contrast in second language pronunciation research, and for good reason. In English, /l/ and /r/ distinguish hundreds of common word pairs: light/right, lead/read, collect/correct, alive/arrive, lock/rock.

Brown's (1991) functional load ranking places the /l/–/r/ contrast among the highest in English. Munro and Derwing's (2006) study confirmed that errors on this contrast significantly reduced comprehensibility ratings.

This contrast is notoriously difficult for speakers of East Asian languages—Japanese, Korean, and Chinese—where /l/ and /r/ either don't exist as separate sounds or map differently. But it affects learners from many backgrounds.

How to fix it:

• Tongue position awareness. For English /l/, the tongue tip touches the ridge behind your upper teeth. For /r/, the tongue curls back slightly without touching anything.
• Practice in word pairs. Start with minimal pairs: light/right, glass/grass, fly/fry. Say them slowly, exaggerating the difference, then speed up.
• Record and compare. Use your phone to record yourself, then compare against a native speaker model. The gap is often smaller than you think.
• Use conversation practice. Tools like speaky.space give you real-time conversation where you can practice these sounds in context—not just in isolation drills.

2. Vowel Length Contrasts

In English, the difference between "ship" and "sheep," "bit" and "beat," "pull" and "pool" comes down primarily to vowel length and quality. These contrasts have extremely high functional load.

In Japanese, vowel length is phonemic—obasan (aunt) vs. obaasan (grandmother) differ by one beat of vowel length, and confusing them ranges from embarrassing to incomprehensible.

In German, vowel length changes meaning systematically: Staat (state) vs. Stadt (city), Höhle (cave) vs. Hölle (hell).

Research by Saito and colleagues (2019), published in Studies in Second Language Acquisition, found that vowel-related errors were among the strongest predictors of reduced comprehensibility in L2 English speech—stronger than most consonant errors.

How to fix it:

• Feel the duration. Long vowels aren't just longer—your mouth is often more tense and the sound more "pure." Short vowels are relaxed and centralized.
• Rubber band technique. Stretch a rubber band while saying long vowels, release for short ones. The physical action reinforces the timing difference.
• Practice with minimal pairs that highlight length. Bit/beat, full/fool, ship/sheep. Focus on timing, not just the sound quality.

3. Word Stress Placement

This isn't a single "sound," but it may be the most important item on this list. In stress-timed languages like English, Portuguese, and Russian, putting stress on the wrong syllable can make a word completely unrecognizable.

Research by Field (2005), published in TESOL Quarterly, found that misplaced word stress was more damaging to intelligibility than segmental (individual sound) errors in English. Native listeners could recover from wrong vowels or consonants more easily than from wrong stress placement.

Consider: DEsert (noun: sandy place) vs. deSERT (verb: to abandon). REcord vs. reCORD. PREsent vs. preSENT.

But it goes beyond these textbook examples. Every multi-syllable English word has a stress pattern, and deviating from it forces listeners to do extra processing work. Zielinski's 2008 study in Prospect found that word stress errors were the single most common cause of communication breakdowns in interactions with L2 English speakers.

How to fix it:

• Learn stress with every new word. When you learn "photography" (phoTOGraphy), mark the stress. Most dictionaries show it.
• Listen for the "bounce." Stressed syllables are louder, longer, and higher in pitch. Tap your hand on the table with each stressed syllable to feel the rhythm.
• Practice with shadowing. Imitating native speakers' rhythm trains stress patterns more effectively than rules.
• Use connected speech awareness to understand how unstressed syllables naturally reduce in fluent speech.

4. Aspiration and Voicing Contrasts

In English, the difference between /p/ and /b/, /t/ and /d/, /k/ and /g/ involves both voicing (whether the vocal cords vibrate) and aspiration (a puff of air after the sound). These contrasts carry enormous functional load: pin/bin, time/dime, coat/goat, pack/back, two/do.

For speakers of languages where these contrasts work differently—Arabic, many Indian languages, Mandarin Chinese—the English system can be disorienting. In Mandarin, for instance, the contrast is between aspirated and unaspirated voiceless stops (/pʰ/ vs. /p/), not between voiced and voiceless (/b/ vs. /p/) as in English.

