Lexical Functional Grammar ACTL08 Lecture 3 1
c. Ram
Subject-verb agreement d. Sita
(1) Ram calegaa Ram-masc will.go-Masc3Sg (Hindi)
(4) Ram Ram- masc ⎡ pred ⎢ ⎢ ⎢ ⎢ f :⎢ ⎢ subj ⎢ ⎣
IP NP (↑ subj)=↓
I ↑ =↓
N ↑ =↓
VP ↑ =↓
Ram (↑ pred) = ‘Ram’ (↑ pers) = 3 (↑ num) = sg (↑ gend) = masc (↑ case) = nom
V ↑ =↓ V ↑ =↓
calegaa (↑ pred) = ‘gosubj’ (↑ subj pers) = 3 (↑ subj num) = sg (↑ subj gend) = masc (↑ subj case) = nom (2) *Sita calegaa fem will.go-Masc3Sg (3) a. (f subj) = g b. calegaa
(f (f (f (f (f
pred) = ‘gosubj’ subj pers) = 3 subj num) = sg subj gend) = masc subj case) = nom
1
(g (g (g (g (g
pred) = ‘Ram’ pers) = 3 num) = sg gend) = masc case) = nom
(g gend) = fem calegaa will.go-Masc3Sg ‘gosubj’ ⎡ pred ⎢ pers ⎢ g :⎢ num ⎣ gend case
⎤ ⎤⎥ ‘Ram’ ⎥ ⎥ ⎥⎥ 3 ⎥⎥ sg ⎥ ⎥ ⎦⎥ masc ⎦ nom
2
Object and subject agreement
(5) ma t˘ am k˘ alaN wel-s@-Nil-am I those reindeer kill-T-DU-1SG ‘I killed these (two) reindeer.’
(7) ma t˘ am k˘ alaN wel-s@-Nil-am I those reindeer kill-T-DU-1SG ⎡ ⎤ pred ‘killsubj,obj’ ⎡ ⎤ ⎥ ⎢ ⎥ ⎢ pred ‘pro’ ⎥ ⎢ ⎦ ⎥ ⎢ subj g :⎣ pers 1 ⎥ ⎢ ⎥ ⎢ num sg f :⎢ ⎡ ⎤⎥ ⎥ ⎢ ⎥ ⎢ spec ‘these’ ⎥ ⎢ ⎣ ⎦ ⎣ obj h : pred ‘reindeer’ ⎦ num dual
(Nikolaeva 1999, Northern Ostyak)
IP NP (↑ subj)=↓
I ↑ =↓
N ↑ =↓
VP ↑ =↓
ma (↑ pred) = ‘pro’ (↑ pers) = 1 (↑ num) = sg
V ↑ =↓
NP (↑ obj)=↓
V ↑ =↓
wel-s@-Nil-am (↑ pred) = ‘killsubj,obj’ (↑ subj pers) = 1 (↑ subj num) = sg (↑ obj num) = dual
Det ↑ =↓
N ↑ =↓
t˘ am (↑ spec )= ‘these’
k˘ alaN (↑ pred) = ‘reindeer’
(6) a. wel-s@-Nil-am
(f (f (f (f
pred) = ‘killsubj,obj’ subj pers) = 1 subj num) = sg obj num) = dual
b. (f subj) = g c. ma
(g pred) = ‘pro’ (g pers) = 1 (g num) = sg
d. (f obj) = h e. k˘alaN
(h pred) = ‘reindeer’
2
3
Agreement and pronominal incorporation
(10) zi-n´a-w´a-lum-a subj- past -obj-bite-indicative ‘They bit them.’
