Instances of a target language contrast being lost in a second language system are well known (e.g. loss of contrast between [r] and [l] in the English of many Japanese and Chinese speakers). The converse of this phenomenon, however, is unexpected (e.g. a hypothetical case of American speakers of Japanese with a contrast between [r] and [l] in their Japanese). Yet, precisely such a phenomenon is found in M(alayalee) E(nglish) (henceforth ME), which exhibits a contrast between alveolar (n, l, r) and retroflex (N, L, R) consonants:

a.	slowness	[sloonas]
	bonus	[booNas]
b.	lawless	[lOOlas]
	folly	[fOOLi]
c.	very	[weri]
	merry	[meRi]

While the ME contrast between retroflex and alveolar consonants is rooted in the parallel contrast in Malayalam, the patterns of distribution and alternation intertwined with the contrast in ME are not found in Malayalam. For instance, the contrast in (a) and (b) is crucially tied up with morphology: in ME, nasals and laterals are never retroflex morpheme-initially, and they are never alveolar after front vowels within the same morpheme. The latter restriction is not found in Malayalam. Alveolars and retroflexes in ME also exhibit patterns of alternation not found in Malayalam, as illustrated by the pair take [Teek] and intake [inteek].

In this talk, we will lay out the facts of some of the patterns in ME that illustrate 'persistence' and 'emergence' in a second language system. We will focus particularly on the patterns of the alveolar-retroflex contrast in ME, and provide an analysis for the facts within a version of Optimality Theory that, in addition to the ranking of universal constraints and stipulations on individual morphemes, permits language particular statements that (i) specify the domains, loci, and triggers in universally underspecified constraints, (ii) select from universally specified options in a constraint, and (iii) (de)active universal constraints. 

The facts and analysis of the patterns of distribution and alternation of the alveolar-retroflex contrast are of special interest to the theory of language contact. We will show that the patterns under consideration cannot be explained in terms of a theory that views patterns of a second language system (the offspring) as a mere combination of the patterns of the target language (superstrate), the first language (substrate), and universals. Instead, we will argue for an analysis in which the unique properties of an offspring system that distinguish it from its parents evolve as a response to the conflicting pulls from the parent languages, within a universal grammar space.