Suppose a language's syntax is the set of rules for distinguishing the class of things that are expressions of that language, and a language's semantics is the set of rules for assigning meaning to those expressions. A language is a language if and only if it has a syntax and a semantics. Linguistic communication can only occur in a language. So, if Sheng lacks a syntax and a semantics, then it is not a language, and linguistic communication cannot occur in Sheng, or between Sheng and non-Sheng speakers. But linguistic communication does occur between Sheng and non-Sheng speakers, as Mwaura proves by giving examples of various bits of regularly-formed Sheng that he understands and disapproves of. It follows that communication between Sheng and non-Sheng speakers occurs, so Sheng has a syntax and semantics. Now, the grammar of a language is, very roughly, the set of rules for communicating meaningfully in it. So, it's reasonable to suppose that if Sheng has a syntax and semantics, it has a grammar.
This is just as one would expect. Indeed, recent research by the linguist Chege Githiora* has decisively established that Sheng is a proper dialect of Standard Swahili and that it shares Standard Swahili's grammatical structure.
*Githiora, Chege (2002) "Sheng: peer language, Swahili dialect or emerging Creole?", Journal of African Cultural Studies, Vol. 15:159-193.
UPDATE: Potash was already on the case. Check out his hilarious post:
...My parents were brought up on Shakespeare and the bible. The Shakespeare just in case assimilation, of the Kaffir, could be achieved but the bible mostly to tame the heathen- you cannot sjambok a vodoo priest quoth the Native Commissioner. No limey.No.