Both think and this begin with th, but they do not begin with the same sound. If you replace them with sink and zis, more bite-sized tongue exercises may not fix the problem: you also need to hear and control voicing.
English has two dental fricative sounds commonly written /θ/ and /ð/. Start with their shared tongue position, add or remove vocal-fold vibration, then move into words and a short sentence.
One tongue position, two sounds
Place the tip of your tongue gently against the underside of your upper teeth, or just between the upper and lower teeth. Let a narrow stream of air pass over it. The contact is light: biting down hard only makes the sound harder to sustain.
- /θ/ is voiceless. Air passes through, but your vocal folds do not vibrate. Try think, thank, three, bath.
- /ð/ is voiced. Keep the same tongue position and airflow, but add vocal-fold vibration. Try this, that, these, mother.
Rest two fingers lightly on your throat. Alternate sss and zzz to notice no vibration versus vibration. Keep that feeling as you change to /θ/ and /ð/. Making either sound louder is not the same as making it voiced.
Read the word, not just the spelling
Try a sentence that switches between the two sounds:
I think /θɪŋk/ this /ðɪs/ is my third /θɜːrd/ brother /ˈbrʌðər/.
The th in think and third is /θ/, while this and brother contain /ð/. At first, concentrate only on the four target words. Then put the sentence back together without exaggerating every other syllable.
Word endings matter too: breath /breθ/ is the noun, while breathe /briːð/ is the verb. A final e is a useful clue here, not a universal rule for every word. Check a dictionary's audio and transcription when a new word is uncertain.
Train a contrast instead of repeating a substitution
These pairs let you isolate what changes:
- sink — think: Tongue behind the teeth for /s/, lightly at or between the teeth for /θ/.
- tree — three: Begin /θr/ with friction, not the stop at the start of /tr/. Build it slowly as /θ/ plus /riː/.
- day — they: Keep vibration going for /ð/ instead of replacing it with a plosive.
- breeze — breathe: Compare the final tongue positions for /z/ and /ð/ as well as the vowels.
Listen before copying. Choose which word you heard, check your answer, and only then produce the pair. If you cannot hear the distinction reliably yet, ten fast repetitions can merely reinforce the same guess. The aim is useful contrast, not erasing every feature of your accent.
A one-minute recording drill
- 15 seconds: Alternate /θ/ and /ð/ four times with your fingers on your throat. Keep the airflow gentle.
- 20 seconds: Say think — this — three — those. Notice the switch from voiceless to voiced rather than applying one setting to every word.
- 25 seconds: Record I think this is my third brother twice. Use a slow, accurate version first, then a connected version. Review only the four target words.
If think this keeps tripping you up, put a small boundary between the words, then shorten it as the contrasts become easier. Leave broader linking and reduction work for afterwards; our guide to Connected Speech explains how words change inside fluent phrases.
With MirrorSay, record a short sentence containing think, this, three, or those, shadow the reference, and compare the actual words you produced. Four controlled words today are more useful than rushing through a list of fifty.
If you can hear the contrast from one model but lose it with a new voice, build a High Variability Phonetic Training set with several speakers and words.
After learning /θ/ and /ð/, there is no need to treat every accent feature as an error. Use pronunciation priorities to choose what most affects understanding next.
Further reading
- BBC Learning English: Voiceless Consonants — /θ/ demonstrates the voiceless TH sound with example words.
- BBC Learning English: Voiced Consonants — /ð/ demonstrates the voiced TH sound with example words.
- BBC Learning English: Voiceless Consonants describes voiceless consonants as sounds made without vocal-fold vibration.
*MirrorSay is a free speaking-practice app: say a sentence, hear a more natural version, then shadow it and compare. No sign-up, and speech is processed on-device whenever possible.*