Phonetic variation from the bottom up: evidence from Liverpool English plosives

1
Phonetic variation from the bottom up evidence
from Liverpool English plosives

• KEVIN WATSON
• Lancaster University
• k.d.watson_at_lancaster.ac.uk

2
Introduction

• Phonetics in (and versus) phonology
• What is Liverpool English?
• What is lenition?
• What is Liverpool English lenition?
• The phonetics and phonology of LE stops
• Marrying etics and -ology

3
Phonetics in phonology

• It is well known that phonetics, phonology and
  sociolinguistics have traditionally had an uneasy
  coexistence

4
Phonetics in phonology

• This uneasy relationship can be connected to the
  well-cited distinction between language
  competence and language performance

5
Phonetics in phonology

• Under this view, there would be no point in
  doing research on language variation and
  linguistic theory for variation would have
  nothing to do with competence, and linguistic
  theory would have nothing to do with anything
  outside of competence (Guy 1997 127)

6
Phonetics in phonology

• This view has not disappeared
• e.g. Newmeyer (2003)
• defends the notion that speakers internal
  grammar is separate from their usage
• criticises the use of corpora because they are
  collections of performance data

7
But

• It has been shown time and again that language
  variation is part of speakers competence
• (Docherty, Foulkes, Milroy, Milroy Walshaw
  1997, Docherty Foulkes 2000, Docherty Foulkes
  in press)

8
So

• We should be pursuing a view of language and
  linguistics that is as encompassing and
  integrative as possible. It is simply good
  science to try to explain the broadest possible
  range of facts and to take note of all relevant
  data. Hence our theories should de designed to
  have utility in accounting for both language
  structure and language use. (Guy 1997 141)

9
The underlying theme of the talk

• The relationship between language structure and
  language usage
• That is, between
• Phonology
• Phonetics
• Sociolinguistics

10
What is Liverpool English?

• One third Irish, one third Welsh, and a third
  catarrh
• Liverpool English is said to have been born
  around the 1850s when high numbers of Irish
  people arrived in the city
• Not just the Irish, though
• Popularly (and probably falsely) thought to
  originate because of air pollution

11
Liverpool English Phonology

• STRUT and FOOT - /?/
• DANCE - /a/ not /??/
• START vowel is fronted to /a/
• NURSE and SQAURE MERGER
• LOOK and COOK have /u/
• diphthongal

12
Liverpool English Phonology

• The NURSE/SQUARE merger is to the front vowel
  /??/ rather than the central /??/ found in other
  northern English accents
• The word-final weak vowel is typically ?
  (walker, winter)
• Dental fricatives ? and ? are often stopped
  (e.g. this ???? and there ????)
• /r/ is tapped (e.g. mirror ????)
• /k/ is often realised as a fricative (dock ???
  week ????)
• /t/ can surface as t ?? ? ? ? ?
• Glottal stops are rare - particularly in
  word-final and intervocalic position

13
Lenition in Liverpool English

• voiceless stops sometimes lack complete closure
  in certain syllable-final environments, so that
  varieties of fricatives ? t? ? result for /p,
  t, k/ in such words as snake ?????, short
  ?????, and daughter ??????. (Wells 1982 371)

14
Lenition in Liverpool English

• A classic example of spirantisation can be found
  in the city of Liverpool, where the voiceless
  stops p, t, k have become the voiceless
  fricatives ? ? ? respectively, and the voiced
  stops b, d, g have become the voiced fricatives
  ?, ?, ? respectively, in non word-initial
  environments. (Radford et. al. 1999 121)

15
What is lenition?

• a cline of phonetic weakening (Hickey 1996
  182)
• a form of articulatory softening whereby a
  phonological stop is affricated or aspirated, or
  can be realised as a fricative (Sangster 1999
  1)
• a systematic reduction process, often resulting
  in deletion, which affects certain consonants
  depending on their position within the word or
  phonological phrase (Escure 1975 5)

16
Lenition trajectories
1
3
4
2

• Lass 1984 178

17
If these are lenitions, what is lenition?

• Kirchner (1998 2) says lenition has been
  largely ignored in the theoretical literature

18
If these are lenitions, what is lenition?

• articulatory phonology (Browman Goldstein 1989,
  1992, Hind 1996),
• acoustic phonetics (Lavoie 1996, 2000, 2001),
• dependency phonology (Anderson Jones 1974,
  Anderson Ewen 1987, Ewen 1995),
• government phonology (e.g. Kaye, Lowenstamm
  Vergnaud 1985, 1990, Harris 1990, 1994, Honeybone
  2001, 2002),
• optimality theory (Kirchner 1998, 2000).

19
Lenition in GP approaches

• Some background to elemental phonology
• Segments are made up of smaller privative
  elements (not e.g. binary distinctive features)
• The same elements can occur in consonants and
  vowels
• Elements have the potential to be independently
  interpretable

20
Some Elements

• palatality, coronality, dorsality
• occlusion, frication
• nasal
• spread, voice

21
Segments

• /t/ /s/ /h/
• coronality coronality
• occlusion frication
• spread spread spread

22
Lenition as element loss
Loss of occlusion
Loss of coronality
Loss of spread

• t ? ts ? s ? h

23
Lenition in Liverpool English

• Stops at all places of articulation are lenited
  to a certain extent
• The widest range of realisations are found for
  /t/ e.g. t, th, ts, ht, st, s, h, ?, ?
  (interestingly not ?)

