For primary frequency ratios, f2/f1, in the range 1.1-1.3, the fixed-f1 ('f2-sweep') phase derivative of the 2f1-f2 distortion product otoacoustic emission (DPOAE) is larger than the fixed-f2 ('f1-sweep') one. It has been proposed by some researchers that part or all of the difference between these delays may be attributed to the so-called cochlear filter 'build-up' or response time in the DPOAE generation region around the f2 tonotopic site. The analysis of an approximate theoretical expression for the DPOAE signal [Talmadge et al., J. Acoust. Soc. Am. 104, 1517-1543 (1998)] shows that the contributions to the phase derivatives associated with the cochlear filter response is small. It is also shown that the difference between the phase derivatives can be qualitatively accounted for by assuming the approximate scale invariance of cochlear mechanics. The effects of DPOAE fine structure on the phase derivative are also explored, and it is found that the interpretation of the phase derivative in terms of the phase variation of a single DPOAE component can be quite problematic. (C) 2000 Acoustical Society of America.