From print to speech.

Letters appear in ⟨angle brackets⟩, contrastive sounds in /slashes/, and a particular pronunciation in [brackets]. IPA is a sound alphabet: /j/ is the glide at the start of yes, never automatically the English sound in jet. ˈ marks the start of a stressed syllable; ˌ marks secondary stress; . is a syllable boundary; ː means long; ʰ means a puff of aspiration; ʲ means a simultaneous raised front tongue; ʷ means rounded lips; ̃ above a vowel means nasal airflow. An apostrophe in a language’s spelling can represent a glottal stop, an ejective, a boundary, or something else: consult its chapter. Broad transcriptions deliberately omit predictable detail. They are not acoustic measurements.

The alveolar ridge is the bump behind the upper front teeth. The hard palate is the firm roof farther back; the soft palate is the flexible rear roof; the uvula hangs behind it. Bilabial sounds use both lips; labiodental sounds use bottom lip and upper teeth; dental sounds use the teeth; alveolar sounds use the ridge; postalveolar sounds use the area just behind it; palatal sounds use the middle tongue against the hard palate; velar sounds use the tongue back against the soft palate. Retroflex means the tongue tip is drawn back; it does not require extreme curling. Pharyngeal sounds narrow the throat below the tongue root. Start gently; throat pain is a reason to stop and reset, not evidence of success.

Stops briefly seal airflow, then release it: /p t k/. Fricatives maintain a narrow noisy opening: /f s x/. Affricates begin with closure and release into friction at roughly the same place: /tʃ/. Nasals close the mouth path but open the nose path: /m n ŋ/. Approximants narrow without sustained friction: /w j/. A tap /ɾ/ touches once; a trill /r/ lets airflow produce repeated contacts. A lateral /l/ lets air around the sides of the tongue. Put two fingertips lightly on your throat and alternate /ssss/ and /zzzz/: vibration helps distinguish voicing. Do not press on the throat or force a trill.

Hold a light strip of tissue in front of your lips. Compare the p in English pin and spin: the first usually produces a stronger puff. Practise /pa pʰa ba/ as three separate targets. In Mandarin, Korean, Hindi and several other languages, English spelling b versus p gives a misleading model. Aspiration is a delay before the following vowel starts voicing; a voiced consonant uses vocal-fold vibration. A breathy voiced consonant combines voicing with extra airflow. An ejective compresses air above a closed glottis; it is neither an ordinary aspirated stop nor simply a louder consonant.

For /i/, bring the tongue high and forward without closing the air passage. Open gradually through /e ɛ a/. Move the tongue back for /ɑ ɔ o u/. Lip rounding is a separate choice: hold the tongue of /i/ while rounding to reach /y/; round /e/ for /ø/ and /ɛ/ for /œ/. For /ɯ/, use the tongue position of /u/ with unrounded lips. Central /ə/ is a relaxed mid vowel, but it is not the correct substitute for every unfamiliar vowel. Keep a pure vowel steady; an English-style glide at the end changes /e/ into something closer to /eɪ/. Use a mirror for lips and jaw; the tongue position still needs listening and feedback.

Length changes duration, stress makes a syllable prominent, and tone changes pitch. Practise the same vowel at two durations while keeping pitch steady, then at two pitches while keeping duration steady. A mora is a timing unit: a Japanese long vowel uses two moras even within one syllable. A long consonant holds its closure or friction; do not insert a vowel between two written consonants. Tone numbers use 1 for low and 5 for high relative to your own comfortable speaking range. 35 rises; 51 falls. These are teaching targets, not musical notes. Questions and emotion can shift the whole pitch range without erasing lexical contrasts.

First identify the language variety and script conventions. Mark letter groups before single letters. Divide the word into syllables; identify stress, vowel length and tone where relevant. Apply the language’s local rules (such as a consonant changing before a front vowel). Then apply word-edge and connected-speech rules. Say it slowly without adding extra vowels, and finally at a conversational pace. If spelling leaves stress, vowels or tone unspecified, look up that information and write it beside the word. Guessing cannot recover information that the writing system does not encode.

Choose one contrast. Read its mouth instructions; say alternating examples five times; then mix their order. Next listen to unfamiliar examples without looking and identify the sound. Recognition can lag behind production or the reverse. Record a short phrase on your own device and compare duration, stress, pitch and consonant release one at a time. Work from sound → syllable → word → short phrase → normal sentence. Keep a short error log: intended sound, actual difference, physical adjustment, retest. Five careful minutes on one contrast is more useful than repeatedly reading an entire page with the same error.

A description can explain how to form a sound; it cannot certify that your own production matches. Device speech, where offered, is synthetic and may use the wrong regional accent or fail to preserve a contrast. It is supplementary listening, not a native-speaker recording or an assessment. Historical pronunciations are reconstructions or named reading traditions. Living languages also vary: choose a consistent target, learn recognition of alternatives, and do not treat a different dialect as a mistake. The chapters teach a reading procedure and explicitly identify spelling ambiguities; no text-only resource can guarantee native-like pronunciation.