Syntax – Week 7, Lecture 2
1
Grammatical Relations and
Morphosyntactic Alignment
1. Review
Last lecture we saw how word order and constituent structure can be used
to define grammatical functions (such as subject, object, oblique etc.). The
location of these grammatical functions is defined structurally (for
example, the subject is the specifier of the IP). Because of this, every
speaker of a given configurational language (say English) knows how to find
these grammatical functions.
This way, English speakers know who did what to whom. This is how the
process works: A given verb such as lie says: “OK – as part of my meaning,
I’ve got two arguments (that is to say, entities engaged in a relation of lying).
One of the arguments is required, and that is the guy doing the lying. You
will find that guy in the subject. I also optionally have the guy being lied to.
If you are going to talk about that guy, you’ll find him in the oblique
grammatical function.”
Because all speakers share the meaning of verb lie, and know how to find
the grammatical functions, when they hear a sentence using this verb,
they know how to find which entity lied to whom.
Summary — the main ways in which GRs may be encoded are:
● by word order / configurationally
● and/or by case marking
● and/or by verb agreement.
Many languages use a mixture of these strategies.
Syntax – Week 7, Lecture 2
2
2. Grammatical Functions and Case
In today’s lecture we are going to examine case in greater detail. First,
let’s examine the process by which speakers of a case language can
understand who did what to whom.
In the sentence below, both farmer and gate have case marking:
(Lat) Agricola portam videt
farmer.m.sg.nom gate.f.sg.acc see.3.sg.pres
“The farmer sees the gate”
This case marking identifies the grammatical function of the NP. The
process by which who does what to whom is then communicated in a way
similar to English. Each verb defines the arguments it has, and says where
those arguments will be found (in the subject for example).
However, within case marking systems, there are several ways case can be
marked.
3. Different Coding Systems (Morphosyntactic Alignment)
English still has a case system in some of its pronouns:
English:
She goes He goes
She sees him He sees her
Diyari (South Australia)
Karna wapayi Wilha wapayi
man goes woman goes
‘The man is going’ ‘The woman is going’
Syntax – Week 7, Lecture 2
3
Karna-li wilha nhayiyi Wilha-li karna nhayiyi
man woman sees woman man sees
‘The man sees the woman’ ‘The woman sees the man’
Clearly we have differences here in the coding of grammatical relations.
Let us take as fundamental a distinction between intransitive verbs (1-
place verbs) and transitive verbs (2-place verbs). We now define three
core grammatical relations:
(a) S — the single NP argument of an intransitive construction:
English Diyari
The man goes Karna wapayi
The kangaroo goes Jukurru wapayi
(b) A — the NP argument of a transitive construction that (usually)
represents the agentive role:
The man spears the kangaroo Karnali jukurru thatiyi
(c) P — the NP argument of a transitive construction that represents the
(prototypical) patient role:
The man spears the kangaroo Karnali jukurru thatiyi
Note that the way that English and Diyari group their case markers differs.
So in Diyari, the S and P are unmarked, and the A is marked. In English the
S and A are marked one way, and the P is marked another.
There are several ways these three grammatical functions are grouped
in the languages of the world according to morphological and syntactic
behaviour of NPs. These different systems are known as systems of
Syntax – Week 7, Lecture 2
4
morphosyntactic alignment or case alignment (or just “alignment”):
(a) S and A functions are treated the same and P is different. In such a
system S+A is nominative and P is accusative:
Examples are Latin, English (morphologically for some pronouns, eg. 1sg),
and the vast majority of other European languages but also elsewhere
(esp. Africa, South Asia). Languages which pattern this way are called
“nominative/accusative languages”.
(b) S and P functions are treated the same and A is different. In such a
system S+P is absolutive and A is ergative:
Examples are Diyari (and many Australian Aboriginal languages), many
languages of the Americas (e.g. Inuit, Tsimshian) but also elsewhere.
Languages which pattern this way are called “ergative/absolutive”
languages (or sometimes just Ergative languages).
