posted
June 22, 2023
Project overview
Part of my Ph.D. comprehensive examination consists of writing two full, single-authored research papers of publishable quality. This is the first one of those papers. I was inspired by a study written by Bresnan et al. (2007). In their original paper, the authors challenged the traditional reliance on made-up examples in linguistics. They showed that by analyzing real-world language patterns using advanced statistical techniques, we can uncover a much richer and complex picture of how grammar actually works. This is particularly evident in phenomena like the dative alternation, where speakers have a choice between seemingly interchangeable phrasings, like I gave the book to Mary vs. I gave Mary the book.
Consider this – if we compare (i) Who sent the box to Germany? with (ii) Who sent Germany the box?, speakers will likely find (ii) somewhat odd; this is because the animacy of the recipient (Germany – an inanimate noun) is one important factor that influences the choice between one phrasing vs. the other (Winter, 2020). In contrast, when there is an animate recipient, the (ii) structure/type seems to work just fine, e.g., Who sent Sarah the box? Other factors governing the dative alternation are the length of the constituents, the type of noun phrase (e.g., pronominal vs. non-pronominal), whether the recipient/patient was mentioned in previous discourse or not, among other factors. Bresnan et al. (2007) showed that linguistic predictors must be examined using multilevel modeling, because such factors are nested and co-dependent, acting simultaneously upon the outcome variable.
What I did in my own paper was to explore whether speakers of other languages show similar usage patters of the dative alternations compared to English speakers. Callies & Szczesniak (2008)
Gries & Deshors (2015)
Doing the analysis with R inside Emacs, using the package Emacs Speaks Statistics (ESS).