Angle configurations and the S -centroid: modeling changing planar locations in the vowel space

1. Anne Fabricius, Roskilde University Dominic Watt, University of York and J.P French Associates Angle configurations and the S-centroid: modeling changing planar locations in the vowel space

2. Outline • Background to the paper • Aim of the research • Introducing the S-centroidanchormethod • Application to somevowel data from British English (SSBE/Modern RP) • Discussion and furtherimplications Angle configurations and the S-centroid

3. Background • Quantitativemethods in a quantitativediscipline • The art and science of vowelchange (Labov 1994) • Earlierworkthisstudy is basedon • Watt and Fabricius S-centroidmethod (Watt and Fabricius 2002) • TRAP/STRUT rotation in RP (Fabricius 2007) and angle methodsusingvoweljuxtapositions • Testingnormalisationmethods (Fabricius, Watt and Johnson 2009) for geometrically-relatedproperties • Here: Combining the two (normalization and modellingchangingvowelloci distributions around the centroid) Angle configurations and the S-centroid

4. A summary of the original S-centroidnormalizationmethod Adapted from Fabricius 2007: 300 Angle configurations and the S-centroid

5. Modified W&F (tested in Fabricius, Watt and Johnson 2009) Flynn 2010: slide 16 For variety-specificmodifications of the Watt and Fabricius S-centroidmethod, seealsoDurian, forthcoming and Bigham 2008. mW&F uses F1 of [a] only Angle configurations and the S-centroid

6. (Geometrical) Centroid ≠ Spectral Centroid • In geometry, the centroid, geometric center, or barycenter of a plane figure or two-dimensional shape X is the intersection of all straight lines that divide X into two parts of equal moment about the line. Informally, it is the "average" (arithmetic mean) of all points of X. • (From Wikipedia) Angle configurations and the S-centroid

7. Peeters 2004 (Geometrical) Centroid ≠ Spectral Centroid Angle configurations and the S-centroid

8. Aim of thispaper • To present the method • To investigate the methodologicaladvantages for sociophonetics of representing relative planar locations as vectors vis à vis a pre-definedgeometrical centroid of the vowelspace • To ask whetherthis supports other arguments in favour of a centroid-basednormalisationmethod? (in the spirit of Fabricius, Watt and Johnson 2009) Angle configurations and the S-centroid

9. The definition of the S-centroid point • Makenoclaimshereabout the centroid’sperceptualsignificance (although the conceptdoes feature in someperception/normalisation research, eg Deterding 1990) • Usedhere as a geometric point to investigate and illustrateproperties of vowel distributions in F1/F2 space Angle configurations and the S-centroid

10. Somepreviousphonetic studies using the centroid concept • Effect of speech disorderson the vowelspace and distance of vowels from the centroid (e.g. in stuttering: Blomgren, Robb and Chen 1998) • Pickering 1986 formalisedperipheralitymeasured as dispersion from a centroid, context speech perception research • Hyper- and hypo-articulation, ie clear and indistinct speech: (Lindblom 1990, 1996, Ferguson & Kewley-Port, 2002; Picheny, Durlach, & Braida, 1986) • Whiteside 2001; NB definition of centroid usedherediffers from thispresentationsinceaxesarederiveddifferently (using Bark differences) Angle configurations and the S-centroid

11. Announcing the S-centroidanchormethod • R-algorithmdeveloped by Daniel Ezra Johnson after an idea by Anne Fabricius • Availablehere (The Modern RP Page) Angle configurations and the S-centroid

12. F2 90° 180° F1 0° S-centroid point -180° -90° Angle configurations and the S-centroid

13. F2 90° u i 180° F1 0° S-centroid point -180° -90° a Angle configurations and the S-centroid

14. Angle configurations and the S-centroid

15. So far…. • Documentingchanges by measurements in degrees vis a vis a stable point, ratherthaneyeballjudgments of relative placement • Couldalsobeused in combinationwithEuclidean/Cartesian distances (as in Fabricius 2007, Richards, Haddican and Foulkes 2009) • Quantificationenablesfurtherstatisticaltesting • Has potential applications in determining the nature of centre versus peripheryin the vowelspace (Labov 1994) in a more reproducibleway Angle configurations and the S-centroid

16. Demonstration of planar locations and theirrelationships • RP generationalvowel data from Hawkins and Midgley 2005 and Moreiras 2006, plus Fabricius 2009 • using R script devised by Daniel Ezra Johnson • Thisexample: short vowel system with lines connectingaveragevowelloci Angle configurations and the S-centroid

17. Female speakers, 1998 cohort (Data from Fabricius 2009) Angle configurations and the S-centroid

18. Female speakers, 2008 cohort (Data from Fabricius 2009) Angle configurations and the S-centroid

19. Male speakers, 1998 cohort (Data from Fabricius 2009) Angle configurations and the S-centroid

20. Male speakers, 2008 cohort (Data from Fabricius 2009) Angle configurations and the S-centroid

21. The new alternative: the S-centroidanchormethod • To makethese types of configurations more easilycomparable • By using the S-centroid point as anchor • deriving angles vis-à-vis the centroid point • The S-centroid point is common to ALL speakers in the sample sincetheyare all normalisedusing the W&F (or mW&F) method • Advantage: the S-centroiddoes not move over time Angle configurations and the S-centroid

