Does This Make Me Look Parsable? Spelling Variation and Morphological Structure

What is the relationship of a word like parsable (or parsible) to its morphological constituents parse and able/ible? Much previous work has shown that pronunciation variation reflects processes underlying speech production, such as lexical access and retrieval (e.g. Bell et al. 2009, Gahl 2008) and morphological organization (Kemps et al. 2005; Blazej & Cohen-Goldberg 2015; Plag et al. 2015; Ben Hedia & Plag 2017). In this talk, we argue that spelling variation (aka mistakes) may similarly be used as a window onto the mental lexicon.   

English contains pairs of very similar suffixes such as –able/-ible, -ence/-ance, and -ency/-ancy.  Morphological analyses tend to treat such pairs either as “spelling variants” of single suffixes, or as separate suffixes with identical meaning. Discussions of such pairs usually focus on etymology and on the free vs. bound nature of the derivational bases. However, much actual  written usage of these pairs is left unaccounted for by any of these discussions, as in the following examples:

“Hello - is siri audable on the Apple watch ?”

“ambivalance is irritating and not helpful”

“Under the tuteledge of Coke and Sheila, Tillery began her studies in Afro-Cuban percussion.”

In this talk, we argue that spelling variation in these words reflects aspects of their morphological structure. Based on previous research on morphological productivity and boundary strength (Hay, 2001, 2003; Hay & Baayen 2002, 2005; Baayen & Plag 2009), we hypothesized that correlates of morphological segmentability, such as the relative frequency of base vs. derived frequency, should affect spelling.

We tested this hypothesis using data from Twitter. Consistent with our hypothesis, spelling variation of words partially reflects morphological segmentability, along with other lexical properties, such as lexical frequency.