1. The various members of the '"family" must show phonetic simito one another, in other words be related in character.

2. No member of the "family" may occur in the same phonetic context as any other.

The extreme form of the "physical" conception as propounded by D. Jones (64) and shared by B. Bloch and J. Trager (45) excludes all reference to non-articulatory criteria in the grouping of sounds into phonemes. And yet it is not easy to see how sounds could be assigned to the same phoneme on any other grounds than that substitution of one sound for the other does not give rise to different words and different meaning. This approach may seem to be vulgarly materialist since it views the phoneme as a group of articulatorily similar sounds without any regard to its functional and abstract aspects.

Summarizing we may state that the materialistic conception of the phoneme first put forward by L.V. Shcherba may be regarded as the most suitable for the purpose of teaching.