What it does
Phonology describes sounds as bundles of distinctive features: [±voice], [±sonorant], [±continuant] and so on. Sounds that pattern together in a rule (all voiceless stops, all high vowels) usually share features that no other sound in the language has, and are called a natural class. This tool does the bookkeeping that phonology exercises and papers need:
- a feature matrix for any list of IPA segments, with +, − and 0 (not applicable);
- the features a chosen set shares;
- the smallest feature sets that pick out exactly that set within your inventory, or a clear “not a natural class” with the extra segments that get in;
- find by features: type [+voice −sonorant] and see which segments match;
- export as a LaTeX tabular or CSV.
It works well with the IPA keyboard and the vowel formant chart.
How to use it
- Type or paste the language’s segments in the first box, separated by spaces. Affricates need a tie bar (t͡ʃ) or a ligature (ʧ), because “tʃ” is read as two segments. The buttons fill in an example inventory, every consonant on the IPA chart, or every vowel.
- Diacritics after a symbol change its features: ʰ [+spread glottis], ̤ breathy [+spread glottis], ̰ creaky and ʼ ejective [+constricted glottis], ̥ [−voice], ̬ [+voice], ̃ [+nasal], ʷ [+labial, +round], ʲ [+high, +front], ˠ velarised, ˤ pharyngealised, ̩ [+syllabic], ̯ [−syllabic], ̪ dental, ː [+long]. Other diacritics are kept in the symbol but not turned into features; the page says which.
- Tap column heads in the matrix, or type the segments in Chosen segments, to test a set as a natural class.
- Tick Hide features that are the same for every segment to shorten the matrix.
- Download the matrix as CSV or as a LaTeX tabular (features down, segments across). Compile the .tex with XeLaTeX or LuaLaTeX and a font with IPA symbols.
How the values are set
Every value comes from the symbol’s IPA description by a rule, not from a table: obstruents (plosives, fricatives, affricates, clicks, implosives, ejectives) are [−sonorant]; plosives, nasals and affricates are [−continuant]; affricates alone are [+delayed release]; central approximants and glottals are [−consonantal]; labials, coronals (dental to retroflex, plus alveolo-palatal) and dorsals (palatal to uvular, and every vowel) get their place features, and features such as [anterior], [distributed], [round], [high] and [back] are 0 where their place does not apply. Labiodental and sibilant fricatives and affricates are [+strident]. Close and close-mid vowels are [+tense]; the others are [−tense].
Frameworks differ. Some make laterals [−continuant], [h] and [ʔ] [+sonorant], palatals [+coronal] as well as dorsal, or use [ATR] instead of [tense]. Check the values against the framework your course or paper uses.
Worked example
The example inventory opens with p t k chosen. They share 14 features, and the tool reports p t k = [−delayed release, −voice]: the voiceless segments that are stops. Choose m n ŋ instead and the answer is the single feature [+nasal]. Choose s z ʃ ʒ t͡ʃ d͡ʒ and two sets of two features each work: [−labial, +strident] or [+coronal, +strident]. Choose p z and the page says it is not a natural class: every segment with all their shared features also includes t b d f θ s ʃ v ð ʒ t͡ʃ d͡ʒ, the other obstruents made in front of the velum. In Find segments by features, [+voice −sonorant] lists b d ɡ v ð z ʒ d͡ʒ.
FAQ
Why is the minimal set not the one in my textbook? The tool finds the smallest sets within your list, and lists every set of that size. Textbooks often add redundant features to make the class easier to read; [−sonorant, −continuant, −voice] for p t k is also correct, just longer.
Can I change a value? Not in the page; download the CSV and edit it, or adjust the LaTeX.
Is anything stored? No. Nothing is saved or sent; the page works offline once loaded.
Where the numbers come from
- Feature values are derived on this page by rule from each symbol's standard IPA description (place, manner and voicing; vowel height, backness and rounding), in the tradition of Chomsky and Halle (1968, The Sound Pattern of English) with place features as in modern textbooks (e.g. Hayes 2009, Introductory Phonology). No published feature chart is copied; frameworks differ on details.
Standard scientific values, such as constants, atomic weights and melting points, are facts found in every textbook; we name where ours were checked. If you are an author or publisher and want a value, a credit or a table changed or removed, tell us and we will act quickly.
Last updated 2026-11-09.