There are two main sources of variability in natural languages: sociolinguistic (probabilistic) variability and parametric variation. Although sociolinguistic variation has long been recognized as an integral part of the linguistic system, its impact on the acquisition of grammar has barely been explored. We present experimental data from Chilean Spanish (ChS) and Brazilian Portuguese (BrP), two languages in which there is significant sociolinguistic variation in the production of plural morphology but which also differ parametrically in the licensing of bare count nouns in argument position. We argue tha the acquisition patterns in the two languages differ, but are consistent with the idea that input variability may cause ambiguity and delay acquisition of categorical properties.

In both ChS and BrP, lack of plural morphology in the NP does not guarantee that the referent is to be interpreted as a singleton (one) or as a mass term. In ChS syllable-final /s/ is subject to a lenition process that reduces [s] to [h] or zero. This also affects the realization of plural /-s/, resulting in the syntactic plural feature sometimes realized as zero. In BrP, agreement is also subject to variation, but the variation is restricted to number and person morphology which is sometimes not overtly realized, again causing ambiguity. In both languages, the variability is probabilistic and dependent on both linguistic and extralinguistic factors (e.g., syntactic category and social class). While the ChS variability is, for the adult, a low level phonetic variability, the BrP variability targets agreement only (Miller 2007; Cepeda 1995; Pereira & Scherre 1995). Additionally, in BrP but not in ChS, numberless bare count nouns can appear in subject and object positions with a singular or plural interpretation depending on the context (Schmitt & Munn 2003). ChS bare count nouns are only allowed in a restricted set of contexts (objects of possession verbs, for example) (Espinal & McNally 2011). This fact provides more evidence to the child that lack of plural does not always mean singular-one in BrP than in ChS.

In this paper I present three studies examining the interpretation of plural morphology and its absence in three contexts in both Brazilian Portuguese and Chilean Spanish. Study 1 examines children's ability to use number to make count/mass distinctions in comparative contexts such as (1). (1) Quem tem mais corda/cordas? (Brazilian Portuguese) Who has more string/strings? Study 2 examines children's ability to interpret bare count singulars and bare plurals in contexts such as (2). (2) Que menina tem livro/livros? (Brazilian Portuguese) Which girl has book/books? Study 3 examines children's ability to interpret the indefinite singular and the indefinite plurals in contexts such as (3). (3) Dá um livro/uns livros pra tartaruga. (Brazilian Portuguese) Give a book/some books to the turtle. The results show that both the language and the type of variability matter. Plural morphology is ignored more often by Chilean Spanish speaking children than Brazilian children. On the other hand, and much as predicted by the parametric variation, lack of plural morphology does not guarantee a one interpretation in the bare count noun experiment and the mass/count experiment more often for Brazilian children than for Chilean Spanish speaking children. I examine the results with respect to current analyses of the plural and analyses of bare singulars/bare plurals in both languages.