← All modules

Estimate how many syllables an English word has, by counting vowel groups and applying a list of corrective patterns.

Version

1.0.3 zef:coke

Depends

none beyond the core

License

Artistic-2.0

Its own test suite

1 file, green

Checked

2026-09-15 against Raku++ 3.28.0 and Rakudo 2026.08

Where it lives

raku.land · source

Install it #

$ rakupp install Lingua::EN::Syllable

zef install Lingua::EN::Syllable writes the same store; either installer leaves the module usable by both engines.

What it is for #

Every readability score — Flesch, Flesch–Kincaid, Gunning fog, SMOG — is words per sentence and syllables per word. The second half needs a syllable count, and a dictionary lookup is both too slow and too incomplete for arbitrary text.

This distribution is the heuristic version: count the vowel groups, then apply a list of corrections for the patterns English uses to break the rule.

Counting syllables #

use Lingua::EN::Syllable;

for <cat table syllable beautiful university Mississippi
     strength onomatopoeia> -> $w {
    say sprintf('%-16s => %d', $w, syllable($w));
}

cat              => 1
table            => 2
syllable         => 3
beautiful        => 4
university       => 5
Mississippi      => 4
strength         => 1
onomatopoeia     => 7

Most of those are right. table is two, beautiful four, university five, Mississippi four, strength one.

Case and punctuation #

use Lingua::EN::Syllable;

for <banana Banana BANANA> -> $w {
    say sprintf('%-10s => %d', $w, syllable($w));
}
say '';
for "don't", 'hello-world', 'co-operate' -> $w {
    say sprintf('%-14s => %d', $w.raku, syllable($w));
}

banana     => 3
Banana     => 3
BANANA     => 3

"don't"        => 1
"hello-world"  => 3
"co-operate"   => 4

Case is folded correctly. Hyphenated input is counted as a whole, which is right for hello-world and means the routine is not purely per-word — feed it tokens, not phrases.

The one thing to know #

The function has a floor of 1 and never returns 0, including for input containing no letters at all.

use Lingua::EN::Syllable;

for '', '   ', '123', '---', '42.7' -> $w {
    say sprintf('%-10s => %d', $w.raku, syllable($w));
}
say '';
say 'there is no return value that means "I could not analyse this"';

""         => 1
"   "      => 1
"123"      => 1
"---"      => 1
"42.7"     => 1

there is no return value that means "I could not analyse this"

The most common use of a syllable counter is a readability score computed over a tokenised document — and a tokeniser that leaves numbers, punctuation fragments or empty strings in the stream will silently inflate the total by one per junk token. That bias lands directly on the metric, and the metric is the only thing you were computing.

Filter the token stream yourself: skip anything without a letter in it.

Where the two engines differ #

Nowhere. Every word, every edge case and every case variant produced identical output on Raku++ and Rakudo, with no exception raised for any input tested.

The accuracy is what you would expect from a vowel-group heuristic, and the errors are not symmetric. Two worth knowing:

use Lingua::EN::Syllable;

for 'queue'   => 1,
    'rhythm'  => 2,
    'onomatopoeia' => 6,
    'cat'     => 1,
    'table'   => 2 -> $p {
    my $got = syllable($p.key);
    say sprintf('%-14s counted %d, actually %d  %s',
        $p.key, $got, $p.value, $got == $p.value ?? 'ok' !! 'off');
}

queue          counted 2, actually 1  off
rhythm         counted 1, actually 2  off
onomatopoeia   counted 7, actually 6  off
cat            counted 1, actually 1  ok
table          counted 2, actually 2  ok

queue has one syllable and counts as two — four vowels in a row defeat the grouping. rhythm has two and counts as one, because it has no written vowel at all. Over a document those errors partly cancel, which is why the heuristic is good enough for a readability score and not good enough for anything that needs a per-word answer.