Spoken English and pronunciation

In short

Spoken English is the skill of producing English that a listener anywhere understands at first hearing: correct speech sounds, correct word and sentence stress, natural rhythm and intonation, and fluent delivery without strain. Pronunciation is judged on intelligibility, not on sounding British or American - the working target in India is a clear, globally understood General Indian English. It is examined in the first-year university course (AICTE model curriculum "Communicative English" / UGC CBCS AECC English) through IPA transcription, stress marking and the English Language Lab practical with viva. In placements it decides the HR round, the telephonic round, group discussion and client-facing shortlists, and it is machine-scored by tests such as Versant (Pearson) and SVAR/AMCAT at IT services, BPO, banking and airline recruiters.

Explained

Speech sounds and the IPA

• Phonetics studies how sounds are produced and heard; phonology studies how a language organises them into meaningful units.
• Phoneme: the smallest sound unit that changes meaning (/p/ vs /b/ in *pin/bin*). An allophone is a variant that does not change meaning (aspirated [pʰ] in *pin*, unaspirated [p] in *spin*).
• Minimal pair: two words differing in exactly one phoneme - *van/wan*, *thin/tin*, *sip/ship*, *bad/bed*, *cot/coat*. Minimal pairs are the standard diagnostic and drilling tool.
• Standard English (RP / BBC English, the model in most Indian syllabi) has 44 phonemes: 24 consonants + 20 vowels.
• 24 consonants by manner: 6 plosives (p b t d k g), 2 affricates (tʃ dʒ), 9 fricatives (f v θ ð s z ʃ ʒ h), 3 nasals (m n ŋ), 1 lateral (l), 3 approximants/semivowels (r w j).
• A consonant is described by place + manner + voicing: /t/ = voiceless alveolar plosive; /ʒ/ = voiced palato-alveolar fricative; /ŋ/ = voiced velar nasal.
• 20 vowels = 12 pure vowels (monophthongs) iː ɪ e æ ɑː ɒ ɔː ʊ uː ʌ ɜː ə + 8 diphthongs eɪ aɪ ɔɪ əʊ aʊ ɪə eə ʊə. Five triphthongs (aɪə in *fire*, aʊə in *hour*) are listed in some syllabi.
• A vowel is described by tongue height, tongue position (front/central/back), lip shape and length: /iː/ = long high front unrounded; /ɒ/ = short low back rounded; /ə/ = mid central (the schwa).
• Write phonemes between slashes /kæt/; square brackets [ ] are used for actual phonetic detail. Reference: Daniel Jones, *Cambridge English Pronouncing Dictionary* (18th edition is the one usually cited); online Cambridge/Oxford entries give IPA plus audio.
• Spelling is not pronunciation. English has 26 letters for 44 sounds, so count vowel sounds, not vowel letters - *through* has 1 vowel sound, *beautiful* has 3 (not 5).

Syllable, word stress and weak forms

• A syllable = optional onset + obligatory nucleus (a vowel sound) + optional coda. *strength* = CCCVCC, one syllable; *de-vel-op-ment* = four.
• Stress marks go before the syllable: primary ˈ, secondary ˌ - /ˌedʒʊˈkeɪʃn/.
• Two-syllable words: most nouns and adjectives take first-syllable stress (ˈtable, ˈhappy); most verbs take second (deˈcide, beˈgin).
• Stress-shifting pairs (noun/adjective vs verb): ˈrecord-reˈcord, ˈpresent-preˈsent, ˈobject-obˈject, ˈincrease-inˈcrease, ˈpermit-perˈmit, ˈconduct-conˈduct, ˈexport-exˈport.
• Neutral suffixes do not move stress: -ing, -ed, -ly, -ness, -ful, -less, -er, -able (ˈcomfort → ˈcomfortable).
• Stress-attracting suffixes take the stress themselves: -ee (employˈee), -eer (engiˈneer), -ese (Japaˈnese), -ette (cigaˈrette), -oon (balˈloon), -ique (uˈnique).
• Pre-suffix stress - stress falls on the syllable immediately before: -ity (aˈbility), -ion (ˌappliˈcation), -ic (ˌecoˈnomic), -ical (ecoˈnomical), -ious (inˈdustrious), -graphy (phoˈtography), -logy (biˈology), -meter (therˈmometer).
• Compound nouns stress the first element (ˈblackboard, ˈgreenhouse, ˈbus stop); an adjective + noun phrase stresses the second (a green ˈhouse, a black ˈboard).
• Weak forms: unstressed function words reduce to schwa - *to* /tə/, *of* /əv/, *and* /ən/, *for* /fə/, *was* /wəz/, *can* /kən/, *them* /ðəm/. Pronouncing every function word in full is the single biggest cause of "flat, word-by-word" English.

