The prosodic word and accent patterns in Swedish
Tomas Riad, Stockholm UniversityWorkshop on the Prosodic Hierarchy,
September 5, 2008University of Tromsø
Distribution of accent varies in interesting ways
•
Conditions: –
Room
–
Locality–
Initiality
–
Presence of stress–
Presence of morphological information
–
or prosodic information
•
What is the relevance of Pω
structure?
Empirical area
•
Interaction of Lexical accent and Auftakt (anacrusis) exposes several properties.
•
The pattern is interesting in itself.
•
Looking at Pω
structure may give us hints regarding the relationship between lexical and postlexical accent.
Prosodic words in Swedish
•
Culminativity:–
One phonological stress within the minimal Pω
–
One main-stress within the maximal Pω
•
Rhythmic grouping:–
One can tell trochaic and iambic rhythm apart(ˈtetra-)(ˌvinet) vs
(ˈbäck-)(raˌvinen)
.] [x [. x’the tetrapak wine’
’the brook ravine’
Prosodic words in Swedish
•
Syllabification:–
Minimal Pω
is the domain for syllabification
•
Accent:–
accent 2 assignment would also seem to take place within the Pω
–
lexically and post-lexically•
But how?
Privative distinction•
lexical tone
+ prominence tone
= accent 2
•
prominence tone
+ Ø
= accent 1
Accent 1 Accent 2
ˈs y r a k ˈe l a k ǀ ǀ L*H L% H*LoHL%
•
Lahiri et al. (2005): accent 1 is specified•
Morén (2006): neither accent is specified
•
Riad (1996 et seq.): accent 2 is specified•
Bruce (1977 et seq.): both are specified, but 2 is marked.
•
I will look for support that accent 2 is morphologically specified, mostly in suffixes.
Plan•
Interaction of lexical accent 2 (specified) and Auftakt/anacrusis (initiality challenged) in lexical accent assignment.
•
Support for the lexical specification of tone in suffixes.
•
Prosodic structure of words.•
Integrate the analysis of accent variation with Pω-structure.
Simplex polysyllabic forms
Accent 2 Accent 1σˈσ Ø ka1ˈmel, gar1ˈdin
ˈσσ 2ˈsommar, 2ˈbyx-or, 2ˈtvätt-a
1ˈtango, 1ˈfänrik, 1ˈfeg-is
ˈσσσ 2ˈlöp-are, 2ˈglad-are 1ˈsyfilis, 1ˈKanada
σˈσσ ve2ˈrand-a, go2ˈrill-a bog1ˈser-a, bo1ˈlero, be-1ˈtal-a
σσˈσσ kara2ˈok-e, pape2ˈgoj-a mili1ˈtär-er,
σσˈσσσ speci2ˈal-are speci1ˈell-are
Patterns
•
Most syllabic suffixes correlate with accent 2 in canonical contexts (ˈσ-σ); a few don’t (-is)
•
It’s mostly suffixes that carry lexical accent information (and that are stable across dialects, Bruce 1998): -are2
, -a2
, -e2
, -ning2
, -ig2
, -nad2
, -ar2
, -or2
, - ande2
, …
Patterns
•
Unsuffixed Auftakt forms get accent 1mi1ˈnister, bo1ˈlero, an1ˈsjovis
•
Suffixed Auftakt forms variably get accent 1 or 2 bog1ˈser-a,
go2ˈrill-a
•
There’s a distinction between suffixes vis- à-vis Auftakt.
Central Swedish ”strong acc 2 suffixes”
-areN
, -aN
, -eN
,-orPl
, -orN
, -skaN
, -igA
Internal Prosodic Auftakt
(ana- crusis)
”weak acc 2 suffixes”
-ingN
, -andeVAN
, -arPl
, -aV
, -arV, -ligA
, -arecomp
, -astsup
a ˈglad-are ‘happier’ 2
b speciˈell-are ‘more special’
1 2
c ˈlöp-are ’runner’ 2
d speciˈal-are ‘a special’ 2 1
e bogs-ˈer-a ‘to tow’ 1 2
f bogs-ˈer-ing ‘towing’ 1 2
g bogs-ˈer-are ‘tower’ 2 1
”Strong”
>> Auftakt >> ”Weak”
”Weak”
and ”strong”
asymmetry correlates roughly with ”verbal”
and ”nominal”
(cf.