Catford's (1987) ranking places stop consonant voicing contrasts consistently in the high functional load category. Getting these wrong doesn't just add accent—it changes the word you're saying.

How to fix it:

• The tissue test. Hold a tissue in front of your mouth. For /p/, /t/, /k/ at the start of a word, the tissue should flutter from the puff of air. For /b/, /d/, /g/, it shouldn't.
• Feel your throat. Place your hand on your throat. Voiced sounds (/b/, /d/, /g/) vibrate; voiceless ones (/p/, /t/, /k/) don't.
• Practice in initial position first. Start-of-word contrasts carry more functional load than mid-word or end-of-word contrasts.

5. The /θ/ and /ð/ Confusion — Wait, This One Doesn't Matter Much

Surprised? The English "th" sounds (/θ/ as in "think" and /ð/ as in "this") are some of the most-drilled sounds in ESL classrooms. But research consistently shows they have low functional load.

Brown (1991) ranks the /θ/–/t/ contrast and /ð/–/d/ contrast among the lowest functional load pairs in English. There simply aren't many common word pairs distinguished only by "th" vs. "t" or "d" (think/tink? Not a word.).

Munro and Derwing (2006) confirmed this: errors on low functional load contrasts like "th" substitution had minimal impact on comprehensibility. Listeners barely noticed.

So what actually matters in position 5? Let's talk about what you should practice instead:

The Real #5: Fricative Contrasts (/s/ vs. /z/, /f/ vs. /v/, /ʃ/ vs. /ʒ/)

These voiced/voiceless fricative pairs distinguish many common words: seal/zeal, face/phase, fine/vine, sip/zip, few/view. The /s/–/z/ contrast in particular has high functional load—it's also the difference between singular and plural (bus/buzz, price/prize) and present/past tense markers.

Many learners from languages without voiced fricatives (Mandarin, Cantonese, Korean, Arabic varieties) substitute voiceless versions, which can obscure grammatical information as well as word meaning.

How to fix it:

• Buzzing practice. Start by sustaining a long "ssssss" then slowly turn on your voice to get "zzzzz." Feel the vibration.
• Grammatical awareness. English plural -s is pronounced /z/ after vowels and voiced consonants (dogs, trees, pens). Getting this wrong sounds unnatural and can cause confusion.
• Word-final practice. Many languages don't have voiced consonants at the end of words. Practice save, five, buzz, phase with the voicing sustained all the way through.

6. Nasal Confusions (/n/ vs. /ŋ/ and /m/ vs. /n/)

The distinction between /n/ (as in "sin") and /ŋ/ (as in "sing") is subtle but carries real functional load in English: sin/sing, thin/thing, ban/bang, ran/rang, win/wing.

For speakers of many languages—including several Chinese dialects, Portuguese, and some Southeast Asian languages—the /n/ and /ŋ/ distinction may not exist or may work differently in their L1.

Additionally, in many languages, confusing /m/ and /n/ is common when they occur before other consonants or at word boundaries. While /m/–/n/ has moderate functional load in English (some/sun, name/lane in rapid speech), it can be more significant in other languages.

Research by Kang and Moran (2014), published in Language Assessment Quarterly, found that certain nasal errors contributed meaningfully to lower proficiency ratings in L2 English speaking assessments.

How to fix it:

• Tongue position check. For /n/, your tongue tip is on the ridge behind your teeth. For /ŋ/, the back of your tongue hits the soft palate (same position as /k/ or /g/, but with air through the nose).
• The "sing" test. Say "sing" and hold the final sound. Now try to say "sin" and hold the final sound. Notice where your tongue is in each case.
• Practice -ing endings. Many learners say "goin" instead of "going." The /ŋ/ sound is not /n/ + /g/—it's its own sound.

7. Tone and Pitch Patterns (For Tonal Languages)

If you're learning Mandarin, Cantonese, Thai, Vietnamese, or any tonal language, this is your biggest intelligibility challenge—full stop.

Research on Mandarin Chinese shows that tone carries as much informational load as all the vowels combined. A study published in Language by Surendran and Niyogi (2006) calculated the functional load of Mandarin tones and found it to be roughly equivalent to the entire vowel system. Getting tones wrong in Mandarin doesn't just add accent—it can make you literally incomprehensible.