Rhythm, connected speech and intonation

• English is stress-timed: stressed syllables fall at roughly equal intervals and unstressed ones are compressed. Most Indian languages are syllable-timed (every syllable gets near-equal length), which is why Indian English can sound even-paced to a foreign ear.
• Sentence stress falls on content words (nouns, main verbs, adjectives, adverbs, question words); function words (articles, prepositions, auxiliaries, pronouns) stay weak.
• Connected speech processes: linking (*an apple* → /ənæpl/), elision of a sound (*next day* → /neks deɪ/, *sandwich* → /ˈsænwɪdʒ/), assimilation (*ten boys* → /tem bɔɪz/), contractions (*I'll*, *we've*, *doesn't*).
• Intonation is the pitch tune over a group of words. Falling tune: statements, commands, wh-questions (*Where do you ˈwork↘*). Rising tune: yes/no questions, polite requests, items before the last in a list, checking (*Did you ˈsend it↗*). Fall-rise: doubt, contrast, polite disagreement.
• The tonic (nucleus) - the most prominent word - carries the meaning. *I didn't say he stole it* means six different things depending on which word is made prominent.
• Voice modulation = pitch + pace + pause + volume + stress. Pause at thought groups (end of a phrase), not in the middle of one; a planned 0.5-1 s pause replaces a filler.

Mother-tongue influence (MTI): the Indian error map

MTI is the carry-over of first-language sounds and rhythm. It is not a defect to be erased; what must go are the features that break intelligibility or change a word.

• Epenthetic (inserted) vowels in clusters: *school* → "iskool", *speak* → "ispeak", *film* → "filum", *Monday* → "Monday-u". Practise the cluster as one release: /sk/, /sp/, /lm/, and stop the final vowel.
• Spelling pronunciation of silent letters: debt, doubt, subtle, comb (silent b); know, knife (k); write, sword, answer (w); psychology, pneumonia, receipt (p); hour, honest, heir (h); calm, salmon, colonel (l); listen, castle (t); Wednesday, handsome (d); sign, foreign (g); autumn, column (n); island, aisle (s).
• Commonly mispronounced in India (RP): pronunciation /prəˌnʌnsiˈeɪʃn/ not "pro-noun-", mischievous /ˈmɪstʃɪvəs/ (3 syllables), Wednesday /ˈwenzdeɪ/ (2), comfortable /ˈkʌmftəbl/ (3), vegetable /ˈvedʒtəbl/ (3), February /ˈfebruəri/, determine /dɪˈtɜːmɪn/, colleague /ˈkɒliːɡ/, hotel /həʊˈtel/, police /pəˈliːs/, guarantee /ˌɡærənˈtiː/, receipt /rɪˈsiːt/, resume (CV) /ˈrezjʊmeɪ/, suite /swiːt/, queue /kjuː/, entrepreneur /ˌɒntrəprəˈnɜː/.
• Grammatical Indianisms that stand out on a client call: *discuss about*, *revert back*, *cousin brother*, *out of station*, *do the needful*, *prepone*, *myself Rahul*, *doubt* for *question*, *passing out* for *graduating*. Acceptable within India; replace for global audiences.