Smith 2001, Kabak & Plank ms. 2007)
”verbal” ”nominal”2ˈtal-a2
2ˈtal-are2
bog1ˈser-a2
bog2ˈser-are2
han1ˈter-ande2
han2ˈter-are2
han1ˈter-ing2
kara2ˈok-e2
dis1ˈträ-are2
ve2ˈrand-a2
pru1ˈdent-lig2
ge2ˈlé-ig2
”strong acc 2 suffixes”
Internal Prosodic Auftakt
”weak acc 2 suffixes”
Central Swedish -areN
, -skaN
, -eN
, -aN
, -orPl
, -orN
, -igA
-ingN
, -andeVAN
, -arPl
, -aV
, -arV, -ligA
-arecomp
, -astsup
North Swedish -areN
, -skaN
, -eN:Lat
-ingN
, -andeVAN
, -arPl
, -aV
, -arV, -ligA
-arecomp
, -astsup
-aN
, -orPl
, -orN
, -igA
North Swedish
ve1ˈrand-ago1ˈrill-orpro1ˈfessorge1ˈlé-ig
re2ˈal-e kara1ˈok-e
So
•
Internal prosodic Auftakt, i.e. initiality of stress or not interacts with
•
lexical specification of tone (=morphology) in determining accent in simplex forms.
•
Unstressed prefixes be- and för- represent External Auftakt.
Unstressed prefixes be- and för-Central Swedish
External Prosodic Auftakt
Incorporation of be- & för-
”strong acc 2 suffixes”-areN
, -aN
, -eN
, -orPl
,-orN
, -skaN
, -igA
InternalProsodic Auftakt (ana-
crusis)
”weak acc 2 suffixes”
-ingN
, -andeVAN
, -arPl
, -aV
, -arV, -ligA-arecomp
, -astsup
a be-ˈtal-a ‘pay’ 1 2
b för-ˈräd-are ‘traitor’
1 2
c för-ˈakt-lig ’despicable’
1 2
d be-ˈständ-ig ‘lasting’
1 2
•
Internal
Auftakt and External
Auftakt interact differently with accent-2-inducing suffixes.
•
External
Auftakt always inhibits accent 2 in simplex forms. …hm…
•
The presence of two stresses always yields accent 2, however.
Two stresses yield post-lexical accent 2
Central Swedish
Two stress
es yield
post- lexical
acc 2
External Prosodi
c Auftakt.Incorpora
tion of be- & för-
”strong acc 2 suffixes”-areN
, -aN
, - eN
, -orPl
,-orN
, -skaN
, -igA
Inter nal
Pros odic
Aufta kt
”weak acc 2 suffixes”
-ingN
, -andeVAN
, -arPl
, -aV
, -arV
,-ligA
-arecomp
, -astsup
e ˈmellan-ˌmål ’snack’
2
f ˈmellan-ˌmåls- ˌbull-e
‘snack bun’
2 2
g be-ˈtal-kaˌnal ‘pay channel’
2 1
So, compounds will always get accent 2 in this dialect
Properties•
Lexical tone must associate to a stressed syllable (premise)
Initiality•
Post-lexical accent 2 targets the first
stress
with the word tone (always) •
We get lexical accent 2 canonically when stress is at the left edge.
•
In SSw compounds (Skåne) initiality matters for post-lexical accent, too.
Kristoffersen (2000, 257)
•
Tonal Foot is a trochee in which the lexical tone must be initial.
•
≠
Metrical Foot•
The TF occurs at some edge of a Pω
•
This captures the fact that accent 2 occurs in the initial or the penultimate syllable.(not quite surface true for Sw: pape2ˈgoj-or2
-na)
LocalityLocal
Non-local
2ˈnunn-a2
’nun/s’
1ˈoper-a2
’opera/s’2ˈnunn-or2
1ˈoper-or21ˈbil ’car/s’
1ˈkaktus ’cactus/es’
2ˈbil-ar2
1ˈkaktus-ar2
In SSw compounds (Skåne) locality matters for post-lexical accent, too, in the form of a stress clash requirement.
go2ˈrill-a
ω
2ˈlär-are
ω
ω ω ω
ω
2ˈmellan-ˌmåls-ˌbulle
be-2ˈtal-kaˈnal
ω ω
ω
ω
σ
σ för-1ˈtal-a
ω
ω
bo1ˈlero
ω
Prosodic wordssimplex
compound
?
