Upload
casey-mawdsley
View
218
Download
1
Tags:
Embed Size (px)
Citation preview
Adjectival Flection andN-deletion in the Dutch
Dialect Regions
Ton GoemanMeertens Instituut / KNAW - Amsterdam
Variflex Variation in Inflection resumé van het nieuwe NWO-programma
21 augustus 2003 UVA
The role of morphology in a phonological process
Does Morphological conditioning emerge from Phonology?
• A. Anttila “Morphological Conditioned phonological alternations”
• ‘morphological conditions emerge in environments where the phonological conditions are at their weakest’
Are there indications for level ordering:?• N-deletion on every level
• So that uninflected&(un)derived forms show more deletion than inflected forms
Morphological or (morpho)syntactical?
In our case w.r.t N-deletion: Do we have to think primarily in syntactic terms
confronted with morphological categories?
Or, are morphological categories explained away by functional syntactic terms?
N.B. I can not yet answer the parallel question Is morphological conditioning emerging from
syntax?
The case
Word final N-deletion
Adnominal word final -N Word final -N in other Morphological categories
Adnominal -N in Masc. Adjectives
Ene goeien oogst “a good harvest”
Enen droge kerel “a flat character”
Enen houten boom “a wooden beam”
N retained if before:• VOWEL• T/D• B• R
N Deleted: before the other CONSONANTS
N.B.
This phonological conditioning is variable
Variability according to region
Word final -N in other Word Classes Adv/Prep buiten ‘out of’ > buite
Noun sing.: jongen ‘boy’ > jonge
Noun plur.: jongen ‘young animals’ > jonge
Verb infinitive: geven ‘to give’ > geve
Verb pret. particip.: gegeven ‘(been) given’ > gegeve
Verb present: zij geven ‘they give’ > geve
Verb past: zij gaven ‘they gave’ > gave
Language internal factors: Phonology
Word internal phonology:
Influence of preceding consonant• Labial, Labiodental, Coronal, Velar, Liquid
Phrase-phonology:
Influence of following • Vowel, h, t/d, b, nasal, r, other Consonant or Pause
Word internal Phonology: -(e)N after Vowel & Consonant
Darkgreen: 0-12 % to Darkbrown: 88-100 %
Phrasal Phonology
-N before Vowel Non-adnominal(black) Adnominal(brown)
Ten onder?
-N before H….Non-adnominal(black) Adnominal(brown)
-N before T, D….Non-adnominal(black) Adnominal(brown)
-N before B….Non-adnominal(black) Adnominal(brown)
-N before other Consonants….Non-adnominal(black) Adnominal(brown)
How is the fate of Wordfinal -N
in the other
Morphological Classes?
Morphological Classes: N-deletion in green area’s
Adv/Adj/P/Num(mono-) Nsing (mono-) Nplur (bi-morph.)
Darkgreen: 0-12 % (deletion) to Darkbrown: 88-100 %
(retained)
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Morphological Classes: N-deletie in green area’s
Infinitive Past Participle Finite Verb (plur.)
Darkgreen: 0-12 % to Darkbrown: 88-100 %
(deletion) (retained)
Morphology……
What are the relevant Morphological classifications?