Fluency, clarity and what recruiters actually score

• The four things scored, separately: fluency (flow, pausing, fillers), pronunciation (sounds, stress, intonation), grammar and accuracy in speech, vocabulary/range.
• Fluency is not speed. A comfortable public speaking rate is roughly 130-150 words per minute; above ~180 wpm clarity collapses, below ~110 wpm sounds hesitant.
• Fillers - *um, uh, like, basically, actually, I mean, you know, so yeah* - should stay under about 2% of words. Replace with a silent pause.
• Articulation: open the jaw, finish final consonants (*projects* not *projec*), and keep volume steady to the end of the sentence (Indian speakers often fade on the last three words).
• Register: *Could you please send it?* not *Send it*; *I'm afraid I don't have that data* not *That is not possible*.
• Automated screening: Versant/Pearson tests (read-aloud, repeat, sentence build, open response) report an overall score plus fluency and pronunciation subscores; SVAR is used inside AMCAT. Exact scales, cut-offs and subscore names differ by test version and employer, so confirm the spec for the drive you are sitting.
• Benchmarks you may be asked for: IELTS Speaking is banded 0-9 on fluency and coherence, lexical resource, grammatical range and accuracy, and pronunciation; TOEFL Speaking 0-30; CEFR runs A1-C2, with B2 a common client-facing expectation. Aviation is the one legally fixed case - DGCA requires ICAO English Language Proficiency Level 4 (Operational) for flight crew and ATC.

Building it: a practice system

1. Diagnose: record 2 minutes of free speech plus a 25-item minimal-pair list; get a peer to transcribe. Count substituted sounds, stress errors, fillers, wpm.
2. Drill sounds 10 minutes a day on your own 4-5 error sounds using minimal pairs and a mirror (mouth position matters more than repetition).
3. Shadowing: play 30 seconds of a news clip or TED talk, then speak along with it, copying rhythm and intonation rather than words.
4. Read aloud 15 minutes daily, marking stressed words and pause slashes on the page first.
5. Look up every new word's IPA and stress before using it; never guess from spelling.
6. JAM (Just A Minute) and extempore practice: one minute, no filler, recorded; review your own recording - self-listening is the fastest corrective.
7. Resources commonly used in Indian colleges: the English Language Lab software (practical + viva), EFLU/British Council pronunciation courses, NPTEL/SWAYAM soft-skills courses, Cambridge/Oxford online dictionaries for audio.
8. Realistic timeline: individual sounds improve in 3-6 weeks of daily drilling; rhythm and intonation take months. Claiming an "accent change" in a week is not credible.

Key points

• Intelligibility - the real goal; being understood first time beats imitating any native accent.
• 44 phonemes - 24 consonants + 20 vowels (12 monophthongs + 8 diphthongs) in RP.
• Phoneme vs allophone - phoneme changes meaning, allophone is a harmless variant.
• Minimal pair - two words differing in one sound; the standard diagnostic and drill.
• Schwa /ə/ - the commonest English vowel; lives in unstressed syllables and weak forms.
• Syllable count = number of vowel sounds, not vowel letters.
• Stress mark ˈ goes before the stressed syllable; ˌ marks secondary stress.
• Noun-verb stress shift - ˈrecord (n) vs reˈcord (v); same spelling, different stress.
• Suffix rules - -ee/-eer/-ese/-ette take stress; -ity/-ion/-ic/-ical/-logy/-graphy put it on the syllable before; -ing/-ed/-ly/-ness leave it alone.
• Compound noun stresses the first word (ˈgreenhouse); adjective+noun stresses the second (green ˈhouse).
• Stress-timed vs syllable-timed - English compresses unstressed syllables; Indian languages do not.
• Connected speech - linking, elision, assimilation and contractions make speech sound natural.
• Intonation - fall for statements/wh-questions/commands, rise for yes-no questions and politeness, fall-rise for doubt or contrast.
• MTI - mother-tongue influence; fix only what changes a word or blocks understanding.
• Epenthesis - inserted vowels in clusters ("iskool", "filum"); drill the cluster, not the vowel.
• Speaking rate - about 130-150 wpm; fillers under roughly 2% of words.
• -ed and -s endings have three pronunciations each, set by the preceding sound.
• ICAO Level 4 - the only statutory spoken-English requirement in India (DGCA, flight crew and ATC).

