Proponents of a ‘feature pool’ approach to creolization (e.g. Mufwene 2001, Aboh & Ansaldo 2006) have claimed that the emergence of the new grammar is driven by the syntax-discourse prominence, markedness, and frequency of available features, with typological similarity or dissimilarity of the languages involved playing a crucial role in the competition and selection process. This paper takes a closer look at the predictions of a feature pool-based approach to creolization and tests whether these predictions are borne out by the facts. Three case studies from the Surinamese creoles and Sri Lanka Malay show that the feature pool approach suffers from a number of conceptual, theoretical, and empirical problems. The typology alone of the languages involved in the contact is not a good predictor for the outcome of language contact. The feature pool approach neglects processing constraints: one can only select from what one can process. ‘Creolization’, as in the case of the emergence of the Surinamese Creoles, is not ‘exceptional’, but happens in contact situations in which second language acquisition plays a significant role. The processing restrictions inherent in second language acquisition play an important role in shaping the structural outcome. ’Admixture’, as in the case of Sri Lanka Malay, is not ‘exceptional’ either, but happens in different situations and shows different processes at work. And these processes allow structural outcomes that are very different from those found under the conditions of second language acquisition.