Morphological Theory:classes are allocated differently by morphological characteristics: 9 divisions
1 Syntactical Word classes (A P N V ) eigen (A), tegen (P), Pasen (Nsing.), glazen (Nplur.),
krijgen (V)
2 Finite-Infinite jongen ‘young animals’(finite) – jongen ‘boy’ (sing.),
tegen (P) zij lopen (finite) -lopen (infinitive), gelopen (Past
Participle)
3 Derivation-Inflection• krijgen (plur.) V• krijgen (infinitive) N, gekregen (past participle) N
Flection after/on) Derivation (default)
4 Affixal-Nonaffixal = Bi-morphemic-Monomorphemic
• jongen Nplur. ‘young animals’• jongen Nsing.‘boy’
5 Finer grained morphological classification than A P N and V: Afin, APinf, NOMinf, NOMfin, Infinitive, Pastparticiple,
Present, Past
6 Morphological Theory; Booij 1994Lexical Phonology:- Inherent inflection- Contextual inflection
Wordclass Inherent inflection Contextual inflection
N number
A comparative superlative [number~gender~definiteness] (portmanteaumorpheme)
V infinitive participle tense [number~person](portmanteau morpheme)
7 Morphological Theory: Kiparsky (1972):
Strong-Weak Features
Weak Strong
case number
verb agreement(in languages with no pronoundeletion)
tense
gender
verb agreement(in languages with pronoun deletion)
8 Theory: Kiparsky (1994) w.r.t. Word final t-deletion: strata
Morphological conditioning by constraint interaction Deletion is the outcome by different ordering of PARSE
with a family of SyllableWellFormednes Conditions:
• SyllWFF-root ≈ no inflection eigen
• SyllWFF-stem ≈ irregular inflection kregen (plur. Pret.) gekregen (part. Pret.)
• SyllWFF-word≈ regular derivation & flection krijgen (inf.) krijgen (plur. Pres.)
9 Theory: Morpho-syntactic (Kean):N-Plur and Infinitive are derivational, all other Verbal forms are inflectional
Agrammatic patients Deficit where the ‘minimal word’ remains intact
Patients drop inflection in A, V
(un)derived words - derived Nplur. and Infinitive - remain intact
Survey of Morphological features in 9 Morphological Models
Morpholo-gical Class
Word Class
Finite-Infinite
Affixal-Nonaffixal
Derivation -Inflection
Inherent-Contextual (Booij)
Strong-Weak (Kiparsky 1972)
Syllable WFF
(Kiparsky 1994)
Morpho-syntax (Kean)
Afin A Fin Affix Inflected Contextual Weak Word Inflected
APinf A/P Inf Nonaffix None Mixed Mixed Root (Un)Derived
NOMinf N Inf Nonaffix None Inherent Strong Root (Un)Derived
NOMfin N Fin Affix Inflected Inherent Strong Word (Un)Derived
Infinitive N Inf Affix Derived Inherent Strong Word (Un)Derived
Pastpartic N Inf Affix Derived Inherent Strong Stem Inflected
Present V Fin Affix Inflected Contextual Weak Word Inflected
Past V Fin Affix Inflected Contextual Weak Stem Inflected
Selected items:
114 wordforms
-Adjectives
-Adverbs
-Prepositions
-Nouns
-Verbs
Data selected from GoemanTaeldemanVanReenen-database
http://www.meertens.knaw.nl/projecten/mand/MANDintroE.html
Social characteristics in database
N Range MeanSex speaker m/f 252 / 117Age 25-84 61Occupational Prestige 13-89 46Sex fieldworker m/fmostly younger than speaker
201 / 158
Dialect areas& Number of localities
Regions Number of localities
River area dialects 31
North East dialects 152
West dialects 81
Zeeland dialects 29
South East dialects 76
What are the structural differences
between dialect area’s?
