Spoken Syntax: The phonetics of giving a hand in New Zealand English - PowerPoint PPT Presentation

Spoken Syntax: The phonetics of giving a hand in New Zealand English (2006) Jennifer Hay and Joan Bresnan

Exemplar Theory in Syntax No explicit grammar rules → generalize over past experiences see constraint-based grammars (HPSG, LFG...) or Cognitive Grammar probabilistic grammars: stat. learning of abstract structures

Le bus arrivait à 7h. (Imp.) → The bus always arrived at 7. Hier, le bus est arrivé à 7:10. (Perf.) → Yesterday, the bus arrived at 7:10. completed vs. habitual/setting the scene

General Metaphor Solid vs. Fluid Please come in. (Imp.)

Stored Phrases Natives fixate on idioms shorter early linguistic representations highly concrete

Exemplar Theory in Phonetics

Lexical diffusion: frequent words lead sound change

Perception affected by speaker's perceived gender, age, class, dialect actual overlap due to dialect, speaker variation, random variation in production: systematic bias (Lindblom 1984)

Systematic bias More variation with increased usage → approaches Gaussian distribution BUT phonetic variability decreases, e.g., up through late childhood → entrenchment: averaging over exemplars

Syntax + Phonetics Combined Syntactic variables associated with social groups liaison loss in infrequent phrases in French palatalization at probable word boundaries did you vs. good you

What is old and modern NZE?

Australia colonized in 1788 (New South Wales) NZ first mentioned in Murders Abroad Act 1817 colony by 1841 113,000 settlers 1846 Australian and NZE close to British

1887: McBurney remarks tendency to Cockney and variability in different towns (others deny variability) From 1880: annual school inspector reports no 'provincialisms' unitl 1900 The Triad: 'genteel' pronunciation ('may' for 'my', 'bay' for 'by', 'cray' for 'cry') 1936: The Mother Tongue in NZ , Arnold Wall

The Triad, 1910: 'there is nothing to distinguish their speech from that of a highly cultured Englishman in England... I am merely just now observing that a dialect, and that not a defensible one, is gradually becoming fixed in the Dominion among the children and younger adults'

Mobile Unit: three tours 1946-1948 Origins of NZE project (ONZE) 1996 couple hundred speakers still being analyzed

NZE: Its Origins and Evolution Maori influence on NZE purely lexical though there's Maori English/'bro talk'; probably very recent all essential NZE traits from BE (ONZE) NZE homogenous (except Southland burr, maybe Scottish)

TRAP vowel NZE uses raised [ε] raising already an RP feature /æ/ has been lowered in RP for at least 40 years, away from Cockney association

Tendencies Drεss → dress kɪt → kɘt 'fush and chups' recent (only few early tokens) START → vowel fronted TRAP-BATH split STRUT → open and relatively fronted

FLEECE → [ ə i] GOOSE → central or slightly fronted [ʉ] or [ ə ʉ] back variant before dark /l/ LOT → rounded nowadays, above [ɔ] THOUGHT, NORTH, FORCE, (POOR) → /ɔ:/ FOOT → raised and fronted to raised [ɘ] generally no FOOT-GOOSE merger NEAR-SQUARE merger

ex ungue leonem

Syntax: grammar as an analogical generalization over stored memories of phrases Phonetics: lexical items as distributions of stored memories with phonetic detail

/æ/ raising more advanced when referring to limbs than in 'give/lend a hand' /ɪ/ centralization more advanced in 'give a chance' (abstract) than 'give a pen' → phrases maybe stored with phonetic detail

hand 59 speakers born 1857-1900 → 5579 tokens of /æ/, 3284 raised strong lexical frequency effect in logistic regression model 92 'hand' tokens

rasing more likely in 'hand' tokens

limb: washed his hand, put one hand up, wash by hand give: give a hand, try one's hand, turn one's hand (to), have a hand in other/figurative: left-hand turn, in good hands, on the other hand, close at hand

33% 90% 76%

Chain Shift /æ/ raising → /ɛ/ raising → /ɪ/ centralization (1900-1930 most radical) frɛnd → fri:nd /i/ diphthongization

give speakers born 1896-1931 53 tokens of 'give' /ɪ/ centralization as [ɪ ̙ ] or [ɨ] 3886 tokens of /ɪ/ → lexical frequency effect

give DA-transfer: give presents, give a plate of food DA-abstract: give the horses a spell, give us the strap other, i.e., passives, preposed forms, phrasal verbs, recipient implied a.o.: given licorice, what our parents could afford to give us, had to give it away, give it up

Tendencies abstract give → double obj: I gave him the idea NOT I gave the idea to him centralization most likely with dative alternation with abstract semantics (give horse a spell, give the strap) and with later-born speakers 70% abstract vs. 15% transfer

Discussion give a hand not advanced advanced more advanced variant = frequent contexts? Frequency of abstract vs. transfer meaning? Or average token frequency per type?

give a hand 75x ← leading sound change give a call 25x relevant as a cognitive category give a watch 10x give a present 10x

US and NZE corpus 2794 dative-alternation 'give' tokens → 60% abstract → most frequent and most advanced in sound change individual token frequencies of abstract types greater than of transfer types 73 vs. 28 that appeared at least 5 times (chance, type, right vs. money, dollars, one)

hand not nearly as much data 54% of tokens refer to limbs → most frequent and most advanced

Guesswork give a hand > hold out your hand, her hand was cold, her hand is bigger than her face... phrases with limb meaning less frequent Why is /æ/ more raised for limbs?

'give' as a verb and 'hand' as a noun verb + object > subject + verb → stored individually 'take my hand' frequent enough to be stored → slower sound change

give usually has a theme give + theme: quite restricted 'give + x' should be frequent → abstract ones more frequent than transfer types → abstract ones more advanced

So, 'give a hand' 'giving a hand' > 'give x' 'give a hand' < 'hand' → storage of context info leads to different pace

! We want to find out if phrases are stored → look at correlation between phrasal frequency and sound change BUT you can't calculate lexical frequency without knowing which phrases are stored phrases could be stored on word-level and activate through context spreading → syntactic + semantic + context/social info

(First) Alternative different representations of same lexical item hand: 'limb' entry 'helping out' entry give: 'possession transfer' entry 'abstract' entry → a lot of syntactic analysis in lexical retrieval

Third option some factor correlates with meaning and advancement in sound change e.g. 'limb' more frequent in focus positions than 'non-limb' e.g. words with pitch accent more variable

Against Lexical Diffusion /t/-deletion as a result of contextual support and frequency? → would be more frequent where more predictable but /æ/ not reductive 'give' reductive?

centralization hints at reduction low semantic load: 'give a hand' vs. 'give a chance' abstract 'give': affect s.o. 'give an apple/towel/pen' → stable meaning

reduction likely with abstract 'give'? low semantic load → stylistic use, extreme phonetic variants? doesn't work for 'hand' → 'hand' and 'give' different processes?

Wrap-up lexical diffusion exists → also evidence for advancement differences between syntactic items → what are those items and are they stored? sound change of phrases not predicted by exemplar theories of syntax or phonetics alone

Outlook whether phrase storage is phon. detailed or not → either lex. and phrase storage are different → or lexical access has to account for syntax 'give a lecture' vs. 'give a damn' 'sheep' sg and pl stored as one lex. item? grammaticality depends on speakers?