Formulas and examples

Example 1 - diagnosing a JAM recording. A student speaks for 60 s, producing 112 words, 9 fillers, and 14 s of total pause.

• Speaking rate = 112 ÷ 1.0 min = 112 wpm (below the 130-150 band).
• Phonation time = 60 − 14 = 46 s = 0.7667 min → articulation rate = 112 ÷ 0.7667 = 146 wpm.
• Pause ratio = 14 ÷ 60 × 100 = 23.3%; filler density = 9 ÷ 112 × 100 = 8.0%.
• Reading: when he is talking he is fast enough (146 wpm); the low overall rate comes from pausing and filling. Fix = fewer, planned pauses, not faster speech. At 2% the allowance is 112 × 0.02 = 2.24, i.e. at most 2 fillers.

Example 2 - fitting a 2-minute answer. A prepared answer runs 320 words; the slot is 2 minutes.

• Deliverable at 140 wpm = 140 × 2 = 280 words → the script is 40 words too long.
• Time needed for 320 words at 140 wpm = 320 ÷ 140 = 2.286 min = 2 min 17 s.
• Speeding up to fit would need 320 ÷ 2 = 160 wpm - above the comfortable 130-150 band. Cut 40 words instead.

Example 3 - syllables, stress and suffix shift.

Same root, three different stress positions across four words - proof that stress is governed by the suffix, not the spelling.

Example 4 - the -ed and -s endings. Rule: /ɪd/ after /t/ or /d/; /t/ after any other voiceless sound; /d/ after any other voiced sound. For -s: /ɪz/ after sibilants (s, z, ʃ, ʒ, tʃ, dʒ); /s/ after other voiceless; /z/ after other voiced.

• stopped /t/ · washed /t/ · laughed /t/ · begged /d/ · played /d/ · called /d/ · wanted /ɪd/ · decided /ɪd/
• cats /s/ · books /s/ · dogs /z/ · pens /z/ · boxes /ɪz/ · watches /ɪz/

Count check: of the 8 verbs, 3 take /t/, 3 take /d/, 2 take /ɪd/.

Example 5 - minimal-pair diagnostic and a call-round intelligibility check. A student records 25 minimal pairs; a peer identifies 18 correctly.

• Accuracy = 18 ÷ 25 × 100 = 72%; 7 errors: /v-w/ 3, /θ-t/ 2, /ʃ-s/ 1, /æ-e/ 1.
• 20 minutes of daily drill allocated by error share: /v-w/ 20 × 3/7 ≈ 8.6 min, /θ-t/ ≈ 5.7 min, the other two ≈ 2.9 min each.
• In a mock telephonic round the same student speaks 60 keywords and the listener transcribes 51: intelligibility = 51 ÷ 60 × 100 = 85%, i.e. roughly 1 word in 7 lost. Voice-process employers typically want the high 90s, but published cut-offs vary by test and role.

Either model is acceptable; mixing them inside one sentence is what sounds odd.

Where it is used

• IT services and consulting: client calls, daily stand-ups and stakeholder demos at TCS, Infosys, Wipro, HCLTech, Cognizant, Accenture and Capgemini; spoken English is screened at the HR/GD stage and often again before client allocation.
• BPO, voice and customer support: Concentrix, Teleperformance, WNS, Genpact, Infosys BPM, Amazon and bank call centres use automated spoken tests (Versant/SVAR type) as a first filter; neutral, high-intelligibility speech is the core job skill.
• BFSI and sales roles: relationship managers, insurance and fintech sales at HDFC, ICICI, Axis, Bajaj Finserv - pitching, objection handling and compliance scripts spoken to customers.
• Aviation, hospitality, retail and railways front-line: IndiGo, Air India, Taj, Oberoi, Vande Bharat on-board staff - announcements and safety briefings; pilots and ATC must hold ICAO English Language Proficiency Level 4 under DGCA rules.
• Teaching, training, PSU and civil-services interviews: UPSC/SSB/bank PO interviews, GATE-based PSU personal interviews, campus teaching assistantships and NET-qualified lecturing all assess spoken clarity.
• Media, voice-over and ed-tech: anchors, dubbing, podcast and course narration, where IPA-level control of stress and intonation is a daily production tool.