Discern the significant factors that condition the variation
• The relative strenght of those factors
A. What is the influence of social context:• Social characteristics of speakers and interaction
with fieldworkers
B. Influence of linguistic structural factors:
linguistic social
Unexplainable restfactors
N-deletion
0%…………………………………………………100%
The Strategy is Partitioning
Partition off all unexplained variation Partition off all non-linguistic variation
(social factors and interaction in fieldwork)
Then the linguist gets his part The real linguistic variation, uncontaminated
by non-linguistic factors: linguistic structure
Patterns in the dialectsGeneral structure of models used:
Linguistic variables
Explanatory Phonological variables• Wordphonology: Influence of preceding Consonant
• Phrase-phonology: Influence of following Vocaal, h, t/d, b, nasal, r, other Consonant or Pause
Explanatory Morphological variables• The 9 Morphological Classifications allocated differently
by Theoretical positions
Structure of models 2Social factors
Main effects• Sex speaker• Age speaker• Occupational prestige speaker• Sex fieldworker
Social Interactions (accommodation of speaker)• Sex speaker x Sex fieldworker• Age speaker x Sex fieldworker• Occupational prestige speaker x Sex fieldworker
N.B. Fieldworkers (students) were young and their Occupational
prestige is rather undefined yet
Given the areal differences: analysis by Region (5)
• This gives 5 (Region) * 9 (Morph. Classifications) = 45 statistical models to test for
Which of the models are best? Those morphological allocations that
explain most of the variation Measure % Variance explained
Over all 5 regions
In what follows: Morphological Models are ordered from
left to right as best to less
Percent explained variance in 9 different Morphological Models
North 51 48 48 48 51 51 48 48 48
Zealand 49 41 40 38 38 38 41 36 35
South 35 32 32 32 32 32 31 32 31
RiverArea 26 24 24 24 22 22 22 22 22
West 15 14 13 13 11 11 12 11 11
Models
Morph Class
Deriv / Infl
Kip-SyllWFF
Affix/Non-affix
Kip-Strong/
Weak
Word Class
KeanAgrammatism
Inherent/Contextual
Finite/
Infinite
Best ………………………………………………………………………………………………. Less
Choose the best Morphological Model (9) over 5 geographical Regions:Percent explained Variance:
0
20
40
60
NorthZealand
SouthRiverArea
West
MorphClassDerivation-Inflection
Kiparsky-SyllWFFAffix-NonAffix
Kiparsky_Strong-WeakWordclass
Kean-AgrammatismInherent_Contextual
Finite-Infinite
MorphClassDerivation-InflectionKiparsky-SyllWFFAffix-NonAffixKiparsky_Strong-WeakWordclassKean-AgrammatismInherent_ContextualFinite-Infinite
Best model is Morphological Classes,
followed by Derivation-Inflection
Conclusion 1
A division between inflected and non-inflected forms has less explanatory power
Therefore, a stratum view based on these divisions is not applicable here
Conclusion 2: Social effects All main social factors and interactions are
significant Except:
• North-East main effect: sex fieldworker• West main effects: age, occupational prest
interaction: age*sex fieldworker• Zealand main effects: sex speaker, occupational
prestinteraction: sex
speaker*sex fieldworker• South main effect: occupational prestige
interactions: all
Concentration on Linguistic Main Effects
Model by Region
To see where Morphology outweighs Phonology
The effect of Word internal Phonology is less than that of Phrase Phonology
Mean(@n_Ndel_@)
1.299
0
0.840109
Sex speaker
mv
Age
46 7661.7332
Occup_prestige
13 7843.789
Sex-fieldworker
mv
PostCV
cor lab
labdentlabvel
liq
nasvel
vel
Word_Init_Cons
B H
TD
cons
liqnas
pausa
voc
Morphol_class
apfinapinf
infinitive
nomfinnominf
pastparticiple
pres pret
River Area
Mean(@n_Ndel_@)
1
0
0.830749
Sex speaker
mv
Age
25 8461.0296
