Connected Speech: Linking, Elision, and Assimilation

Ask a classroom of learners to read a sentence aloud and you will hear every word pronounced in crisp isolation. Record a native speaker producing the same sentence at conversational pace and the boundaries between words practically dissolve. Sounds glide into one another, vanish altogether, or morph to match their neighbours. These shifts — grouped under the label "connected speech" — are not slurring or shortcuts. They are a systematic, predictable layer of English pronunciation that every fluent speaker uses and every fluent listener expects to hear.

Defining Connected Speech

Connected speech is a catch-all term for the systematic adjustments that happen when words run together instead of standing alone. In careful, slow pronunciation, each word keeps its dictionary form — the so-called citation form. In running speech, the sounds at the edges of each word start reacting to the sounds next to them, producing changes that are entirely expected once the patterns click into place.

These adjustments exist for a reason. Producing every word in full, isolated form would be slow, tiring, and strange-sounding. Connected speech lets the mouth take the shortest reasonable route between sounds while still preserving the rhythm that makes English feel like English.

For learners, this layer matters for two concrete reasons. First, a lot of what feels like "native speakers talk too fast" is really "native speakers use connected speech and I haven't tuned my ear to it yet." Second, your own speech starts to sound fluent — rather than stilted and over-articulated — the moment you stop fighting these processes and let them happen.

Where These Changes Come From

Connected speech grows out of a combination of physics and rhythm. A few factors drive it.

Economy of Articulation

The tongue, lips, and jaw have to travel between positions for every sound. Some of those transitions are costly; others are easy. Connected speech trims the costly ones by letting the articulators glide rather than hit every target precisely.

The Pull of Stress Timing

English is stress-timed: stressed syllables want to land at roughly regular beats. When several unstressed syllables pile up between two stressed ones, they have to compress to keep the beat, and processes like elision and reduction do the compressing.

Tempo

The faster the speech, the less time each sound gets, so connected speech processes intensify. Even at unhurried speeds, though, these processes are already in play.

Linking: Tying Sounds Across Word Boundaries

Linking happens when the final sound of one word blends seamlessly into the opening sound of the next, with no pause or glottal break between them.

Consonant-to-Vowel Linking

When a consonant ends one word and a vowel begins the next, the consonant behaves as if it had moved to the start of the second word. This is the form of linking most listeners notice first.

Vowel-to-Vowel Linking

When two vowels bump into each other across a boundary, speakers usually slide a glide between them to smooth the transition. This overlaps with the intrusion processes described further down.

Linking /r/

In non-rhotic varieties such as most British English, a silent final "r" reappears when the next word starts with a vowel. This is the linking /r/.

Elision: Sounds That Vanish

Elision is the outright dropping of a sound that would be present if you pronounced the word in isolation. It is probably the most unnerving process for learners, because missing sounds can push a familiar word right past recognition.

Dropping /t/ and /d/

The most frequent elision involves /t/ and /d/ sandwiched between other consonants. This happens across word boundaries and inside words too.

Dropping Vowels

Unstressed vowels — especially the schwa — often get swallowed entirely in fast speech, taking a whole syllable with them.

Dropping /h/

The /h/ at the start of unstressed function words tends to disappear in the flow of a sentence. Words like "he," "him," "her," "his," "have," and "has" routinely lose that initial /h/ when they aren't stressed.

Assimilation: Sounds That Reshape Themselves

Assimilation is the shift a sound makes toward the shape of an adjacent sound. Most commonly it happens at word boundaries, where the final consonant of one word adjusts to anticipate the first consonant of the next.

Place Assimilation

The usual flavor in English involves place of articulation. The alveolar trio — /t/, /d/, /n/ — is especially eager to shift.

Voicing Assimilation

A sound can flip its voicing to match a neighbour. English uses this less than some languages, but it does turn up.

Coalescence

Coalescence is a special case in which two adjacent sounds collapse into a brand-new third sound. The classic examples involve /t/ or /d/ followed by /j/.