’to slander’ ’snack bun’
’pay channel’
’teacher’
The Pω
is recursive (Ito & Mester 2006, Kabak & Revithiadou 2007)
Non-isomorphism: Prosodic structure often doesn’t reflect
morphological structure be-jaka ‘acknowledge’ jaka ‘say yea’
be-frynda ‘make friends’ †frynda be-segla ‘seal’ †segel ‘seal’ be-tala ‘pay’ tala ‘speak’ för-leda ‘lead astray’ leda ‘lead’ för-gråa ’become grey’ grå ’grey’ för-spröda ’become brittle’ spröd ’brittle’ för-gubbning ’geezering’ gubbe ’geezer’
(be-(tal-a)ω)ω’
(för-(gubb-ning)ω)ω’ Evidence of some productivity
Non-isomorphism
Morphologically left-branching:[varg-skinn-s]-mössa ‘wolf skin hat’
Morphologically right-branching:Barn-[skinn-mössa] ‘children’s skin/leather hat’
Prosodically identical:2((Hˈvarg)(ˌskinns)(LHˌmössa)ω)ω2((Hˈbarn)(ˌskinn)(LHˌmössa)ω)ω
Non-isomorphism
Also true of phrases included in compounds (more later)
Vi ska gå
i första ((maj)-(tåget)ω)ω[ ]NP
… and in syntax/prosody: embedded/flat(This is the cát)(that chased the rát)(that ate the mált)[ [ [ [ [ [ ]]]]]]
Diagnosing the minimal PωSyllabification appears to be the indicator of
minimal Pω
in ordinary compounding in Gmc. (cf. Hall and Kleinhenz 1999), etc.
•
Onset challenge
(ˈhärm)ω(ˌapa)ω härm.a.pa *här.ma.pa compound (favoˈrit)ω(antipaˌti)ω fa.vo.rit.an.ti.pa.ti *fa.vo.ri.tan.ti.pa.ti (ˈBred)ω(ˌäng)ω bred.äng *bre.däng (ˈsokrates)ω(ˌanda)ω so.kra.tes.an.da *so.kra.te.san.da (ˈvan)ω(ˌartad)ω van.a.ʈad *va.na.ʈad derivation (ˈfuling)ω *ful.ing fu.ling (ˈhärmande)ω *härm.an.de här.man.de derivation/inflection (ˈgrinar)ω *grin.ar gri.nar inflection
Diagnosing PωBehaviour of (unstressed) ”strong”
and
”weak”
suffixes is identical regarding syllabification
They are all internal to Pωmin
(2ˈgla.d-a.re)ω
(2ˈba.g-a.re)ω’happier’
’baker’,
(spe.ci.1ˈel.l-a.re)ω
vs. (spe.ci.2ˈa.l-a.re)ω’more special’
’a special’
speciˈalarespeciˈellare
ω
Initiality: distance•
Pattern dependent on suffix
rather than on
prosodic form lexical encoding of tones
(for1ˈmell-are)ω
vs. (lun2ˈdell-are)ω’more formal’, ’song by Lundell’
(speci1ˈell-are)ω
vs. (speci2ˈal-are)ω’more special’, ’a special’
(traditio1ˈnell-are)ω
vs (instrumen2ˈtal-are)ω’more traditional’, ’instrumental piece of music’
•
Measured distance seems unimportant
Diagnosing Pω’
Incorporation in simplex morphological context
(be-1(ˈtala2
)ω
)ω’
(för-1(ˈära2
)ω
)ω’
σ be- tala
ω
ω’
Prefixes be- and för- are not
part of the minimal Pω
Syllabification in ω
and ω’Onset challenge again:för-.ä.r-a rather than *fö.r-är.-a
’bestow’
or *fö.r-ä.r-athus:(för.-(ä.r-a)ω
)ω’
rather than *(för.-(är.)ω
-a)ω’(be.-(ta.l-a)ω
)ω’
’pay’
or *(fö.r-ä.r-a)ω
Not so for ha.v-an.de ’pregnant’, hå.r-ig ’hairy’
The test is somewhat unstraightforward, but it works in slow motion.
