Constituent order Flashcards
What is constituent order?
- intuitive notion
- “word order”
- Last night [the oldest friend I ever had] met [Jane’s younger brother].
- S, V,and O
#SAO #universal primitives
S = S, A O = O
Fixed constituent order
In some languages it is clear what the order should be.
> As in English.
Children like lollies
S V O
?Lollies like Children
with some variation > Marked constituent order. > Licensed by discourse structure. >CONTRASTIVE FOCUS: Children like lollies. [SVO] Lollies, children like. [OSV]
Children hate spinach. Lollies, children like.
> Lexicalised formula
Believe you me.
*Stop you talking
Dominant constituent order
Languages can allow more variation than English.
> And still have a prevalent basic constituent order.
Russian (Slavic, Russia)
> Languages w rich morphology (case marking).
> Avoids confusion between participants.
- NOT English where word order is a primary clue.
> (Often the only clue.)
- clear preferred order
- ALL LANGS have more than one order
» EVEN when treated as “fixed constituent order” (eng, russian)
Deviations exploited for special functions in discourse.
>One of the orders is dominant
Six logical possibilities of constituent order, in order
SOV - 45% SVO - 42% VSO - 9% VOS - 3% OVS - 1% OSV - 0%
all attested in at least one lang.
Not evenly distributed
Why is constituent order not evenly distributed?
Random distribution = c. 16%.
There must be other factors.
> EXTERNAL factors.
> So far no clear understanding of what they are.
But we do have a good idea of the distribution.
> And elaborate INTERNAL explanations for it
Further observation of constituent order distribution
GREENBERG’s Universal 1
96% have S before O.
Greenberg 1:
In declarative sentences with nominal SUBJECT and
OBJECT, the dominant order is almost always one in
which the S precedes O
Functional explanation for Greenbergs Universal 1?
Comrie’s (1989) ‘subject saliency.’
> Subject is Agent: initiates and controls the action.
> Object is Patient: being acted on and affected by the event.
> Agency makes subjects salient (most noticeable/important) to human cognition.
- Cognition reflected in syntactic organisation. - In prototypical transitive clauses S comes before O.
Further observations: Is there a preference for contiguity of V and O? Why?
91% have O next to V (left or right).
Syntactic explanation?
Rules of phrase structure
> in Govt and Binding Theory
> Generative grammar
S –> NP(subject); VP
VP –> V; NP(object)
ultimately also a cognitive explanation
What about OSV ?
OSV violates both the subject saliency principle. And Govt and Binding postulates.
> And it’s the rarest of all.
> But it does occur.
But no psycholinguistic
tests on these matters.
Implicational Universal 5.
If a language has a dominant SOV order and the genitive follows the governing noun, then the adjective likewise follows the noun
Jane yogurt eat
+ yogurt of jane # of jane yogurt
» yogurt fresh # fresh yogurt
Predictions from V,S, and O
we can predict ordering of other pairs:
Nouns and adjectives >Gberg's Universal 5 Nouns and genitives Adpositions and nouns Affixes and roots, etc.
Languages tend to be
consistent in the way they
order different components
Lehmann’s correlations
1973, 1978
Generalised based on order of O/V pair - Leaks ! v general - see table VO vs OV prefix suffix ?word first ?word after noun +gen/adj gen/adj+noun Main v + aux v Aux v+main v Neg + v v + neg
Venneman : heads and dependents (1973, 1978)
The predictions run both ways
Verbs => nouns
- If verb dependents are BEFORE the verb,
then noun dependents are BEFORE the noun.
Nouns => verbs
- If noun dependents are BEFORE the noun,
then verb dependents are BEFORE the verb.
> Etc.
MORE GENERAL PRINCIPLE (improved degree of generalisation)
Consistency in ordering these pairs:
> Some languages always place heads before dependents.
> Some languages always place dependents before heads.
Verbs: heads for objects, adverbs, negatives dependents.
Nouns: heads for genitives, adjectives, relative clauses dependents
Hawkins: category harmony (1983)
Becomes a matter of proportion.
>If MOST of the verb dependents are BEFORE the verb
then MOST of the noun dependents are BEFORE the noun.
What langs have OSV word order?
4 languages cited in WALS 1. Kxoe (Zimbabwe) 2. Nädeb (Brazil) 3. Tobati (PNG) 4. Wik Ngathana (Australia, near PNG), + Warao, isolate, Venezuela >>>> Yoda’s language
Leaks in Lehmanns/Vennemans theory
> Unproblematic variation.
