The presentation discusses an ongoing change in the TAM-marking system of Juba Arabic (: JA), a pidgincreole spoken in the South Sudan. The TAM-marking system of JA is based upon two core proclitc markers: bi= and ge= (which cannot combine among themselves). Tosco (1995) provided an outline of the JA TAM system along the following lines: •Ø-marked verbal forms: imperative; after modal verbs; past punctual for nonstative verbs; usually imperfective for statives and in nominal (verbless) sentences; •bi: irrealis (future, conditional, …): 1.alélamátarbi=nézil tonightrainbifall ‘tonight it will rain’ •ge: nonpunctual: 2.ánalísag(e)=istákal 1SGstillgework ‘I am still working’ As to the expression of the habitual, it is well known that natural languages (and, among them, pidgins and creoles) often choose to “assign” habituality to either an irrealis or a progressive marker. Portuguese-lexified creoles such as São Tomense and Cape Verdean belong to the first group, while many French- and English-based creoles choose the other option. In JA the expression of habituality was instead split between ge and bi on the basis of its “actuality” vs. “generality” or “potentiality:” 3.ánage=rúafisukkúluyom 1SGgegotomarketallday ‘I go to the market every day’ vs. 4.makwesítabi=kárabumóya NEGgood2SGbispoilwater ‘it is not good that you spoil water’ While 3. reports an action which takes place regularly (although very possibly not at the time of speaking), 4. describes a possible state of affairs which could, but not necessarily will, come into being. Additional TAM markers include in primis the anteriority marker kan, which can combine with either bi= or ge= for expressing a counterfactual conditional or a past progressive. The JA system therefore neatly expresses tense and aspect by means of the two core markers bi= and ge= and it could be defined as aspect-based; it was possibly derived from the aspectual system found in JA’s lexifier, Sudanese Colloquial Arabic, where proclitic bi= overlaps with the auxiliary gaa‘id (a possible source for JA ge) in marking the habitual and the progressive of action verbs (Ali and Miller 1986). Recent (Summer 2013) fieldwork in Juba has shown that this system is largely superseded nowadays by a mood-based system where ge= is devoted to the marking of a realis (including the habitual), while bi= specializes for marking an irrealis mood, including factual conditionals: 5.kélibdege=kóre dogDEMgecry ‘The dog barks’ (i.e., ‘Dogs bark’) 6.kanítadúgukélibdeúobi=kóre if2SGbeatdogDEM3SGbicry ‘If you beat the dog, it will bark’ The presentation will discuss the possible causes of such a dramatic change. It is remarkable that, while the Arabicization of JA is still very much in progress, the new TAM system represents a radical departure from Arabic.