Initiality: barrier
Barrier: Ø
vs. (ω
(bog2ˈser-are2
)ω
vs.
(be-1(ˈtal-are2
)ω
)ω’
’tower’
’payer’(mo2ˈras-ig2
)ω
vs.
(be-1(ˈhör-ig2
)ω
)ω’
’”morassy”’
’eligible’(ge2ˈlé-ig2
)ω
vs.
(för-1(ˈsikt-ig2
)ω
)ω’
’jellyish’
’careful’
Evidence for incorporation as such: 1. Simplex incorporation in phrasal context
(många2
)
a. Rostbiff och lax – det är (för 1(många2
)ω
)ω’läckerheter’roast beef and salmon –
that’s too many
delicacies’b. Rostbiff och lax – det är för 2(många2
)ωläckerheter’roast beef and salmon –
that is delicacies for
many
(people)’
Example from Malmgren (1992)
2. Incorporation is morpheme dependentbe-, för-, but not des- or kon-de.sin.fi.ce.ra ;
des-infi2cer-are2
de.sin.te.gre.ra ;
des-inte2grer-are2
kon-1ter-a2
~kon-2ter-are2
för ’too’, but not för ’for, to’
2. Incorporation is morpheme dependent
Some sensitivity to the head word in phrasal incorporation (Sara Myrberg, p.c.):
för 1liten2
, för 1stora2
, för 1vanlig2
’ordinary’but
för 2/1egen2
’odd’, för 2/1dyig2
’muddy’
för 2ˈfårˌaktig ’too sheepish’, för 2ˈhemˌtrevlig ’too cosy’
(*för 1ˈfårˌaktig)
So no incorporation into compounds
3. Norwegian ”stress retraction”……is also sensitive to the difference between ω
and ω’.
Kristoffersen (2000, 165)protes1ˈtere > 2ˈprotesˌterebekka1ˈsin > 2ˈbekkaˌsin ’snipe’selek1ˈsjon > 2ˈselekˌsjonbe1ˈtong > 2ˈbeˌtong ’concrete’be-1ˈtone > 2ˈbe-ˌtone ’stress’
This phenomenon is better thought of as (pseudo-) compound formation.
3. Norwegian ”stress retraction”…More general versions of this phenomenon
occur, e.g. Ärtemark, Dalsland (Noreen 1915, 39):2ˈbe-ˌsked ’word’
be-1ˈdrövlig ’terrible’
2ˈge-ˌmen ’simple’
för-1ˈstå ’understand’2ˈKatˌrina, 2ˈfranˌsoser ’venereal disease’, etc.
Cf. alsopoˈlice > ˈpoˌliceDeˈtroit > ˈDeˌtroit
So,
•
This motivates a structure where be- and för- are differently connected to the Pω, than internal Auftakt.
•
How now is the structure related to accent?
Lahiri et al. (2005), Wetterlin (2007): Accent 1 is specified
…and accent 2 is default.A central argument for them is precisely the
behaviour of be- and för-words: If 2tal-a has accent 2, why doesn’t be-1tal-a have accent 2?
Their answer: lexical specification of be-1
, för-1
.and lexical specification always dominates.
Structure-based analysis•
I have suggested a difference in prosodic structure.
•
Now we must make that structure responsible for the accent pattern, rather than the prefixes themselves.
•
NB. There are no equivalent prefixes that induce/admit accent 2, so we might as well get the accent behaviour from structure rather than specification.
Structure-based analysis•
Prima facie problem (for this perspective): (be-1(tal-a2
)), (för-1(gör-are2
)), etc. seem to fulfill the central criteria: stress is initial in ω, and locality is met.
