We investigate effects of word order, (embedded) clause boundaries as well as sentence parsing on intonation by analysing read speech elicited from native speakers of Japanese, a language with word order alternation. All stimuli contain one dative phrase, which belongs to either the embedded or matrix clause. a. NP1-nom [NP2-nom/dat NP3-dat/nom NP4-acc V1 Comp] V2 b. NP1-nom/dat NP2-dat/nom [NP3-nom NP4-acc V1 Comp] V2 We compared: 1) & 2) nominative–dative vs. dative–nominative order (in a and b, respectively) to test the word order effect; 3) the embedded vs. matrix dative phrase placed at NP2 to test the clause boundary effect. Speakers read the stimuli twice, once instructed to read the sentence silently to parse it before reading it aloud for recording, and once to start reading aloud as soon as the sentence appears on the computer screen, to test the parsing effect. Words are either all lexically unaccented or accented to check the effect of the lexical pitch accent. Results show significant effects on f0 by lexical pitch accents and parsing, but only partial effects by clause boundary, and no effects by word order.