To honor Jorge on this special occasion, I would like to present a critical review of one of his best-known works, Deletion in Coordinate Structures. Alas, I still haven't read it. Also, I don't remember enough about syntax. None of this would be a problem if it were June, since we professors are unemployed over the summer. But as the deadline for submission draws near, I thought it wiser to switch topics.

Instead I'm going to provide the beginnings of an Optimality Theoretic analysis of Jorge's soccer playing. I've collected a lot of data on the topic over the past decade. It turns out that most of a player's soccer strategies are a function of a hierarchy of a number of fixed, violable constraints in a strict dominance relation. Or at least, when viewed from the perspective of OT it sure seems that way.

First some notational conventions:

(1)	J	=	Jorge
	E	=	An enemy, i.e., someone on the other team
	o	=	The soccer ball
	J(o)	=	Jorge controls the soccer ball
	E(o)	=	The enemy controls the soccer ball
	#J	=	Jorge is injured
	#E	=	The enemy is injured

Here are the constraints needed for the analysis:

(2)	J(o)
	*E(o)
	*#J, *#E

Implicit in the first two constraint formulations are some fundamental markedness notions involving Jorge's playing. In this regard J(o) and *E(o) might be compared to ONSET and NOCODA respectively. The existence of J(o) and nonexistence of E(o) predicts, all else equal, that there will be events in which Jorge strives to control the ball, and events in which Jorge is not involved; crucially, there are no events predicted in which Jorge strives to give control of the ball to a member of the opposite team. Similarly, given *E(o) and no *J(o), there will be events in which Jorge attempts to remove control of the ball from a member of the opposite team, and events in which Jorge is not involved. However, there should be no events in which Jorge strives to wrest control of the ball from himself. These predicted markedness patterns correlate very highly with observed fact, though there are questions involving interpretation of the data that go beyond the scope of this unfinished squib.

To begin to see how these constraints are ranked, consider the following tableau. Inputs consist of two or more players, a ball, and an optional '#', meaning that someone will have to be hurt to get the ball. Candidate (a) below involves Jorge in control of the soccer ball, while it is the Enemy who controls the ball in (b). Given the ranking *E(o) >> *#J, candidate (a) wins, in spite of the injury implied to Jorge.

(3)	/ J o E # / *E(o) *#J a. F #J(o) E   * b.     J #E(o) * !

This ranking predicts that faced with the choice, Jorge will choose to get control of the soccer ball rather than avoid injury. Such events are abundantly attested in soccer games, and so this prediction is strikingly confirmed. Note that there are less obvious consequences of the analysis that also turn out to support it. For example, Jorge will play soccer even when he has the flu. This is obviously injurious to him, but it ensures that the Enemy will control the ball less.

At this point the rest of the analysis, and the typological consequences, should be clear. I have a soccer game to go play.

Happy birthday, Jorge.