•
Yet, neither strong nor weak suffixes appear to be able to realize their lexical accent….hm…
•
We return to the issue of accent in betala after discussing compounds:
•
Unstressed be- and för- can incorporate into compounds as stressed.
4. Prosodic incorporation of be- and för- in compounds
2ˈför-ˌan-ˌleda ’occasion, cause’2ˈför-ˌor-ˌsaka ’cause’2ˈför-ˌo-ˌrätta ‘cause injustice to; insult’2ˈför-ˌöd-ˌmjuka ’humiliate’2ˈför-ˌall-ˌmänliga ’make public’
(~0för-ˈall-ˌmänliga)
2ˈbe-ˌled-ˌsaga ’accompany’
(~0be-ˈled-ˌsaga)2ˈbe-ˌar-ˌbeta ’work, tr.’
Meanings are often causativeThe pattern is marginal, and doesn’t take place in phrases
(*ˈför-ˌfår-ˌaktig ’too sheepish’)
Incorporation into compounds only happens with full Pω
((ˈöd)ω
(ˈmjuka)ω
)ωmax
0för
X
((ˈöd)ω
(ˈmjuka)ω
)ωmax
unstressed prefix
(ˈför)ω
((ˈöd)ω
(ˈmjuka)ω
)ωmax create prosodic word
2((ˈför)ω
(ˈöd)ω
(ˈmjuka)ω
)ωmax incorporate
2((ˈbe)ω
(ˈled)ω
(ˈsaga)ω
)ωmax
Exhaustive parsing•
Incorporation of the unstressed prefixes into compounds indicates that compounds are exhaustively
parsed into lower projections of the
Pω.
( ω ω ω )ωmax
((för)(an)(leda))
( ω ω ω ω’ )ωmax
((hem)(kun)(skaps)(be(tyg)))
•
Cf. phrasal layering forcing ω-status onto function words phrase-finally (Selkirk 1996, Itô & Mester 2007, 200ff)
(He wanted [tu]ω
/*[tə])Φ
but he couldn’t(What did you look [æt]ω
/*[ət])Φ
yesterday
•
Pω
incorporation into compounds also takes place in phrasal prosody (in some dialects), in so-called particle verbs.
Particle verbs in Nw and NSw
2ˈkomme + 1ˈover > 0komm1ˈover or [2ˈkomˌmover]2ˈkomme + 2ˈetter > 0komm2ˈetter or [2ˈkomˌmetter]1ˈfinner + 1ˈut > 0finner1ˈut or [2ˈfinneˌrut]
Kristoffersen (2000, 288)
jag ska 2((ˈslå)ω(ˌin)ω)ω an åt de2((ˈskick)ω(ˌå)ω)ω den 2((ˈfå)ω(ˌut)ω)ω den
Görel Sandström (p.c.)
0gå ˈut
Ф
ω ω>
ω ω
ˈgå ˌut
ω
Compounds
•
Compounds and similar structures are Pωmax
2((ˈhärm)ω
(ˌapa)ω
)ωmax
’copy cat’2((ˈmellan)ω
(ˌmåls)ω
(ˌbulle)ω
)ωmax
’snack bun’
2((ˈvan)ω
(ˌartad)ω
)ωmax
’delinquent’
Compounds
CSw Compounds get accent 2 post-lexically –
two stresses accent 2
–
pitch events at first and last stress–
Lexical tonal information is superseded
–
Morphological composition is superseded
Compounds
ˈtull+ˈpack+ˈhus+ˈkarl-s+ˈlag
ω ω ω ω ω
ω
H* -----------------------------------L*HL%
H* ------
L*HL%
be-ˈtal-kaˈnal
ω ω
ω’
ωmax
σ
ˈbe-ˈled-ˈsaga
ω ω ω
ωmax
H* ---------
L*HL%
The prosodic shape of compounds indicates a simple prosodic structure.
ˈtull+ˈpack+ˈhus+ˈkarl-s+ˈlag
ω ω ω ω ω
ω
’customs packing house men’s (work)team’
H* -----------------------------------L*HL%
Stresses associated with pitch accents are prominent, whereas intervening stresses are not.
