Transcript Word Order Typology

Word Order
Typology
(based on Comrie 1981, Croft 2003, Song 2011)
Founding father: the 60s
Joseph Greenberg (1963 talk at First Dobbs
Conference on Language Universals) - ts-ts!
 30 (+140) languages
 1966 ‘Some universals…’ – 45 universals


Universal 1: In declarative sentences with nominal
subject and object, the dominant order is almost
always one in which the subject precedes the object
VSO -> prepositional (almost absolute)
Methodological preliminaries
Issue of basicness:
 Full NPs, prototypical definite S and O,
independent clause, stylistically neutral…






Pronominal clitics misbehave
Main and dependent clauses diverge
‘Thetic’ utterances
Interrogative may have special orders… etc.
Frequencies, morphological unmarkedness
Flexible word orders not considered
Methodological preliminaries
Croft on basicness:

Not restricted to a grammatical subclass



There goes the ice cream truck.
*There crushes the ice cream truck.
No pragmatic / semantic specialization

Philosophy I’ve always enjoyed.
Methodological preliminaries
Croft on basicness:

Less complex constructions


It’s Hana that Federico likes
Frequency

Maybe a reliable proxy for all of the above
Methodological preliminaries
Misnomer:
 Not word order typology
 Rather, order of major constituents


Cf. Greenberg’s order of meaningful elements
We’ll have to live with it, though
Methodological preliminaries
A model topic for cross-linguistic research:
 Typological patterns with no apparent language
specific validity


We may not care about what basic word order is
while describing an individual language
But: present-day processing-based theories of
explanation (Hawkins) may be relevant for
individual grammars
Main lines of research:

Distribution: Possible orders of clause
constituents



Correlation: looking for universal
implications between orders of different types
of constituents


SOV, SVO, VSO, VOS, OVS, OSV
All attested, but unequally
Adpositions, adjectives, possessors, RelClauses
Explanation: looking for basic principle of
constituency

Branching? Processing?
Dominant orders:
kiho-ka
saca-lɨl
cha-ass-ta
Keeho-NOM lion-ACC kick-PST-IND
‘Keeho kicked the/a lion.’
(Korean)
khon níi kàt mǎa tua nán
(Thai)
Lladdodd draig
(Welsh)
man this bite dog CLF that
‘This man bit that dog.’
ddyn
killed
dragon man
‘A dragon killed a man.’
(cited after Song 2011)
Rare orders:
manasa ny lamba
ny vehivavy
wash
the clothes the woman
‘The woman is washing the clothes.’
piʔ
kokampö unkiʔ
(Malagasy)
(Panare)
child washes woman
‘The woman washes the child.’
samũũy
yi
qa-wùh
(Nadёb)
howler-monkey people eat
‘People eat howler-monkeys.’ (cited after Song 2011)
Implicational word order
universals: dominance
Dominant vs. recessive values of word order
 If an order only appears as an implicatum,
never as an implicans, it is a dominant order
SOV&NG ‫ ﬤ‬NA
AN ‫~ ﬤ‬SOV&NG
VSO ‫ ﬤ‬NA
AN ‫~ ﬤ‬VSO
NDem ‫ ﬤ‬NA
AN ‫~ﬤ‬NDem
NNum ‫ ﬤ‬NA
AN ‫~ ﬤ‬NNum
Implicational word order
universals: dominance

Dominant vs. recessive values of word order
Dominant order does not bind other values:
DemN
NDem
NA
+
+
AN
+
-
Implicational word order
universals: dominance
Dominant order is more frequent crosslinguistically (fragment of Croft 2003: 62 based
on Dryer 2001 sample)
SO
SV
NRel
DemN
NumN
NA
1
5, 17
24, IX’, XI’ , XII’, XXIII
V’(=18), XI’
VI’(=18), XII’
5, 7, 18, 21, 24, 40, XXI
96%
83%
73%
61%
55%
68%
Implicational word order
universals: harmony
One value is harmonic with other if it occurs
only with this other value; this is a directed
relation:
DemN
NDem
NA
+
+
AN
+
-
Implicational word order
universals: harmony
In a bidirectional universal, the two values
are mutually harmonic – and this is further
assumed (Croft 2003)
Prep
Post
NG
+
-
GN
-
+
Word order universals


