Final lowering: Fact, artifact or dialectal variation?

Abstract

The existence of final lowering has been disputed by Grabe (1998), who suggested that Liberman and Pierrehumbert’s 1984 finding was due to declination, as there was an extra syllable, ‘‘and,’’ between the last two accents in their materials (e.g. ‘‘RASPberries, MULberries and BRAMBLEberries’’). Arvaniti and Godjevac (2003), however, replicated the results of L&P with and without the extra syllable (c.f. ‘‘LIma beans, NAvy beans and SOY beans’’ vs. ‘‘LIma beans, GREEN beans and SOY beans’’). Here, the materials used in A&G were elicited from speakers of Standard British English (SBE). The results confirmed that the disagreement between L&P and Grabe is due to dialect: the SBE speakers did not exhibit final lowering under either condition. They showed less steep F0 slopes, so all accents after the first had similar scaling, while the last accent showed a rise (rather than a fall, as in American English), enhancing the lack of final lowering. These differences show that final lowering is better seen as a phonological device that a linguistic variety may employ in a given melody. Also, the cross-varietal comparison supports the view that F0 downtrends consist of distinct components (declination, final lowering), each of which may be utilized independently of the others.