Certain facts about the stresses are thus automatic: the number of loud stresses, the occurrence of some of the weak stresses. We therefore mark as phonemic only the remaining non-autimatic facts: the place of the loud stress, and where necessary the place of any secondary stress. In a similar way, English contrastive stress (1040 in 'distribution', 'not production') would be discovered, since when it does occur it hardly ever appears more than once between two word junctures.
- Harris (1944), Pag. 190