AbstractThe current study is an analysis of an Urdu speech corpus using a Tone and Break Indices (ToBI) transcription system to develop a model of Urdu intonation. The analysis indicates that Urdu has three pitch accents (L*, L*+H, H*) and boundary tones associated to two phrase types: accentual phrase (AP) boundaries (Ha, La) and intonational phrase (IP) boundaries (L%, H%, LH%). The AP is a pitch bearing unit on a single word, or more than one word in the context of (a) izāfat, (b) conjunctive vāo, (c) case markers, (d) complex postpositions, and (e) complex verbs. Moreover, this study also investigates the tonal structure of declarative, interrogative (wh-questions, yes/no-questions), and imperative (semi-honorific, polite honorific) sentences in neutral focus context using 50 utterances produced by ten speakers. Results indicate that (i) all declarative sentences consist of a series of APs, represented as (aL) L* (H) Ha, except the sentence final AP, represented as (H*) L%. (ii) wh-questions are different from their corresponding declaratives in terms of pitch range and the final boundary tone; (iii) imperatives are different form their corresponding declaratives in terms of final boundary tone.