Two different definitions of phase shifts and time delays are contrasted and shown to match different experimental methods of generating delayed pulses. Phase shifts and time delays are usually defined in terms of a carrier wave in magnetic resonance, but definitions based on the envelope of a single pulse are useful in optics. It is demonstrated experimentally that a frequency domain measurement using spectral interferometry can simultaneously measure phase shifts with an accuracy of 0.1 rad (2 sigma) and time delays with a precision of 40 attoseconds (2 sigma) for 25 femtosecond optical pulses. Envelope time delays are generated by pathlength differences in an interferometer. Constant spectral phase shifts are demonstrated by diffracting pulses from a variable phase volume diffraction grating. Experimental requirements for phase-resolved spectroscopy are outlined. The theory of phase-locked pulse pair techniques is reexamined, and it is concluded that linear experiments with phase-locked pulse pairs are completely equivalent to Fourier transform absorption spectroscopy and do not measure the refractive index or real part of the susceptibility. It is shown that Fourier sine and cosine transformations of truncated time domain signals which do not match the symmetry of the complete signal can produce a false dispersive susceptibility because they are equivalent to Kramers-Kronig inversion of finite bandwidth absorption data. A procedure for shifting pi/2 phase-locked transients by a quarter cycle of delay to generate a transient with a pi/2 spectral phase shift is given. Equations used to calculate femtosecond nonlinear optical signals have assumed carrier wave delays. Modifications to these equations are required when envelope delays are generated by interferometer pathlength differences and modified equations are given. The modified equations yield significantly different results for phase-resolved or interferometric experiments. In particular, the modified equations are needed to calculate indirectly (interferometrically) detected frequencies and the real and imaginary parts of two-dimensional Fourier transform spectra. The role of the refractive index and real part of the frequency domain susceptibility in nonlinear experiments with phase-locked pulse pairs is explored. It is concluded that experiments such as the heterodyne detected stimulated photon echo are insensitive to nonlinear refractive index changes under some circumstances. Finally, modifications of some equations used in the theory of coherent control are needed to match theory with experimental practice.