22. Illustration Angle configurations and the S-centroid

23. FLEECE stable over time: OM, OF: older generation; YM, YF: younger generation Angle configurations and the S-centroid

24. F2 90° FLEECE 180° F1 0° S-centroid point -180° -90° Angle configurations and the S-centroid

25. FOOT moving over time Angle configurations and the S-centroid

26. F2 90° FOOT, YF3 FOOT, older 180° F1 0° S-centroid point -180° -90° Angle configurations and the S-centroid

27. LOT variability, mostlyamongyoungerfemale speakers Angle configurations and the S-centroid

28. STRUT in thisperspectivealso variable; nb Reducedscalehere Angle configurations and the S-centroid

29. To sum up • Whatare the methodologicaladvantages to representing relative planar locations as vectors vis a vis the centroid location of the vowelspace? • Quantifiability, reproducibility, visualevidence backing auditory perceptions • Doesthisargue for the advantages of a centroid-pointbasednormalisationmethod? • Yes, and the methodcouldalsobeadapted to work from the centroid-based Lobanov normalisationalgorithm. • But Lobanov’s normalisationmethod is in some cases toopowerful (close to a standard statisticalnormalisationtechnique) and performslesswell overall (in severaltesting parameters) than mW&F in Flynn’scomparison of 20 normalisationalgorithms (Flynn 2010) Angle configurations and the S-centroid

30. Conclusion • Methodofferedhere as an aid to the sociophonetic community • The illustrative charttemplatewillalsobeavailable online ( MS Excel.crtx file) • NB A Euclidean distance metriccouldbeincluded as well • R-codewillbeavailable and canbeadjusted Angle configurations and the S-centroid

31. References 1 • Bigham, Douglas. 2008. Dialect contact and accommodation among emerging adults in a university setting. Ph.D. thesis. The University of Texas at Austin. • Deterding, David. 1990. Speaker Normalisation for Automatic Speech Recognition, Unpublished PhD Thesis, Cambridge University. • Fabricius, Anne. 2007. Variation and change in the TRAP and STRUT vowels of RP: a real time comparison of five acoustic data sets. JIPA 37:3: 293-320. • Fabricius, Anne. 2009. Short vowels in real time: TRAP, STRUT and FOOT in the South of England. Paper presented at ICLaVE 5, Copenhagen. June 27th 2009. (www.ruc.dk/~fabri ) • Fabricius, Anne H., Dominic Watt and Daniel Ezra Johnson. 2009. A comparison of three speaker-intrinsic vowel formant frequency normalization algorithms for sociophonetics. Language Variation and Change, 21,3:1-23. • Flynn, Nicholas. 2010. Comparingvowelformantnormalisation procedures. Talk given at York Postgraduate Mini-conference, June 10th, 2010. • Hawkins, Sarah and Jonathan Midgley. 2005. Formant frequencies of RP monophthongs in four age groups of speakers. JIPA 30: 63-78. • Labov, William. 1994. Principles of Linguistic Change volume 1: Internal Factors. Oxford:Blackwell. Angle configurations and the S-centroid

32. References 2 • Lindblom, B. 1990: Explaining phonetic variation: A sketch of the H&H theory, in Speech Production and Speech Modeling, edited by W. J. Hardcastle and A. Marchal. KluwerAcademic, Netherlands, pp. 403–439. • Lindblom, B. 1996: Role of articulation in speech perception: Clues from production. Journal of the Acoustical Society of America, 99, 1683–1692. • Moreiras, C. 2006. Anacoustic study of vowel change in female adult speakers of RP. Unpublished undergraduate dissertation, University College London. • Peeters, Geoffroy. 2003. A large set of audio features for sound description (similarity and classification) in the CUIDADO project. http://recherche.ircam.fr/equipes/analyse-synthese/peeters/ARTICLES/Peeters_2003_cuidadoaudiofeatures.pdf • Pickering, J.B. 1986. Auditory vowel formant variability. Unpublished doctoral dissertation, Oxford University. • Richards, Hazel, Bill Haddican and Paul Foulkes. 2009. Exhibiting standards in the FACE of dialect levelling. Paper presented at ICLaVE 5, Copenhagen, June 2009. •  Watt, Dominic and Anne Fabricius. 2002. Evaluation of a technique for improving the mapping of multiple speakers’ vowelspaces in the F1-F2 plane. Leeds Workingpapers in Linguistics and Phonetics. 9: 159-173. Angle configurations and the S-centroid

33. Acknowledgements • Daniel Ezra Johnson • TylerKendall • Nicholas Flynn • Nicolai Pharao Angle configurations and the S-centroid

34. Thankyou for listening!