From values to parameters
Greenberg 1966: “A dominant order may always occur, but
its opposite, the recessive, occurs only when a harmonic
construction is likewise present” Why AN&NDem is not
attested?
because both AN and NDem would be both recessive and
disharmonic
DemN
NDem
NA
+
+
AN
+
-
From correlations to
explanations
VSO / SVO / SOV ~ Pr / Po ~ NG / GN ~ NA / AN
 Out of 24 combinations only 15 attested, and
only 4 widespread:






VSO
SVO
SOV
SOV
&
&
&
&
Pr & NG & NA
Pr & NG & NA
Po & GN & AN
Po & GG & NA
O follows V
O precedes V
Invites for generalizations! From unilateral
implications to language types / profiles
S as a bad predictor – S dismissed
Lehmann (70s)
FPP: Fundamental Principle of Placement
 Concomitance: V & O vs. O & V
 Modifiers are placed on the other side of the
“Concomitant”


Adj, Gen, Rel
Inconsistent languages = languages under
change


Profiles “VO” and “OV” rather than literal VO/OV
Persian – “VO”, but (S)OV
Vennemann (70s)
PNS: Principle of natural serialization


Head (“operand”) vs. modifiers, or dependents
(“operator”)
Order <dependent,head> determined by <O,V>
 Theory-dependent: adpositions should be
considered heads
Vennemann (70s)
OPERATOR
object
adverbial
main verb
adjective
relative clause
genitive
numeral
determiner
adjective
standard of comparison
noun phrase
OPERAND
verb
verb
auxiliary
noun
noun
noun
noun
noun
comparison marker
comparative adjective
adposition
Vennemann (70s): Der Teufel
steckt im detail - overgeneralization
VSO / SVO / SOV ~ Pr / Po ~ NG / GN ~ NA / AN
 Out of 24 combinations, Vennemann allows only 3:





Hawkins counts that this accounts for slightly less than 50%
of his sample


VSO & Pr & NG & NA
SVO & Pr & NG & NA
SOV & Po & GN & AN
SOV & Po & GN & NA
But: Comrie’s ammendment: scale rather than two binary classes
SVO – bad predictor (nonce in Greenberg’s universals)

But: Dryer’s larger sample show that the factor is overestimated:
SVO do pattern with VSO, on the whole (SOV --- SVO – VSO)
Hawkins 1983
Make no exception for me, please!
Complicated system of multi-conditioned
implications
 Pr -> (NA -> NG)
 Pr -> (NDem -> NA)
 Pr -> (NNum -> NA), следовательно:
 Pr -> (Ndem -> NG), Pr -> (NNum -> NG)



Two exceptions! Ammendment:
Pr & -SVO -> (NDem -> NG)
Pr & -SVO -> (NNum -> NG)
Hawkins 1983

Shift from clause to NP constituents; implications
translated into HSP: Heaviness Serialization Principle
in a Prep language, the heavier the constituent, the less
likely it is located to the left of the head noun




heavy
Incipient functional motivation: the ease of processing


Det,Num < Adj < Gen < Rel
+ Det/Num N Gen/Rel
- Gen/Rel N Det/Num
light
Further elaborated in Hawkins 1994, 2004
More complicated with Post languages

+mobility principle( - some consituents are more mobile than
others
Dryer 1992

Large and principled sampling



Rehabilitation of VO~OV typology


600 lgs (1500 in his WALS map)
weighted for geneology
Including arguing for SVO to be indeed VO
Arguing against head – dependent
explanations
Dryer 1992

Against head – dependent explanations




From dependency to “patterning”
V-patterners vs. O-patterners




AN~NA order is unpredictable
Article, auxiliary are predicted in a wrong way
genitives and relative clauses are O-patterners
determiners and numerals are V-patterners
adjective are none-patterners
Uh-uh… calls for explanation!
Dryer 1992
Branching direction theory (BDT).
In a [XY] constituent:



the V-patterner is the non-branching (non-phrasal)
constituent (e.g. noun, article, numeral)
the O-patterner is the branching (phrasal)
consituent
(e.g. genitive phrase, relative clause)
in adjective + noun, none is branching…

at least, none is recursively branching
WOT cornerstones:
a typology of typologies





Greenberg: order in the clause (SOV etc.)
Lehman: order in the clause (“OV”~“VO”)
Vennemann: order in the clause (OV~VO)
Hawkins (early): adposition based (NPcentered), implications and hierarchies
Dryer: back to OV~VO
Photo stock