Interview questions

• Q: What is the difference between accent and pronunciation? A: Pronunciation is whether a word's sounds and stress are correct; accent is the regional or social colouring of speech. Wrong pronunciation changes a word; an accent need not.
• Q: How many sounds does English have, and how are they grouped? A: 44 phonemes in RP - 24 consonants and 20 vowels, the vowels being 12 monophthongs and 8 diphthongs.
• Q: What is a minimal pair and why is it used in training? A: Two words differing in exactly one phoneme, such as *van/wan*. They isolate a single sound so a learner can hear and practise just that contrast.
• Q: Where is the stress in *record*? A: On the first syllable as a noun (ˈrecord) and the second as a verb (reˈcord) - a stress-shifting pair.
• Q: What is the schwa and why does it matter? A: /ə/, the mid central vowel of unstressed syllables and weak forms. Without it, function words stay full-length and English loses its rhythm.
• Q: Why does Indian English sound different in rhythm from British English? A: English is stress-timed and compresses unstressed syllables; most Indian languages are syllable-timed and give each syllable similar length.
• Q: How is the *-ed* ending pronounced? A: /ɪd/ after /t/ or /d/ (*wanted*), /t/ after other voiceless sounds (*stopped*), /d/ after other voiced sounds (*played*).
• Q: Do you need a British or American accent for a client-facing role? A: No - employers ask for intelligibility and consistency. Correct sounds, stress and pace matter; imitating a native accent badly is worse than clear Indian English.
• Q: You have MTI. What have you done about it? A: Name the specific sounds, how you found them (recording plus peer transcription), the drill you run, and a measured change - for example minimal-pair accuracy from 72% to 94% over six weeks.
• Q: What do you do if the listener cannot understand you on a call? A: Slow to about 120 wpm, shorten sentences, stress the keyword, spell critical items using a standard alphabet, and confirm by paraphrase rather than repeating the same sentence louder.
• Q: Which intonation would you use for *Did you send the report?* A: A rising tune, because it is a yes/no question; a fall there would sound like a command or an accusation.
• Q: How many syllables are in *comfortable* and where is the stress? A: Three in normal speech, /ˈkʌmftəbl/, stressed on the first.
• Q: What is the difference between fluency and accuracy? A: Fluency is smooth, well-paced delivery with few fillers; accuracy is correct grammar, vocabulary and pronunciation. Interviews score them separately, and fluency without accuracy still loses marks.

Common mistakes

• Transcribing from spelling instead of sound - writing /ˈbʌtʌn/ for *button* (/ˈbʌtn/) or counting vowel letters to get syllables.
• Putting the stress mark after or over the syllable; the IPA mark ˈ always goes immediately before it.
• Learning a word's spelling and meaning but never its stress, then saying ˈhotel, ˈpolice or ˈengineer in the interview.
• Equating fluency with speed - racing at 180+ wpm, dropping final consonants, and losing the last words of every sentence.
• Chasing an American accent for a few weeks and ending with an inconsistent mix, while the real blockers (/v-w/, /θ/, clusters) remain untouched.
• Inserting vowels into clusters ("iskool", "filum", "Monday-u") and adding retroflex /ʈ/ /ɖ/ for alveolar /t/ /d/.
• Carrying Indianisms (*discuss about*, *revert back*, *doubt*, *myself Rahul*) into a global-client conversation.
• Practising only by reading silently or in your head; without recording and listening back, errors are never heard, so they never change.

Related topics

Self-introduction for interviews · Listening skills · Telephone and video-call etiquette · Group discussion · Presentation skills · Body language and presence