The Atlas of North American English (Labov et al. 2006) defines criteria for participation in certain dialect features in terms of formant benchmarks; e.g., a speaker is considered to have a raised TRAP vowel if their mean normalized F1 of TRAP is less than 700 Hz. Other researchers often compare their own findings to these benchmarks; but the majority of recent research in North American sociophonetics uses the Lobanov (1971) method to normalize formant measurements, producing values that are formally incomparable with Atlas benchmarks. This paper proposes a method of transforming benchmarks into Lobanov-comparable values. Benchmarks are expressed as z-scores relative to the entire Atlas corpus of normalized formant measurements, whose mean F1 is 650.7 Hz (s.d. 150.0 Hz) and mean F2 is 1595.5 Hz (s.d. 435.2 Hz). Thus, for example, a benchmark of 700 Hz in F1 is converted to 0.329 in Lobanov terms. This method is evaluated by comparing the effectiveness of the Lobanov-transformed benchmarks at distinguishing dialect regions to that of the original Atlas benchmarks. Fourteen such benchmarks are evaluated against three isogloss parameters; in 76% of cases, the Lobanov-transformed benchmarks are at least as effective as the original Atlas benchmarks at characterizing the Atlas' dialect regions. Therefore, this transformation can be recommended for researchers who want to compare Lobanov-normalized data to Atlas benchmarks.
Dinkin, Aaron J.
"Adjusting to the New Normal(ization): Adapting Atlas of North American English Benchmarks to Lobanov-Normalized Data,"
University of Pennsylvania Working Papers in Linguistics: Vol. 28:
2, Article 5.
Available at: https://repository.upenn.edu/pwpl/vol28/iss2/5