Learn

Rhythm.

In most English words of more than one syllable, one syllable carries primary stress. Across a phrase those stresses fall into a larger pattern — and the pattern is not the sum of the words.

From one beat to a pattern

Stress settles which syllable of a word carries weight. Rhythm is what happens when a whole phrase of them lines up:

the BAND will PLAY toNIGHT at SCHOOL

ta-DUM · ta-DUM · ta-DUM · ta-DUM — light syllable into strong, four times over.

DRAma STUdents PRACticed LOUDly

DUM-ta · DUM-ta · DUM-ta · DUM-ta — the same alternation, running the other way.

Neither pattern is a property of any single word in it. It emerges from how the words are placed next to each other, which is why the same word can sit in a strong position in one sentence and a weak one in the next.

Four recurring shapes

English keeps coming back to a small number of groupings. The tools on this site work with four of them, the four used in the training materials for the 2026 study:

JACK fell DOWN and BROKE his CROWN

Strong–weak, repeating. Traditionally a trochee. Also PROduce, the noun.

the MOUSE ran UP the CLOCK

Weak–strong. An iamb. Also proDUCE, the verb.

you have BRAINS in your HEAD, you have FEET in your SHOES

Weak–weak–strong. An anapest.

ALL the king’s HORses and ALL the king’s MEN

Strong–weak–weak. A dactyl.

Four is a modelling decision, not a fact about English. It is a defensible inventory — these are the groupings that recur — but other analyses divide the space differently, and the tools deliberately leave things out. There is no strong–strong category, traditionally called a spondee. And a word like banana, whose lexical pattern is weak–strong–weak and would be called amphibrachic in traditional metrics, is not given a word-sized template of its own; the phrase is parsed across it instead. Which brings us to the more interesting point.

Feet cross words; words keep their stress

Rhythmic grouping does not respect word boundaries. Take three words:

a baNAna FELL

The phrase scans as weak–weak–strong then weak–strong — an anapest followed by an iamb. The first foot ends in the middle of banana, and the second begins there.

Meanwhile banana itself has not changed. Its lexical stress is still weak–strong–weak, exactly as it would be spoken alone. Two descriptions coexist: the word keeps its own shape, and the phrase imposes a grouping that straddles it.

This is why marking beats is more than bolding syllables one word at a time. It also explains a design choice you can see in the tools: feet are fitted inside a phrase and never drawn across a phrase boundary. Move a boundary and the rhythm on either side of it changes — the reason phrasing and rhythm cannot really be taught apart.

Beats move to avoid collisions

If alternation is the preference, what happens when two strong syllables end up adjacent? Often, one gives way. Said alone, TennesSEE stresses its last syllable and AIR is stressed. Put them together and the pattern flips:

TennesSEE + AIR → TENnessee AIR

Likewise thirTEEN and MEN, spoken alone, become THIRteen MEN as a phrase. This is the rhythm rule: beats can be added, moved, or dropped to prevent a clash of adjacent strong syllables (Liberman & Prince, 1977; Selkirk, 1986).

It reaches even into words that do not exist. Given a nonsense word, speakers place its stress to keep the alternation going — Planes will COLvane pilots after a weak syllable, but The pins colVANE balloons after a strong one (Kelly & Bock, 1988). The rhythm is doing the deciding, not the lexicon.

This is also why the tools have no strong–strong foot. A clash is the thing English rhythm works to avoid, so a model built around alternation has little use for a category that encodes one.

Not a metronome

Rhythm here means the alternation of relative prominence, not evenly spaced beats. English speech does not divide into equal intervals, and the syllables between beats stretch and compress considerably. A line can feel strongly rhythmic while being, in measured time, quite irregular.

Relative is the operative word. A syllable is not strong or weak in isolation; it is stronger or weaker than its neighbours (Liberman & Prince, 1977). A word keeps its own lexical stress wherever it appears — that much does not move — but whether its stressed syllable carries a phrase-level beat depends on what surrounds it.

Monosyllables move most freely of all. Of goes unstressed in ordinary speech, yet in the trochaic line in the forests OF the night it takes a beat — a placement the metrical frame makes appropriate and isolated speech would not. That item comes from the 2026 posttest, and it is the kind of case that distinguishes reading rhythm from reciting a word list.

Where the strongest beat falls

Within a phrase, one beat is usually the strongest. In neutral prosody it tends to land on the last content word:

The band will play at SCHOOL.
Give it to the CHILDren.

In both, the accent falls on the last content word — which here happens to be the last word — and on that word’s own stressed syllable rather than its final one: CHILDren, not childREN.

The distinction between last word and last content word only shows itself when function words come afterwards:

She gave the KEYS to him.

to him comes last and takes nothing. The main accent stays back on keys, because what it seeks is the last word carrying content.

This is the nuclear stress tendency (Chomsky & Halle, 1968; Liberman, 1975; Liberman & Prince, 1977) — a tendency, and one that context overrides constantly. Ask who will play at school and the accent moves to the answer. Rhythm Reader Pro marks the nuclear beat and also lets you switch the rule off, because a model that cannot be contradicted is not much use for studying readers.

Why it matters for reading

Sensitivity to rhythm travels with reading ability. In children, rhythm sensitivity correlates with growth in fluency and comprehension (David et al., 2007; Holliman et al., 2010), and a recent meta-analysis reports substantial deficits in specific prosodic skills among readers with developmental dyslexia (Mundy & Wood, 2025). Rhythmic structure also appears to support syntactic parsing and memory for text (Kelly & Bock, 1988).

Two cautions about that literature. These associations are largely correlational, so they establish that rhythm sensitivity travels with reading skill, not that training one moves the other. And the frequently cited meta-analytic link between reading prosody and comprehension (r = 0.51; Wolters et al., 2022) was measured on production-based prosody — how expressively people read aloud — rather than on perceptual sensitivity of the kind discussed here. The two are related but not the same thing.

Two studies from this project bear on rhythm directly. The 2018 experiments asked readers to rate capitalized marking placed on the last line of poems in trochaic, iambic, and anapestic meter; marking that matched the poem’s beat was rated more helpful than marking that fought it, across all three meters. That was a preference judgment, with no training and no plain-text control — see the stress page for what that design can and cannot show. The 2026 study ran the same randomized active-control experiment in two independent samples. Participants learned all four patterns above through interleaved practice, immediate feedback, and gradually reduced marking. In both samples the prosody group then outperformed the active control on a minimally marked meter-judgment task, by a small margin (d = 0.25 and 0.34 by sample; 0.31 combined). The stress page sets out what those studies do and do not show, including that transfer to fully unmarked text was not demonstrated and no reading outcomes were measured.

Try it

Rhythm Reader fits rhythm across each phrase of text you paste in and shows the beats in your choice of four styles. Paste in a nursery rhyme and a paragraph of prose and compare: metrical verse tends to sustain a prevailing pattern more consistently, while prose shifts among patterns as the syntax and phrasing change. Verse permits substitution too, and prose can run locally regular for a stretch.

Rhythm Reader Pro names the feet, detects the prevailing meter of a passage, marks the nuclear beat, and lets you move phrase boundaries to watch the footing rearrange itself. It also reports how confident the parse is, which is often the most informative thing on the screen.