Occup_prestige
13
86.9
47.5459
Sex-fieldworker
mv
PostCV
cor lab
labdentlabvel
liq
nasvel
vel
Word_Init_Cons
B H
TD
cons
liqnas
pausa
voc
Morphol_class
apfin apinf
infinitive
nomfinnominf
pastparticiple
pres pret
North
Mean(@n_Ndel_@)
1.125
0
0.916137
Sex speaker
mv
Age
39 8461.5919
Occup_prestige
14 7843.4477
Sex-fieldworker
mv
PostCV
cor lab
labdentlabvel
liq
nasvel
vel
Word_Init_Cons
B H
TD
cons
liqnas
pausa
voc
Morphol_class
apfin apinf
infinitive
nomfinnominf
pastparticiple
pres pret
Wes t
Mean(@n_Ndel_@)
1.104
-0.216
0.6764
Sex speaker
mv
Age
26 7461.4232
Occup_prestige
13
73.4
43.0743
Sex-fieldworker
mv
PostCV
cor lab
labdentlabvel
liq
nasvel
vel
Word_Init_Cons
B H
TD
cons
liqnas
pausa
voc
Morphol_class
apfin apinf
infinitive
nomfinnominf
pastparticiple
pres pret
Zealand
Mean(@n_Ndel_@)
1.144
0
0.742333
Sex speaker
mv
Age
39 7762.3043
Occup_prestige
18.9 89.1
47.5988
Sex-fieldworker
mv
PostCV
cor lab
labdentlabvel
liq
nasvel
vel
Word_Init_Cons
B H
TD
cons
liqnas
pausa
voc
Morphol_class
apfin apinf
infinitive
nomfinnominf
pastparticiple
pres pret
South
PhrasePhon exceedsMorph
…………………Morph exceeds PhrPhon
…………………Morph exceeds PhrPhon
………………….PhrPhon exceeds Morph
………………….PhrPhon exceeds Morph
Conclusions 3 In all regions, Morphology is a significant factor
In two regions: North and West, Morphology is stronger than Phonology
• West: Nouns inhibit N-deletion• North: masc Adj. show N-deletion, all other categories inhibit it
In North and West, phonological conditions (albeit significant) are relatively weaker than in the other regions and morphology is stronger (cfr. Antilla)
In the River Area and in Zealand Morphology could grow stronger
Conclusions (last)
There are only weak indications for level ordering The best model does not embody it
The second best model (Derivation-Inflection) fares best in the North-East Inflection pushes deletion This runs counter to expectation: in level ordering
(un)derived forms should show more deletion
Pending Problems Is morphology emerging from syntax?
Antilla-style: where syntactic factors are weak I have no answer to that now Indications via morpho-syntactic aspects of the
models
Remember ------->> Survey of Morphological features in 9
Morphological Models
Survey of Morphological features in 9 Morphological Models (repeated)
Morpholo-gical Class
Word Class
Finite-Infinite
Affixal-Nonaffixal
Derivation -Inflection
Inherent-Contextual (Booij)
Strong-Weak (Kiparsky 1972)
Syllable WFF
(Kiparsky 1994)
Morpho-syntax (Kean)
Afin A Fin Affix Inflected Contextual Weak Word Inflected
APinf A/P Inf Nonaffix None Mixed Mixed Root (Un)Derived
NOMinf N Inf Nonaffix None Inherent Strong Root (Un)Derived
NOMfin N Fin Affix Inflected Inherent Strong Word (Un)Derived
Infinitive N Inf Affix Derived Inherent Strong Word (Un)Derived
Pastpartic N Inf Affix Derived Inherent Strong Stem Inflected
Present V Fin Affix Inflected Contextual Weak Word Inflected
Past V Fin Affix Inflected Contextual Weak Stem Inflected
Global Character of these Morphological Models
Morphology (Morpho)Syntax
Morphological Classes Major Wordclasses APNV
Derivation-Inflection Derivation-Inflection Affixal-Non-affixal Inherent-Contextual Inherent-Contextual Strong-Weak Strong-Weak Syllable-WF Morphosyntax (agramm.)
The result That Morphological Classes is best Thus
a very traditional Priscinianus-type of morphological model
is better than any other model
Specifically this result suggests
That it is better than models with syntactic ramifications
Ton Goeman +31 (0) 20 - 46 28 532
www.meertens.knaw.nl/projecten/mand/MANDintroE.html
Meertens Instituut Amsterdam-Netherlands
Royal Netherlands Academy of Arts and Sciences
Workshop Variation in InflectionMeertens Instituut August 21, 2003