Unlike Danish, English …
Danish, mainstress[ˈarmˌbånds]ˌur ’wrist watch’
but(ˌ)stats-[ˈbolig-ˌfonden] ‘state dwelling trust’
(ˌ)kalve-[ˈnyre-ˌsteg] ‘veal kidney steak’
English, main, secondary, tertiary…
stress[[ˈLaw deˌgree] [ˈˈlanguage re0
quirement]]
… and some tonal dialects
Sunnmøre (Abrahamsen 2003: 193ff.), prominence tone.ˈdame-[Hˌhår-ˌpynt] ‘hair adornment for
women’[ˈdame-ˌhår]-Hˌpynt ‘adornment made of
women’s hair’ˈherre-[Hˌpels-ˌhuve] ‘fur hat for men’[ˈherre-ˌpels]-Hˌhuve ‘hat made of man fur’
… and some tonal dialects
Malmö
(Bruce 1974, Riad 2003) accent2[ˈmaskˌros]ˌbrand ‘dandelion fire’1ˈklot-[ˌarm-ˌband]-
‘ball bracelet’
ω ω ω
ω
2ˈmask-ˌros-ˌbrand
ω
ω ω ω
ω
1ˈklot-ˌarm-ˌband
ω
Compounds with lexicalized phrases•
Compounds seem not to prosodically incorporate larger units, although the morphology does.
ω ω
Där går första-maj-tåget
ωmax
’There is the 1st of May procession’
Cpd and lexicalized phrases
Data from Masja Koptjevskaja-Tamm (p.c.)ˈAstrid ’A.’0Astrid ˈLindgren ’A.L.’0Astrid ˈLindgren-ˈböcker ’A.L. books’
0Gustav ˈAdolf-ˈdagen ’the G.A. day’0Drottning ˈSilvia-ˈbrytningen ’Queen S. accent’0Helan och ˈHalvan-parˈtiet ’The Laurel and Hardy party’0Romeo och ˈJulia-förˈälskelsen ’The R. and J. love story’0duktiga ˈAnnika-synˈdromet ’The good Annika syndrome’0Svarte ˈPetter-ˈrollen ’The Black Peter/sheep role’min Me0nuhin och Grapˈpelli-ˈskiva ’my M. and G. record’
lexicalized phrases
0första ˈmaj-ˈtåget’the 1st of May procession’
my I-can’t-be0lieve-it’s-not-ˈbutter ˈsandwich
Putative structures((0första ˈmaj)ω
(ˈtåget)ω
)ωmax
?(0första (ˈmaj)ω
(ˈtåget)ω
)ωmax0första ((ˈmaj)ω
(ˈtåget)ω
)ωmax*
lexicalized phrases
0första ((ˈmaj)ω
(ˈtåget)ω
)ωmax
Test with derivational endings ”strong”
and ”weak” which are demonstrably sensitive to Auftakt.
Hon är en riktig första 2(maj-are2
)ω
”strong”’She is a real 1st of May-person’
Vi ska första 2(maj-a2
)ω
hela dagen. ”weak”’We’re going to (do) 1st of May (things) all day’
lexicalized phrases
•
By transitivity, the same structure should apply in compounds.
0första ((ˈmaj)ω
(ˈtåget)ω
)ωmax0Helan och ((ˈHalvan-)ω
(parˈtiet)ω
)ωmax
min Me0nuhin och ((Grapˈpelli-)ω
(ˈskiva)ω
)ωmax
Summing up•
Lexical accent assignment is sensitive to a distance to the left edge for the ”weak”
accent-2-
inducing suffixes (bog1ser-a vs bog2ser-are).
•
Lexical accent assignment is sensitive to an intervening Pω-barrier between the stress and the left edge, also for the ”strong”
accent-2-
inducing suffixes (för-1för-a, för-1för-are). …hm…
•
Postlexical accent 2 assignment is insensitive to both distance and barriers.
Proposal for be- and för-
•
Lexical tonal information is only present/visible when ωmin
= ωmax
•
Any expansion of Pω
entails invisibility of lexical tones ( post-lexicon)
•
Therefore, unless accent 2 can be motivated post-lexically, we should expect
accent 1 (= no lexical tone) in forms like be-tala and för-föra.