Bresnan & Mchombo (1987), Bresnan (2001): Chichewa ˆ (8) zi-n´ a-w´ a-lum-a (↑ pred) = ‘bitesubj,obj’ ((↑ subj pred) = ‘pro’) (↑ subj nounclass) = 10 (↑ obj pred) = ‘pro’ (↑ obj nounclass) = 2
S VP ↑ =↓
(9) njˆ uchi zi-n´a-w´a-lum-a bees subj- past -obj-bite-indicative ‘The bees bit them.’ ⎡ pred S ⎢ ⎢ ⎢ subj ⎢ VP NP ⎢ ↑ =↓ (↑ subj)=↓ ⎢ ⎣ obj N ↑ =↓
njˆ uchi bees (↑ pred) = ‘bees’ (↑ nounclass) = 10
V ↑ =↓
V ↑ =↓ V ↑ =↓
⎤ ‘bitesubj,obj’ ⎥ ⎥ pred ‘bees’ ⎥ nounclass 10 ⎥ ⎥ ⎥ ⎦ pred ‘pro’ nounclass 2
⎡
pred
⎢ ⎢ ⎢ subj ⎢ ⎢ ⎢ ⎣ obj
⎤ ‘bitesubj,obj’ ⎥ pred ‘pro’ ⎥ ⎥ nounclass 10 ⎥ ⎥ ⎥ ⎦ pred ‘pro’ nounclass 2
zi-n´ a-w´ a-lum-a subj- past -obj-bite-indicative (↑ pred) = ‘bitesubj,obj’ ((↑ subj pred) = ‘pro’) (↑ subj nounclass) = 10 (↑ obj pred) = ‘pro’ (↑ obj nounclass) = 2
• Include the optional specification (↑ subj pred) = ‘pro’: well-formed f-structure
V ↑ =↓
• Do not include the optional specification (↑ subj pred) = ‘pro’: ungrammaticality due to incomplete f-structure (no pred for the subject)
zi-n´ a-w´ a-lum-a subj- past -obj-bite-indicative (↑ pred) = ‘bitesubj,obj’ ((↑ subj pred) = ‘pro’) (↑ subj nounclass) = 10 (↑ obj pred) = ‘pro’ (↑ obj nounclass) = 2 • Include the optional specification (↑ subj pred) = ‘pro’ in the fdescription: ungrammaticality due to incoherent f-structure (multiple specification of the pred of the subject, as ‘pro’ and as ‘bees’ • Do not include the optional specification (↑ subj pred) = ‘pro’: wellformed f-structure 3
4
Null anaphora and configurationality
(14)
IP ADJUNCT
IP
Jiwarli (Austin & Bresnan 1996): null anaphora together with nonconfigurationality, no evidence for VP, discontinuous nominal expressions, pragmatically-conditioned word order. Are these phenomena connected, and if so, should our theory of syntax reflect the connection?
IP
(NP-nom)
ADJUNCT I
NP-erg
Aux
(11) Karla wantha-nma-rni jarnpa juma. fire.abs give-imper-hence light.abs small.abs
0 (nom)
0 (acc)
NP-abs
VP
V
(NP-acc)
‘Give me a small fire light.’ First and second person pronouns are nominative/accusative, and thus appear in the specifier of IP (nominative subjects), or the complement of V (accusative objects); null third person nominative/accusative pronouns appear there too. Ergative/absolutive nominals appear in adjoined positions.
(12) Kutharra-rru ngunha ngurnta-inha jiluru. two.abs-now that.abs lie.pres egg.abs ‘Now those two eggs are lying there.’
Predictions of the pronominal-argument account: (13) Mantharta-lu kurrpirli-nha pinya-nyja yanga-rnu-ru. man-erg kangaroo-acc spear-past chase-imperfss-erg
1. First and second person nominative and accusative pronouns should appear in fixed positions in phrase structure.
‘The man speared the kangaroo as (he) chased (it).’ 2. Null anaphors should be restricted to third person. Two possible analyses:
3. Sentences with discontinuous and displaced nominal expressions should obey constraints on the distribution of null pronouns.
1. The ‘pronominal argument’ model (Jelinek 1984): the nominal phrases are actually modifiers, not arguments; the real arguments are pronominal, appearing as clitic pronouns or (possibly unpronounced) agreement markers. Adjuncts are optional and iterable, explaining why we get null anaphora, discontinuous arguments, and free word order – these phenomena are inextricably connected.
4. Ergative/absolutive nominals should show case compatibility patterns of modifiers. But (Austin & Bresnan 1996, 243 ff.): (15) First prediction is incorrect: VSO order is possible with first (and second) person nominative/accusative subject pronouns.
