1) We communicate using sounds - not movements of the tongue or other articulators. Articulatory movements are merely a means to an end. In many ways, it doesn't matter very much how the sounds are generated: whether by a human vocal tract, a speech synthesis program on a computer, a sound recording on CD, or on the TV, or a parrot.
(Jakobson: "We speak in order to be heard")

2) We can actually produce a given sound in various different ways. For example, interdental vs. postdental [θ]; [a] with clenched teeth. Contrarily, a given movement may yield different acoustic effects (e.g. [ti], [ta], [tu]). Therefore, we want to understand the way in which sounds are generated in the vocal tract.

3) Many movements go on together in articulation, but we hear a single stream of sound.

4) Meaningful differences between words are encoded as sound differences. Some natural classes (e.g. labials and velars; retroflexion, rounding, pharyngealization, dentalization) make sense only in acoustic terms. Some aspects of speech (e.g. friction) can only be properly defined in acoustic terms.

Key Topics:

What is sound?
How are sounds are generated in the vocal tract?
How do sounds differ from one another?