Colloquium: Marlyse Baptista, Susan Gelman & Erica Beck - "Testing the Convergence Hypothesis in Creole Genesis"
The literature on language contact (Heine & Kuteva, 2001, 2005; Hopper & Traugott, 1993; Mufwene, 2008; Siegel, 2000, 2008;Thomason & Kaufman, 1988; Thomason, 2001; Winford, 2003) is replete with examples illustrating convergence. This process has been the focus of studies on bilingual speech (Silva-Corvalán, 1994; Muysken, 2002; Toribio, 2004; Bullock and Toribio, 2004; Poplack, Zentz and Dion, 2012) and on creole languages (Sankoff and Brown, 1976; Kihm, 1980; Cassidy, 1988; Mufwene, 2001; Kouwenberg, 2001; Siegel, 2008).
For the past decade, Mufwene (2001, 2008), Siegel (2008) and Aboh (2009) have put forth the framework of competition and selection as a useful tool to account for creole genesis. This framework assumes that linguistic features present in multiple source languages compete for survival in the newly created language. Based on this assumption, two questions emerge: how do the features that participate in creole formation get selected and why are some of the features favored over others?
In this respect, the process of convergence is particularly revealing, as some features from the creole substrates and superstrate are believed to “converge” in the phonological, morphological, semantic and syntactic components of the newly emerged language. Some scholars (Kouwenberg, 2001) have been critical of the methodology used to detect convergence, as it presents convergence as an ad-hoc, ‘after the fact’ type of process without solid evidence.
The objective of this collaborative paper is to fill the methodological gap by making the mechanism observable and by formulating a testable hypothesis to demonstrate it. We present results from an experiment involving 93 native English speakers to test the hypothesis that speakers most readily make use of morphological and semantic elements that converge across multiple sources of input. Building on experimental methods using artificial languages (Hudson-Kam and Newport, 2005, 2009; Billman, 1983; Frank, Tenenbaum, & Gibson, 2013 among others), we independently varied the morphological similarity and the functional overlap between elements in a new “language” and elements in the speaker’s extant language, to assess the conditions under which novel linguistic elements are acquired.
The study design includes three conditions: In the Congruent condition, we introduced two elements that are similar in form to morphemes in English, and with similar function. In the Reversed condition, the morphemes are the same, but the functions are reversed, so that the form/function mapping would no longer correspond to English. In the Novel condition, the functions are constant, but the forms are not an English morphological unit.
We predicted that participants would learn most readily in the Congruent condition. It was unclear at the beginning of the experiment which of the other two conditions would be more difficult: the Reversed condition (which potentially provides the most interference with English, because familiar units are used with novel functions) or the Novel condition (in which the units themselves are wholly novel). Our experiment shows that in the production tasks, the Congruent is the easiest to acquire and the Novel condition the most difficult.