2. The ‘dual structure’ model (Kroeger 1993; Austin & Bresnan 1996): in some languages, phrase structure does not reflect grammatical functions; these languages use the nonprojecting category S for some or all clauses. There are no constraints on the possible grammatical functions of the daughters of S. Nominal phrases are actually arguments of the predicate, not modifiers. The possibility for null anaphora depends on characteristics of the verbs and other predicates of the language, and is not correlated with c-structure properties like word order and discontinuous arguments.
Jimpingka-rninyja ngatha-thu wirta-nyjarri-nha carry-past 1.nom-top boy-pl-acc ‘I carried the boys on my back.’
Pronominal argument model for Jiwarli, following Jelinek (1984) (Austin & Bresnan 1996, 247):
4
(16) Second prediction is incorrect: Null pronouns with first and second person interpretation are possible.
Ngatha-nha tharla-rnunyja warrara-lu thurnti-ngku I-acc feed-past wild.potato-erg vegetable.food-erg juma-nha-purra-thu. child-acc-time-top
Warri yana-artu ngatha kartaju-la. Yana-artu ngulha not go-usit I.nom night-loc go-usit NOTHING jurrinypi-rnu. Ngurru-martu-la kumpa-artu walk-abour-imperfss old.man-pauc-loc sit-usit
‘(They) used to feed me with wild potatoes when I was a child.’ But we do not see this casemarking distinction in discontinuous or displaced phrases in Jiwarli.
‘I never used to go in the night. (I) never used to go walking about. (I) used to live with the old men.’
Conclusions: As in Chichewa, ˆ agreement markers can serve a pronominal function; overt nominals are not modifiers but arguments of the verb when they appear. “Null anaphora” does not necessarily correlate with free word order and discontinuous argument constructions. See Austin & Bresnan (1996) for further discussion.
(17) Third prediction is incorrect. Null subject and object pronouns are not permitted in text-initial sentences. Nevertheless, word order is not fixed in text-initial sentences. The following sentences are text-initial:
Phrase structure consequences: There is an exocentric constituent in Jiwarli, Warlpiri, Latin, other free word order languages, which does not obey the constraints on the relation between phrase structure positions and grammatical functions that we see in X -theoretic categories. For Jiwarli, there is no evidence for clause-level functional categories or constituents such as VP, and the clause consists of S dominating any categories, with any grammatical function:
Tharrarrayil-pa-nha nhaa-lu ngulha parna bower.bird-phon-acc something-erg NOTHING head.abs puthi-rninyja. hit-past ‘Something hit bower bird on the head.’
(19)
(18) Fourth prediction is incorrect. Adjuncts show a three-way casemarking distinction. Adjuncts modifying an intransitive verb show unmarked absolutive case:
S
...
...
...
In Warlpiri, there is evidence for IP but not CP: Tharti yana-ma-rni. quickly.abs go-imper-hence
(20)
IP I
XP
‘Come quickly!’ Adjuncts modifying a transitive verb take ergative case: Tharti-ngku malha-nma ngunha kurtangara quickly-erg press.on-imper that.abs whirlwind.abs puni-ya-rni-rru. go-imperfds-hence-now
I
S
...
...
...
In other languages, there may be evidence for both CP and IP in addition to S; and some languages, e.g. English, do not make use of S at all. References
‘Press (them) down quickly because there is a whirlwind coming.’
Austin, Peter & Joan Bresnan. 1996. Non-configurationality in Australian Aboriginal languages. Natural Language and Linguistic Theory 14(2), pp. 215–268.
Crucially, temporal adjuncts modifying the transitive object bear accusative case:
Bresnan, Joan. 2001. Lexical-Functional Syntax . Oxford: Blackwell Publishers.
5
Bresnan, Joan & Sam A. Mchombo. 1987. Topic, pronoun, and agreement in Chichewa. ˆ Language 63(4), pp. 741–782. Jelinek, Eloise. 1984. Empty categories, case, and configurationality. Natural Language and Linguistic Theory 2(1), pp. 39–76. Kroeger, Paul. 1993. Phrase Structure and Grammatical Relations in Tagalog. Dissertations in Linguistics. Stanford, CA: CSLI Publications. Revised and corrected version of 1991 Stanford University dissertation. Nikolaeva, Irina. 1999. Object agreement, grammatical relations, and information structure. Studies in Language 23, pp. 331–376.
6
