In phonology and historical linguistics, feeding order of phonological rules refers to a situation in which the application of a rule A creates new contexts in which a rule B can apply; it would not have been possible for rule B to apply otherwise.
Suppose there are two rules. Rule A takes in input x and returns output y. Rule B takes in input y and returns input z. When rule B is applied to input x, it will return the same output (x). The following order is called a feeding order:
A: x→y
B: y→z
The opposite of feeding order, the situation in which rule A destroys a certain context so rule B can no longer apply, is called bleeding order.
An example of feeding order can be seen in English.
Rule A is fortis stop insertion, which inserts voiceless plosives into consonant cluster codas consisting of nasals and voiceless fricatives.
A: ∅→[-son, -cont, -del rel, -voi, αplace] / [+nasal] _ [-son, +cont, -voi, αplace]
Rule B is preglottalization, which glottalizes voiceless stops in word-final codas. As a consequence of this rule, all voiceless plosives which make part of a word-final consonant cluster are glottalized.
B: [-son, -cont, -del rel, -voi] → [+constricted glottis]/ _ (C)#
In English, rule A precedes rule B. We can derive the surface form of prince by applying the rules to the underlying form /prɪns/ (x in the generalization above). Using rule A, /prɪns/ becomes prɪnts (y in the generalization above); using rule B, which can now be applied as there is a voiceless stop in a word-final coda, prɪnts becomes prɪnʔts. Thus, the final output form of both rules is [prɪnʔts] (z in the generalization above).
Since rule A created a phonological context in which rule B could apply and because rule B could not apply without the application of rule A, the two rules are in feeding order; that is, rule A feeds rule B.
If the order of rules which are in feeding order is reversed, this is said to be a counterfeeding order.