The four Mandarin tones turn ma into four completely different words: mother (mā), hemp (má), horse (mǎ), and scold (mà). And this pattern extends across the entire vocabulary.

Thai has five tones. Vietnamese has six. In each case, tonal errors create word-level confusion that listeners can't easily resolve from context.

How to fix it:

• Learn tones from day one. Don't treat tones as an add-on. Every vocabulary word must be learned with its tone.
• Exaggerate at first. New tone learners consistently under-produce the contrasts. Exaggerate the pitch contours until they become natural.
• Sing-song practice. Practice tones in isolation, then in two-syllable combinations (tone pairs), then in full sentences. Tone sandhi (tone changes in context) is real—you need to practice in phrases, not just single words.
• Use AI conversation practice. Speaking with an AI on speaky.space lets you practice tones in real conversation without the pressure of a human listener reacting to every error.

The Priority Principle: Not All Errors Are Equal

The central insight from functional load research is simple but powerful: prioritize the errors that actually cause misunderstanding.

Here's a rough priority framework based on the research:

Highest Priority (Fix These First)

1. Word stress placement (in stress-timed languages)
2. Key vowel length/quality contrasts
3. High functional load consonant contrasts (/l/–/r/, stop voicing)
4. Tone accuracy (in tonal languages)

Medium Priority

5. Fricative voicing (/s/–/z/, /f/–/v/)
6. Nasal place distinctions (/n/–/ŋ/)

Lower Priority (Don't Stress About These)

7. "Th" sounds and other low functional load contrasts
8. Precise vowel quality in unstressed syllables
9. Accent features that don't change word meaning

This doesn't mean you should never work on lower-priority items. It means that if you have 15 minutes a day for pronunciation, spend most of it on what matters most.

How to Practice Effectively

Research-backed pronunciation practice follows a clear pattern:

1. Awareness. Can you hear the difference? Before you can produce a contrast, you need to perceive it. Listen to minimal pairs and try to identify which is which.

2. Controlled production. Practice the contrast in isolated words and short phrases. Slow down. Exaggerate.

3. Guided practice. Use the sounds in structured sentences and dialogues. Conversation rehearsal is ideal here.

4. Free practice. Use the sounds in real conversation. This is where tools like speaky.space shine—you get real-time speaking practice in an environment where you can focus on pronunciation without social anxiety.

5. Feedback loop. Record yourself, compare, adjust. Progress in pronunciation is gradual, but it's also measurable.

The Bottom Line

You don't need perfect pronunciation. You need strategic pronunciation. The research is clear: a handful of high-impact sound contrasts account for most communication breakdowns, while many of the sounds learners obsess over (looking at you, "th") barely affect understanding.

Focus on the sounds that carry meaning. Practice them in context, not just in isolation. And remember: the goal isn't sounding like a native speaker—it's being understood clearly and confidently.

That's intelligibility. And it's completely within reach.

---

References

• Brown, A. (1991). Teaching English pronunciation: A book of readings. Routledge.
• Catford, J.C. (1987). Phonetics and the teaching of pronunciation. In J. Morley (Ed.), Current perspectives on pronunciation. TESOL.
• Field, J. (2005). Intelligibility and the listener: The role of lexical stress. TESOL Quarterly, 39(3), 399–423.
• Kang, O., & Moran, M. (2014). Functional loads of pronunciation features in nonnative speakers' oral assessment. Language Assessment Quarterly, 11(2), 176–196.
• Munro, M.J., & Derwing, T.M. (2006). The functional load principle in ESL pronunciation instruction: An exploratory study. System, 34(4), 520–531.
• Saito, K., et al. (2019). Pronunciation features and communicative abilities in L2 English. Studies in Second Language Acquisition, 41(4).
• Sewell, A. (2021). Functional load and the teaching-learning relationship in L2 pronunciation. Frontiers in Communication, 6, 627378.
• Surendran, D., & Niyogi, P. (2006). Quantifying the functional load of phonemic oppositions, distinctive features, and suprasegmentals. In O. Nedergaard Thomsen (Ed.), Competing Models of Linguistic Change. Benjamins.
• Zielinski, B. (2008). The listener: No longer the silent partner in reduced intelligibility. Prospect, 23(3), 7–23.