More general problem with relative clauses.
The head-dependent principle leaks.
Relative clauses in particular.
VO languages tend to place relative clauses after nouns.
> OV languages also place relative clauses after nouns.
Or: relative clauses are always after nouns.
> Notion of ‘heavy constituent’ always to the right.
Not meant to NOT have exceptions?
observation stands, but empirically not valid for relative clauses
Branching Direction theory
Dryer’s (1992) alternative solution (to leaks in rel clauses)
> Eliminates the notions of head and dependent.
Instead:
> NON-phrasal components: NON-branching.
- Or atomic. (V)
> Phrasal components: branching. (NP)
The girl hit (V) [the boy] NP
Principle:
Languages consistently:
> Either put branching to the right of non-branching.
> Or to the left of non-branching.
E.g. English is right-branching
Hit [NP:(the boy)]
man [RelCl:(whom i hit)]
Oroqen (Manchu-Tungus: China) is left-branching
Benefits of Branching Direction Theory
- Avoids theory-informed notions of head/dependent.
- Improved explanatory power.
- Better accounts for the placement of adjectives.
> Students [HEAD: attend] [DPT: linguistic courses]
vs
The [DPT: tired] [HEAD: student]
under head/dep theory, no explanation for this. branching theory has explanantion:
Students [NB: attend] [B: linguistic courses.]
The [NB: tired] student
The [NB: student] [B: tired to work so hard]
probably still leaks though ****
Internal vs external explanation of branching direction theory?
This is very much an internal explanation.
> I.e. reveals the internal logic of the system.
> External factors are not in focus.
We need more insight about linguistic factors.
> Historical developments (Givon 1977).
> & Effects of contact
impact on typology of constituent order.
has been discussed by venneman VSOSVO VSO ---> FWO FWO ---> SOV SOV----> SVO
Take home point about constituent order
Languages follow consistent principles
of word-ordering.
flexible constituent order
- no dominant order
EG:
Warlpiri (australia) - Anything goes as long as auxiliary is in second position
- NON-CONFIGURATIONAL LANGUAGES.
- Word order in NPs is also flexible.
- NPs can even be discontinuous.
- typically allow prodrop
Determining Constituent order
there R Competing constituent orders - variation (eg. russian) but clear dom.
How do we know?:
1. Native speaker's intuition > dont always have access 2. Frequency > work from texts > count, find highest figure 3. Markedness > basic order likely unmarked (less formal marking than other orders) > true for all levels of grammar
Determining constituent order from Native Speaker’s intuition
Native speakers don’t always have the ‘right’ intuitions.
>We don’t always succeed is understanding their intuitions.
Believe you me.
Lollies, children like.
> Methods of elicitation.
Field and language description methods!
Determining constituent order from Frequency
count > highest freq likely dominant
PROBLEMS: if no obv dominance, lang likely “flexible”
In some languages/texts
most clauses will not contain S and O as full NPs.
» Counts will be less useful.
Or: hard to find enough data.
ALSO: Genres and Registers
Certain orders can be stylistically dominant in your corpus.
»_space; Narrative texts often employ different order.
»_space; VS preferred in temporal sequencing, SV elsewhere.
»_space; Highly informative elements occur pre-verbally.
Same as above: need more data.
Determining constituent order from Markedness
LEAST RELIABLE TEST: lots of guessing
Basic constituent order likely to be unmarked/less formal marking than other orders
- PHONOLOGICALmarkedness
> Neutral intonation.
Children like lollies. (slowly falling pitch, drops sharply at end)
Lollies, Children like. (short intense pitch, brief pause, neutral intonation) - MORPHOLOGICALmarkedness
EG Kutenai (canada)
Sentence 1 VOS
Sentence 2-inverse suffix VSO
Sentence 2 has an inverse suffix, not present in 1.
»>Basic constituent order is as in VOS in 1.
3. SYNTACTICmarkedness EG German sentence 1 - main clause SVO Sentence 2 - relative clause SOV Sentence 2 is a relative clause. >>> Basic constituent order in main clause as in 1 (SVO)
Summary of determining constituent order
- Ideally a grammar may state the answer.
- Or we may ask a linguist.
> Often the data is FRAGMENTARY.
> Information must be inferred from what we have.
Each mistake introduces a bias in the typology.
Theoretical complications of determining constituent order
> > ‘Subject’ and ‘Object’ are theoretical constructs.
- Syntactic analysis is needed before we can decide.
DIFF ANALYSIS may lead to DIFF CONCLUSIONS