One of the requirements for the Introduction to Linguistics course that I teach is a term project, for which I ask students to
In plain language: explain something about how a piece of talk works.
More exactly: analyze the communicative effects of some aspects of one or more linguistic performances, attending to at least two different levels of linguistic analysis.
This is just one part of one introductory undergraduate class (it counts for 20% of the grade), but most of the 120 course participants do something interesting. This year, two students looked at the differences in g-dropping rates between musical performances and interviews, for two quite different performers.
Student AF looked at six of Taylor Swift's songs and three of her interviews. In the interviews, AF found that Ms. Swift used the "dropped g" form ([ɪn] rather than [ɪŋ]) in just 2 out of 52 opportunities. (Interestingly, in both cases these were the first -ing words in the interview.) In contrast, in the six songs, she used the "dropped g" form in 87 out of 103 opportunities.
Student JM looked at three of Aubrey "Drake" Graham's raps, and one of his interviews. In the raps, JM found that Drake used the "dropped g" form in 40 of 49 opportunities; in the interview, just once in 17 opportunities.
There are two effects here: the interview format will influence someone to speak in a more formal style, and therefore to use fewer g-dropping forms; and the two musical genres here (country pop and rap) are both traditionally based on varieties of English where g-dropping is normal. Based on these artists' biographies, AF and JM suggest that their native speech patterns may be closer to the way they speak in the interviews, while their musical performances accommodate to the expectations of the genres they work in.