Intrusion: Sounds That Sneak In

Intrusion is the opposite of elision: a sound turns up even though the spelling gives no warrant for it. The reason is mechanical — speakers insert the sound to ease a tricky vowel-to-vowel transition.

Intrusive /r/

In non-rhotic accents, speakers sometimes plug an /r/ between two vowels where no "r" exists on the page at all. It's a cousin of the linking /r/, but it shows up without any historical justification.

Intrusive /w/ and /j/

Words ending in a close back vowel like /uː/ or /əʊ/ tack on a /w/ before a following vowel. Words ending in a close front vowel like /iː/ or /eɪ/ do the same thing with a /j/.

Geminates: Held Consonants Across Words

When a word finishes with the same consonant that opens the next word, speakers typically hold the consonant a little longer rather than articulating it twice. The result is a single, extended consonant known as a geminate.

Weak Forms and Vowel Reduction

Although it is not strictly a boundary process, weak-form reduction is an essential ingredient of connected speech. About 50 common English words carry two pronunciations: a strong form used under stress or in isolation, and a weak form used everywhere else.

Function words — "and," "but," "for," "from," "of," "to," "a," "the," most pronouns, and most auxiliary verbs — show up almost exclusively in weak forms during natural speech. Using the strong form of these words sounds oddly formal and throws off the rhythm of the whole sentence.

How Contractions Fit In

Contractions are the most formalised kind of reduction in English — so standard that they have their own spelling conventions. In running speech, contractions can reduce further still.

The common set includes forms such as: I'm, you're, he's, she's, it's, we're, they're, I'll, you'll, he'll, she'll, we'll, they'll, I'd, you'd, he'd, she'd, we'd, they'd, I've, you've, we've, they've, isn't, aren't, wasn't, weren't, don't, doesn't, didn't, can't, won't, shouldn't, couldn't, wouldn't.

Double contractions like "I'd've" (I would have), "she'd've" (she would have), and "they'd've" (they would have) rarely appear in edited writing but are everywhere in speech. Recognising them is another gear in your listening toolkit.

Sharpening Your Listening Ear

A lot of listening struggles are connected speech struggles in disguise — not vocabulary gaps or grammar gaps. Here are practical ways to close the gap.

Transcription Drills

Take a short audio clip and write down every word you hear. Compare what you wrote with the official transcript. Wherever your version differs, figure out which process — elision, assimilation, linking — caused the mismatch. Do this regularly and the patterns start announcing themselves.

Slow-Speed Playback

Drop a recording to 0.75x speed. The connected speech features are still there, just slow enough to pick apart. As your ear adjusts, bump the speed back up until you're hearing the same processes at normal pace.

Prioritise the Big Patterns

Instead of chasing every possible modification, drill the handful that do most of the work: consonant-to-vowel linking, /t/ and /d/ elision, weak forms, and coalescence. These four account for the lion's share of the comprehension gap.

Sounding More Natural When You Speak

Producing connected speech is less about forcing changes and more about getting out of the way and letting them happen. Start small, and build up from scripted contexts into spontaneous ones.

Chunk-Based Practice

Rehearse frequent phrases as single rhythmic chunks, not as stacks of individual words. Treat "cup of tea," "have to go," "kind of," and "a bit of a" as one unit each, and the sound changes fall out naturally.

Shadowing

Pick a native speaker recording and imitate in real time, mimicking the rhythm, pitch, and linking rather than chasing word accuracy. Shadowing builds the motor memory that carries the processes over into your own speech.

Attack the Function Words

The single biggest upgrade comes from under-pronouncing the small words. Deliberately keep articles, prepositions, and auxiliaries short and quiet while hitting the content words. Your sentences will immediately sound more native.

Connected speech is the line that divides textbook English from real-world English. Learn these patterns, practise them, and the gap between the classroom and the conversation begins to close — your ear gets sharper, your mouth gets looser, and suddenly the language is coming at you at its normal speed without feeling like a blur.