Alignment allophony and European pitch accent systems

1
Alignment allophony and European pitch accent
systems

• Bob Ladd
• Edinburgh University

2
European word pitch systems

• Two lexically distinct phonetic treatments of
  stressed syllables
• Usually restricted to long/heavy syllables (esp.
  two-sonorant-mora syllables)
• Low functional load / high predictability

3
European word pitch systems

• Scandinavian
• Central Franconian
• Lithuanian
• South Slavic
• Scots Gaelic (?)
• Etc ?

4
Gaelic word-pitch contrasts

• Contrast between underlying long vowel and
  sequence of vowels in hiatus
• VV (lt VCV)
• is phonetically distinct from
• VV ( V)

5
Realisation of Gaelic contrasts (Northern
Hebrides)

• long-vowel cases realised with sustained or
  rising pitch vowels-in-hiatus cases realised
  with clear F0 fall
• bodha po underwater rock vs. bó po
  cow

6
Why not a contact phenomenon?

• Most languages with word pitch systems are found
  in isolated areas
• Most are closely related to other languages or
  lg. varieties that dont have such systems
• In contact situations, word pitch systems tend to
  disappear (e.g. Swedish in Finland)

7
Typical historical triggers

• Apocope
• e.g. Limburg Dutch
• 1dax lt MHG daV days
• vs. 2dax ltMHG dax day
• Loss of intervocalic consonant
• e.g. Scots Gaelic

8
Phonologisation accounts

• Frings 1916 allophonic duration differences
  between mono- and disyllables
• Öhman 1967 allophonic pitch contour differences
  between mono-and disyllables
• Etc.

9
Central Franconian developments (Gussenhoven 2000)

• Sing.
  Pl.
• MHG dax da??
• OSL dax da???
• Apocope dax da?x
• Fake AL ?dax da?x
• Reinterpretation dax da?x
• H

10
Alignment allophony

• Dutch accentual H is aligned earlier with
  phonologically long vowels than short vowels,
  irrespective of actual duration (Ladd et al.
  2000)
• English, Spanish, Dutch nuclear H is aligned
  earlier than prenuclear H (Steele 1986 Nibert
  2000 Schepman et al. forthcoming).
• Greek nuclear L is aligned earlier in
  phrase-final syllable than elsewhere (Arvaniti et
  al., submitted)
• English nuclear H is aligned earlier in
  one-accent phrase than in two-accent phrase (Ladd
  et al., work in progress)

11
German syllabic nasals

• Common for /-?n/ sequences to be realised as
  -?? after most consonants, assimilated in place
  to preceding consonant. E.g.
• /k????/ realised as ?????
• /?????/ realised as ?????
• What do such transcriptions actually mean?

12
German test pairs

• Das Kind ist auf den grossen Stein geklettert
• The child climbed onto the big rock
• Das Kind hat auf den grossen Steinen gesessen
• The child sat on the big rocks
• Das kann sie nicht She cant do that
• Das können sie nicht They cant do that

13
Secondary splits

• German umlaut
• CuCi gt CyCi gt CyC?
• CuCa CuCa gt CuC?

14
Priming effect (Kiparsky 1995)

• the textbooks draw a discreet veil over the
  other cases where the redundant feature simply
  disappears when the triggering environment is
  lost
• the merger of voiced and voiceless consonants
  normally leaves a tone/register distinction only
  in languages which already possess a tone system
  Svantesson 1989

15
If phonologisation is a two-stage process

• No contradiction between my proposal and e.g.
  Gussenhovens both necessary
• Explanation for relatively marginal status of
  word pitch distinctions in Europe
• And yet

16
European sources of alignment allophony

• Common for European languages to have
• Distinctive vowel and/or consonant quantity
• Complex phonotactics (allowing codas and
  consonant clusters)
• Strong dynamic stress
• All are likely to affect alignment and/or to lead
  to sound changes that may affect conditioning
  factors.

17
Realisation of Scandinavian contrasts

• EITHER Accent 2 marked by pitch feature
  (prelinked tone?) (Swedish)
• OR Accent 1 marked by glottal feature (Danish)

18
Realisation of Gaelic contrasts (Southern
Hebrides)

• vowels-in-hiatus cases realised with some sort of
  glottal feature long-vowel cases realised with
  normal phonation
• bodha po/? vs. bó po

19
Central Franconian (Gussenhoven and others)

• Limburg has clearly tonal realisation in which
  Accent 2 cases are marked.
• Cologne has duration involving duration and
  intensity effects in which Accent 1 cases are
  marked.

20
Summary of similarities

• EITHER Accent 2 cases realised by pitch feature
  (Swedish, Northern Hebrides, Limburg)
• OR Accent 1 cases realised by some sort of
  glottal feature (Danish, Southern Hebrides,
  Cologne)

21
Explanation for complementary realisations of
lexical accent

• If its based on alignment allophony, and
• If its hard to phonologise in a European
  language, then
• An obvious way to fit it into the phonology is to
  mark either the early alignment cases (Danish,
  Southern Hebrides, Cologne) or the late alignment
  cases (Swedish, Northern Hebrides, Limburg).
• Note that in Lithuanian and S. Slavic the early
  alignment cases are marked.