Why would an S be treated like a P rather than an A? Note that quite
often, S’s (i.e., subjects of intransitive clauses) do not behave in an
agentive way (i.e., they don’t control the event). Consider:
A
S
P
S
A P
Syntax – Week 7, Lecture 2
5
Patient-like S Agent-like S
Jill died. Jill left.
Jill fell. Jill lept.
Jill slipped. Jill slid.
Jill sneezed. Jill laughed.
Jill was sleeping. Jill was singing.
(c) S, A and P are treated separately (tripartite alignment). This is
relatively rare in the world’s
languages:
Examples are Thalanyji, Wangkumarra (Australian Aboriginal languages).
(d) S has one form and A and P are different. This is very rare, but is found
in some Iranian languages:
Why do you think this system would be rare?
(e) S, A and P are all unmarked, i.e. no case marking or verb agreement.
This is called neutral alignment. E.g. morphology of English nouns (and a
few pronouns like it, you) but not verb agreement or word order. Neutral
morphological alignment is also common cross-linguistically (many
S
P
A
S
A
P
Syntax – Week 7, Lecture 2
6
languages of east and SE Asia, parts of Africa). BUT neutral alignment is
not common syntactically – e.g. these languages tend to distinguish A and
P through word order.
Active-Stative Systems
An alignment system in which S is marked like an A or like a P depending
on the kind of event being described and the degree of control the
participant is seen as having over it.
Recall from earlier that some S’s are more patient-like (non-
volitional/non-agentive) and some are more agent-like:
Patient-like S Agent-like S
Jill died. Jill left.
Jill fell. Jill lept.
Jill slipped. Jill slid.
Jill sneezed. Jill laughed.
Jill was sleeping. Jill was singing.
In Active-Stative systems, if the ‘subject’ is semantically more like a
patient/undergoer/experiencer, it gets marked like a P. If it’s more
agentive, it gets marked like an A.
A
S
P
Syntax – Week 7, Lecture 2
7
Eastern Pomo (SW USA)
wi ceexelka wi baatecki
1sgSP slip/slide 1sgSP get_bumped
‘I’m slipping’ ‘I got bumped (accidentally)’
haa ceexelka haa baatecki
1sgSA slip/slide 1sgSA get_bumped
‘I’m sliding’ ‘I got bumped (on purpose)’
There is a fair amount of agreement across languages as to what counts
as a controlled act and what does not. However, such classifications are
also subject to cultural effects. For example, would you class ‘getting
drunk’ as an active or stative type of predicate?
Syntactically, in such languages, SA and SP may be grouped together as a
single grammatical function, despite the differences in treatment
morphologically. But in some such languages, it may be a better analysis
to recognise two core grammatical functions of (say) ‘Actor’ and
‘Undergoer’: Acehnese has been argued to be such a language.
Actor and Undergoer macroroles:
Actor: the participant which performs, effects, instigates, controls
the situation.
Undergoer: the participant which does not do any of these things,
but rather is affected by the situation.
Marked vs. unmarked case forms
Apart from the neutral alignment system, the above systems of alignment
all involve a distinction where some core grammatical arguments are
marked differently to others – e.g. Nom-Acc systems mark S/A arguments
one way and the P argument differently. BUT this doesn’t have to mean
that S/A gets one suffix and P gets a different suffix. Often, languages just
mark one of the two options, and leave the other unmarked:
Syntax – Week 7, Lecture 2
8
Case Core Grammatical
Arguments
Form of Marking
Nominative S,A usually Ø or relatively less marked
Accusative P usually some positive form of
marking e.g. suffix
Absolutive S,P usually Ø or relatively less marked
Ergative A usually some positive form of
marking e.g. suffix
Recall the examples from Diyari earlier:
Karna wapayi Wilha wapayi
man goes woman goes
‘The man is going’ ‘The woman is going’
Karna-li wilha nhayiyi Wilha-li karna nhayiyi
man woman sees woman man sees
‘The man sees the woman’ ‘The woman sees the man’
Observe here that the ergative suffix -li marks the A argument, while the
S/P arguments are simply unmarked (they don’t get any special suffix or
other inflection).
