# What's the difference between counterbleeding, bleeding and feeding?

Bleeding is when rule A prevents rule B from applying. But counterbleeding is when two rules are ordered too late to bleed. I see counterbleeding the same as feeding.

Let's say you have some segment X. Rule A and B may only apply to X. A applies first and changes X->Y. Next, B applies to Y but it is in vacuous application, so nothing happens. In this scenario A is said to bleed B.

Now let's say we have rule A and B, but this time let's just say A applies to X or Y. Now, when we apply rule B first. Let's say B changes X->Y , and next A applies to Y. If A had applied first, then it would have bleeded B. This seems to be a feeding relationship.

Am I missing anything?

I'm going to start by going through some more detailed definitions of the various rule orderings, with examples:

If rule A creates the environment for rule B to apply, then rule A feeds rule B, e.g.

A: ɣ → ∅ / V_V
B: e → i / _e

/teɣe/
A: tee
B: tie
[tie]

If the order A B is a feeding order, then the order BA is a counterfeeding order, i.e. applying the rules in order BA would fail to feed, e.g.

/teɣe/
B: ----
A: tee
[tee]

Counterfeeding is opaque in the sense that the surface form violates the generalisation that [ee] sequences are generally altered via the raising rule (B).

If rule A destroys an environment that could have triggered rule B, then A bleeds B, e.g.

A: V → ∅ / VVC_
B: stop → fricative / V_V

/heidinir/
A: heidnir
B: ----
[heidnir]

If the order AB is a bleeding order, then order BA is a counterbleeding order. So, applying the rules in the order BA would fail to bleed:

/heidinir/
B: heiðinir
A: heiðnir
[heiðnir]

Counterbleeding is an opaque rule ordering: The surface form violates the generalisation that fricatives appear between vowels.

Note that based on the definitions i've given, counterbleeding is not the same as feeding. In the feeding example, rule A creates the environment for rule B to apply - if rule A hadn't applied, then rule B couldn't apply. In the counterbleeding example, application of B simply does not prevent rule A from applying - even if rule B hadn't applied, rule A would still have been able to apply. Let's consider your concrete example in light of this. Let's say that X is a stop and Y is a fricative, and the environment in question is intervocalic, so:

"Let's say you have some segment X. Rule A and B may only apply to X. A applies first and changes X->Y. Next, B applies to Y but it is in vacuous application, so nothing happens. In this scenario A is said to bleed B."

A: stop → fricative / V_V
B: stop → [-voice] / V_V

We can see that in the order AB, application of A will bleed application of B. Exactly.

A: aza
B: ----
[aza]

"Now let's say we have rule A and B, but this time let's just say A applies to X or Y. Now, when we apply rule B first. Let's say B changes X->Y , and next A applies to Y. If A had applied first, then it would have bleeded B. This seems to be a feeding relationship."

A: [-sonorant] → ∅ / V_V
B: stop → fricative / V_V

A: aa
B: ----
[aa]