Proposal for be- and för-•
Lexical information is no longer relevant at the post-lexical level:
2(ˈsommar2
)
2((ˈmid)(ˌsommar2
)(ˌdansen))2(ˈför-a2
) (för-1(ˈföra2
))
•
Unifying property of compounds and be- and för- formations: They’re projections of ω
”ert”
inert
Conditions on post-lexical accent 2 (or why betala can’t get accent 2)
•
Postlexical accent 2 is the result of two pitch accents being assigned to one maximal prosodic word: –
One final phrasal accent (L*H, the prominence tone, the focus tone of Bruce 1977 and others),
–
one initial word tone (H*).
•
TBU is a stressed syllable (head of ω)two TBUs required for postlexical accent
2, viz. a compound.
ˈtull+ˈpack+ˈhus+ˈkarl-s+ˈlag
ω ω ω ω ω
ω
H* -------------------------------L*HL%
Association is required of both tones.
Cf. Gussenhoven 2005 for English phrases
Align(Ф,T*,Rt): The right edge of every Ф
coincides with a pitch
accent
Align(Ф,T*,Left): The left edge of every Ф
coincides with a pitch
accent
The result for CSw is a ”compound rule”, however.
Asymmetry:
•
In simplex as in compounds, the word tone of accent 2 is always associated. (Unassociated = unrealized.)
•
The prominence tone can be bumped by the lexical
word tone
(H*) and remain
unassociated in simplex, •
but the prominence tone can’t be bumped by the postlexical
word tone
(H*) (in CSw)
Accent 1 Accent 2
ˈs y r a k ˈe l a k ǀ ǀ L*H L% H*LoHL%
’angry’ ’mean’
Accent 2, compound
ˈm e l l a n ˌm å l ˈm e l l a n ˌm å l ǀ ǀ ǀ H* L*HL% H*LoH L%
’snack’
*
Lexical word tone may bump the prominence tone,
But postlexical word tone may not
•
Therefore, post-lexical accent 2 is not an option in betala.
Accent 1 Accent 2
b e ˈt a l a b e ˈt a l a ǀ ǀ L*H L% H*LoHL%
’pay’
*
Recursivity and layering
Ito & Mester 2007: •
Recursivity from the ω
and up
•
Strict Layering from the ω
and upSomewhat challenged in Swedish
•
No Recursivity from the Ft and down•
Unstrict layering from the Ft and down
σ för- ˈtala
ω
ω’
σ för 1ˈmånga
ω
ω’
σ för 2ˈmånga
ω
Ф
ω ω ω
ω
2ˈför-ˌan-ˌleda
ω ω
ˈför-ˌmånga
ω*
ω ω
2ˈför-ˌtala
ω*
σ gå 1ˈetter
ω
ω’*
0gå 2ˈetter
Ф
ω ω
2ˈgå-ˌetter
ω
ω ω
a
bc
d
Layers identified for CSw
•
min ω: syllabification domain, one stress•
max ω: exhaustive parsing into ω
(stressing be-, för-), domain of post-lexical accent 2.
•
Adjunction of unstressed be- and för- to min ω
indicate an intermediate expansion:
ω’
The connection between tonal grammar and accentual
distribution in compounds•
CSw and other dialects have so-called connective accent in compounds
•
SSw and other dialects don’t
The prominence
tone
is left- aligned
The prominence
tone
is associated and right-
aligned (and left-aligned, too, in most of them)
•
Left-alignment only of prominence tone admits accent variation in compounds
•
Interaction of prosodic factors: clash, Auftakt, locality,…
•
With morphological factors: linking element, morphological grouping, lexical tonal specification of first member, word class, …
South less South
even less
South
Central Sw.