Syntax – Week 7, Lecture 2
9
Verb agreement
S, A and P arguments can also control verb agreement in different ways.
E.g. in some languages like English, only S/A arguments control the verb
agreement, but not P arguments. This points to a nominative-accusative
system. Consider:
I sleep. He sleeps.
I see the dog / the dogs / you He sees the dog / the dogs / you.
Some further examples from Kroeger (2005) Analyzing Grammar: An
introduction, pp.112-5:
Nominative-accusative verb agreement
Huasteca Nahuatl (Mexico) – bold = NOM, italics = ACC
ni-kotSi ‘I sleep’ ni-mic-ita ‘I see you (sg)’
ti-kotSi ‘you (sg) sleep’ ti-netS-ita ‘you (sg) see me’
Ø-kotSi ‘he sleeps’ Ø-netS-ita ‘he sees me’
Ergative-absolutive verb agreement
Tabulahan (Indonesia) – blue = ABS, orange = ERG
manahomi-ä’ ‘I fell down’
na-keki’-ä’ ‘It bit me’
ku-painsangi-koa’ ‘I told you (pl)’
la le’ba’-koa’ ‘you (pl) are leaving’
ung-allingk-ä’ ‘you (pl) bought (it) for me’
Lakhota exercise.
Syntax – Week 7, Lecture 2
10
Consider the following data. Comment on the affixes attached to the verb
stem: what do they indicate?
1. Ø-wa-kte ‘I killed him/it.’
2. ma-Ø-kte ‘He killed me.’
3. ma-ya-kte ‘You killed me.’
4. ni-Ø-kte ‘He killed you.’
5. Ø-Ø-kte ‘He killed it/him.’
6. wa-hi ‘I arrived.’
7. ya-u ‘You are coming.’
8. Ø-lowa ‘He sings.’
9. ma-khuzhe ‘I am sick.’
10. ni-haske ‘You are tall.’
11. Ø-khata ‘He/she is hot.’
12. hokshila ki Ø-khata ‘The boy is hot.’
13. hokshila ki matho wa Ø-Ø-kte ‘The boy killed a bear.’
14. matho wa Ø-wa-kte ‘I killed a bear.’
Note: patient prefixes agent prefixes
1 ma wa
2 ni ya
3 Ø Ø
order of prefixes is: P – A – V;
either A or P may occur when verb is intransitive
if full NPs are present their order appears to be A – P – V
NB: ki = definite & wa = indefinite particles
Syntax – Week 7, Lecture 2
11
Split Ergativity (Mixed Ergative Marking Systems)
In the early 70’s, Silverstein started looking for larger generalizations cross
linguistically about the way languages mark case. Silverstein (among
others) noticed that when a language has ergative/absolutive case
marking, what tends to occur is that some aspect of the language, either in
the case marking or in the verb agreement, uses nominative/accusative
case marking instead of solely ergative/absolutive case marking. That is, the
language splits between nominative/accusative and ergative/absolutive
case marking.
Silverstein asked whether there was a cross-linguistic generalization that
could be made about where this split occurs. He found that generally, a
split occurs somewhere on an animacy hierarchy.
Silverstein’s (or animacy) hierarchy:
1&2 prons > 3 prons > proper names & kin terms > human CNs > animate CNs > inanimate CNs
NOMINATIVE-ACCUSATIVE -> <- ERGATIVE-ABSOLUTIVE Example: Dyirbal a. Ngaja balan jugumbil baninyu. 1sg the woman came S S V(intrans) ‘I, the woman, came.’ b. Ngayguna balan jugumbil balgan. 1sg the woman hit P P V(trans) ‘(Someone) hit me, the woman.’ Syntax – Week 7, Lecture 2 12 c. Ngaja banggun jugumbiru bayi yara balgan. 1sg the woman the man hit A A P V(trans) ‘I, the woman, hit the man.’ • Compare the 1st person pronouns: ngaja (S/A) and ngayguna (P) – this looks nominative/accusative. • Compare the common NPs meaning ‘the woman’: balan jugumbil (S/P) and banggun jugumbiru (A) – this looks ergative/absolutive.