🎵ToBI (Tones and Break Indices)

ToBI — Tones and Break Indices — is not a phonetic alphabet. It is a system for transcribing intonation and prosody: the melody of speech, and how tightly words are joined together. It was created by speech scientists in the early 1990s to give prosody the kind of shared standard that the IPA gives to segments, so that prosodically labelled speech databases could be exchanged between research sites. The original design target was English, specifically Mainstream American English.¹²

The tiers

A ToBI annotation is layered over a recording as parallel tiers, time-aligned to the waveform and pitch track:¹

Two points trip people up. Tone labels like H* are phonological categories, not raw fundamental-frequency (F0) values — they describe how a pitch movement functions, not its exact hertz. And break indices are not punctuation: a break index of 4 marks a full intonational-phrase boundary as heard, which may or may not line up with where a comma would go.¹

One name, two scopes

"ToBI" refers to two different things. In its narrow sense it is the original English (MAE_ToBI) system. In its broad sense it is a framework — a set of design principles for building a prosodic transcription system for a particular language. The Ohio State guidelines are explicit that the original ToBI was developed for major varieties of English and was not intended to cover other languages, though its principles can be adapted. That adaptation has produced a whole family of language-specific systems.¹³

The lesson for anyone labelling data: you cannot take English ToBI labels and apply them to another language. You need that language's ToBI system, whose inventory of tones and break conventions may differ — K-ToBI, for instance, uses five minimal tiers and separates phonological from phonetic tone.³

Reliability and tools

ToBI was validated with an inter-transcriber reliability study — the 1992 report describes twenty transcribers making some 20,000 labelling decisions and reports high agreement, which is what let ToBI serve as a shared standard rather than one lab's private notation. In practice, ToBI tiers are annotated in Praat TextGrids (or Emu, or older xwaves-style tools), which is where prosody transcription meets the Praat workflow LinguaCommons audio work uses.¹²

Errors to avoid

• Calling ToBI an alphabet — it is a prosodic annotation system, layered over speech in tiers.¹
• Applying English ToBI labels to another language instead of using that language's ToBI system.¹³
• Reading tone labels (H*, L-) as raw F0 measurements rather than phonological categories.¹
• Treating break indices as punctuation rather than perceived juncture strength.¹
• Reusing the Ohio State training materials without checking their licence.¹

Learn more

• Beckman & Elam, Guidelines for ToBI Labelling (v3, Ohio State) — The official guide: tiers, labels, training materials, and domain limits.
• Beckman & Elam — ToBI guidelines, HTML version — Browsable overview, tone tier, break-index tier, and conventions.
• Praat — our companion guide — Where ToBI tiers are actually annotated: Praat TextGrids.
• The IPA — our companion guide — The segmental standard ToBI was designed to complement, not replace.