/t/
/k/
24
Lenition as lax articulation

• Articulation is generally lax in Scouse this
  applies to the lower lip as well as the tongue
  and the active articulator exerts little pressure
  on the passive one. For stops, the pressure is
  often insufficient to make or maintain the
  closure, as that these consonants are often
  impressionistically fricatives or affricates
  (more precisely the cardinal categories of
  stop, fricative and affricate are inappropriate
  for the description of Scouse consonants.
• Knowles (1973 107)

25
Lenition as lax articulation

• The approximation of the articulators for the
  fricatives tends to be less close in Scouse
  than is usual in RP.
• Knowles (1973 107)

26
Questions

• What is lax articulation exactly?
• The articulatory target is complete closure
• Lenition is articulatory undershoot
• Random and unstructured
• Is the articulation of LE plosives random?

27
Lenition inhibition

• Lenition in LE isnt completely random
• Prosodically strong positions (utterance-
  initial, word-initial) inhibit lenition (e.g.
  Escure 1975)
• And segments that share elemental material are
  stronger than those that dont (Honeybone 2001)

28
Lenition inhibition

• cat lenition to stage 2 possible
• milk lenition to stage 2 possible
• halt lenition to stage 1 possible
• thank lenition to stage 1 possible
• school no lenition possible

29
Lenition inhibition

• For Honeybone (2001) the presence of lenition is
  governed by the phonological system. LE lenition
  belongs in /phonology/ and not in phonetics.
• However, in considering only the stage of
  lenition (e.g. affricate, fricative etc) the
  details of the phonetics are overlooked
• If the phonology of lenition is not random, is
  the phonetics?

30
Questions

• In a prosodically weak position, is any phonetic
  variation in the surface form random and
  unstructured?
• Or is there evidence for a particular target over
  another? That is, is the phonetic variation
  structured?

31
The Data

• 16 speakers (9 female and 7 male)
• Hangman elicitation task
• Stops in all places of articulation
• 2458 tokens

32
/g/ stage of lenition, male/female
33
/p/ - stage of lenition, male/female
34
The phonetics of /p/

• 3 attested realisations
• 1. stop closure and burst transient
• 2. stop closure and extended release
• 3. no stop closure

35
/p/ - burst aspirated release
lttapgt
ltharpgt
36
/p/ - stopless variant
ltpipgt
37
The phonetics of /p/
Burst
Ext release
Fricative
n
38
The phonetics of /p/

• Men lenite to Stage 2 more than women
• Variation within Stage 0 lenition

39
/k/

• Two attested realisations
• aspirated release
• fricative

40
/k/ - stage of lenition by speaker
41
/k/ - phonetic variants

• Fricative
• palatal, fronted velar, velar, back velar,
  uvular.
• Straight-forward case of place assimilation to
  the vowel?

42
/k/ - variant by gender

• Assimilation to the vowel attested in male speech
  but not in females
• Not a straightforward case of assimilation but
  also an indexical marker

43
/t/

• Stages of lenition for /t/ t, ts, s, h

44
/t/ - stage 2 lenition by speaker
45
/t/ - phonetic variants

• ts at varying degrees of oral approximation

46
/t/ - phonetic variants

• Aspirated and affricated variants
• Pre-aspirated/affricated variants

47
/t/ - phonetic variants
48
/t/ - phonetic variants

• controlled sibilants

hat M1
49
/t/ - phonetic variants

• dynamic sibilants

out F10
50
/t/ - variants by gender
51
Whats going on?

• Its a fricative (i.e. a Stage 2 lenition)
• but speakers are aiming for a particular kind of
  fricative
• A particular phonetic realisation is the target
  for these speakers

52
Articulatory Gestures
53
Articulatory Gestures
54
Summary

• Men and women speakers might reach the same stage
  of lenition (that is, have the same phonological
  structure)
• But the phonetic realisation of that stage (the
  gestural organisation) is often different
• Individual speakers are also consistent in their
  choice of low-level phonetic realisation

55
Summary

• Solenition in Liverpool English can not be
  explained sufficiently in terms of the nature of
  the sugsegmental elements
• The surface forms vary in a structured way, yet
  have the same elemental makeup
• The timing of the elements would have to be
  handled in the phonetic component of the
  grammar

56
Summary

• Butlenition in Liverpool English can no longer
  be just because of lax articulation
• Lenited forms occur too frequently and
  consistently
• They are indexical markers
• They cannot be articulatory by-products or
  articulatory undershoot

57
So what is it then?

• Neither? Or both?

58
Marrying the etics and -ology

• We need both articulatory gestures and underlying
  subsegmental material
• How do we get from one to the other?

59
A usage based model of phonology

• Gestures are pre-linguistic units which become
  units of linguistic contrast (babbling to speech,
  see Vihman Croft 2005 for overview)
• Browman Goldstein (1989 202) the emergence
  of the linguistic significance of gestures can be
  considered a function of the particular language
  environment in which the child finds itself
• The variability in the speech signal
  necessitates, and creates, abstraction
  (Pierrehumbert, Beckman Ladd 2000 292)

60
Begin to separate out the contrastive information
from the indexical
Recognition of elements for contrast
Phoneme?
Statistical recognition of similarities between
phonetic events
Indexical linguistic information at the same
time
Word template
61
Concluding remarks

• Fine grained phonetic variation must be included
  in a model of phonology
• It is insufficient to ship it of to the phonetic
  component of the grammar
• A model of phonology which incorporates indexical
  information and lexical contrast information is
  necessary

62
Concluding remarks

• A cognitive, usage based model of phonology is a
  step in that direction

63
Concluding remarks
Competence

• Phonetics

Phonology
Sociolinguistics