nonce compounds Malmö 1973
K-stad 1973
H-stad 1973
Sthlm 1973
a. pros. non-local ˈtaxi-ˈgris 1 1 1 2 b. linking-s, morph. non-
ˈskog-s-ˈhals 1 1 1 (48%)
local, pros. local ˈskog-s-ˈhals 2 (52%) 2 c. Auftakt baˈnan-ˈkust 1 1 (35%)
pros. local baˈnan-ˈkust 2 (65%) 2 2 d. pros. non-local ˈlax-choˈklad 1 2 2 2
e. participle ˈin-ˈkläckt 1 1 2 2
f. monosyll., ˈklot-[ˈarm-ˈband] 1 (68%)
morph. non-local, pros. local
ˈklot-[ˈarm-ˈband] 2 (32%) 2 2 2
g. l,r,n pros. local ˈcykel-ˈplank (l,r,n) 1 (43%)
or non-local) ˈcykel-ˈplank (l,r,n) 2 (57%) 2 2 2 h. clash, local ˈblod-ˈprins 2 2 2 2
i. clash, local [ˈmask-ˈros]- ˈbrand 2 2 2 2 j. lex. spec. ˈsommar2-ˈträsk 2 2 2 2
Some references
Abrahamsen, Jardar E. 2003. Ein vestnorsk intonasjonsfonologi. NTNU, Trondheim.Bruce, Gösta. 1974. Tonaccentregler för sammansatta ord i några sydsvenska stadsmål,
in: Platzack, C. (ed.): Svenskans beskrivning 8, 62–75.Bruce, Gösta. 1977. Swedish word accents in sentence perspective. (Travaux de l’institut
de linguistique de Lund 12) CWK Gleerup, Lund.Bruce, Gösta. 1998. Allmän och svensk prosodi. Praktisk Lingvistik 16. Lund University.Elert, Claes-Christian. 1970. Ljud och ord i svenskan. Uppsala, Almqvist & Wiksell.Ito, Junko & Armin Mester. 2007. Categories and Projections in Prosodic Structure. OCP
4, Rhodes.Ito, Junko & Armin Mester. 2006. Prosodic Adjunction in Japanese
Compounds. Proceedings of FAJL 4, Osaka, 2006
Kristoffersen, Gjert. 2000. The Phonology of Norwegian. (The phonology of the world’s languages.) Oxford University Press, Oxford.
Lahiri, Aditi, Allison Wetterlin & Elisabet Jönsson-Steiner. 2005. Lexical specification of tone in North Germanic. Nordic Journal of Linguistics 28, 1, 61–96.
Malmgren, Sven Göran. 1992. Om tonaccent i centralsvenskt standardspråk –
och i svenska ordböcker. Svenskans beskrivning 19, 206–214. Lund University press.
Morén, Bruce. 2006. Central Swedish noun inflection and prosody: synchrony at its best (and no underlying tones!). Rough draft for project group (nov. 2006), CASTL, Tromsø.
Noreen, Erik. 1915. Ärtemarksmålets ljudlära I. Norstedts, Stockholm.
Riad, Tomas. 1998b. Towards a Scandinavian accent typology, in: Kehrein, W. & Wiese, R. (eds.) Phonology and Morphology of the Germanic Languages, 77–109. (Linguistische Arbeiten 386) Niemeyer, Tübingen.
Riad, Tomas. 2006a. Scandinavian accent typology. Sprachtypol. Univ. Forsch. (STUF), Berlin 59 (2006) 1, 36–55.
Riad, Tomas. 2003. Diachrony of the Scandinavian accent typology, in: Fikkert, P. & Jacobs, H. (eds.) Development in Prosodic Systems (Studies in Generative Grammar 58). Berlin/New York: Mouton de Gruyter. 91–144.
Riad, Tomas. 2008. Prosodi i svenskans ordbildning. Ms. Stockholm university.Selkirk, Elisabeth. 1996. The prosodic structure of function words. In Morgan,
James L. & Katherine Demuth (eds.) Signal to syntax. Bootstrapping from speech to grammar in early acquisition. Lawrence Erlbaum Associates: Mahwah, NJ. 187–213
Smith. Jennifer L. 2001. Lexical Category and Phonological Contrast. In R. Kirchner, J. Pater, and W. Wikely (eds.). PETL 6: Proceedings of the Workshop on the Lexicon in Phonetics and Phonology. Edmonton: University of Alberta, 61–72.
Wetterlin, Allison. 2007. The Lexical specification of Norwegian tonal word accents. Ph. D. diss. Konstanz.