7#,-;  ffffrbrbrbrrvr|sdsd*sxg t t&uP*uzCf uPt<uPuP%uPuPuPuPuPuPMinimalist Syntax & Comparative Grammar University of Maryland, College Park 12th May 1994 Agreement Alternations Colin Phillips MIT Introduction: On Doing without D-Structure D-Structure can be dispensed with if conditions on D-Structure can be (i) dropped or (ii) moved to another component. What D-Structure conditions might there be? a. Pure GF-q relations (Chomsky 1986)trace theory made this unnecessary long ago (cf. Koster 1978, Kitagawa 1986) b. Conditions on choice of lexical itemscan all bad choices be ruled out by other components of grammar (PF, LF, derivation)? Dispensing with (a) is straightforward: theta-theory applies at LF. Dispensing with (b) is not so obvious. Is it possible to freely select any array of lexical items from the lexicon? Are there cases where an expression is impossible not because it is phonologically ill-formed, not because it is semantically ill-formed, not because there is no convergent derivation, but purely because a different array of items should have been selected at the outset? If so, we are forced to impose D-Structure conditions. Test CaseComplex Inflectional Systems Yimas (Foley 1991): Lower Sepik family, northern New Guinea. 250 speakers, rapidly dying. Extremely complex verbal inflectional systemsplit ergative, with many additional complications. 5 CasesNominative, Accusative, Ergative, Absolutive, Dative 4 NumbersSingular, Dual, Plural, Paucal (3-7 humans) 16 Genderssee below Inflection for an argument varies according to which other inflections are present. What does a child need to know to learn a complex inflectional system? just fill in slots in large paradigm can sytactic knowledge help? Alternation I: EPP driven alternation Yimas shows typical polysynthetic properties: free ordering of independent phrases liberal omission of independent phrases affixes on verb reference all arguments (no case marking on NPs) Person-based split-ergative system. Ergative/Absolutive system for 3rd persons: (1) a. pu- n- tay 3rd person object 3pl-abs 3sg-Erg see He saw them. b. pu- wa -t 3rd person intransitive subject 3pl-Abs go Perf They went. c. na- mpu- tay 3rd person transitive subject 3sg-Abs 3pl-Erg see They saw him. Nominative/Accusative system for 1st/2nd persons: (2) a. ant- ka- tmuk -t 1st person intransitive subject poss 1sg-Nom fall perf. I almost fell down. b. pu- ka- tay 1st person transitive subject 3pl-Abs 1sg-nom see I saw them. c. pu- Na- tay 1st person object 3pl-abs 1sg-Acc see They saw me. Every Yimas verb (affirmative) must contain an Absolutive marker. This requirement overrides other principles of agreement marking. (3) a. pu- nan- tay 3rd person transitive subject 3pl-Abs 2sg-Acc see erg abs They saw you. b. kapwa-Nkra- tay 2nd person transitive subject 2dl-Abs 1dl-Acc see nom abs You two saw us two. (4) a. Nominative & Accusative are Arguments, result of incorporation. b. Ergative & Absolutive are Agreement, product of Spec-Head relation. (see Phillips 1993, to appear b)  Requirement for Absolutive affix is reflex of Extended Projection Principle. (6) Extended Projection Principle (first try) Agr10 must agree with a nominal in its specifier Alternation II: EPP again If one of the complementizer prefixes in (7) is present, the requirement for an Absolutive marker no longer applies. (7) m- Relative clause complementizer ta- Negation ka- Likely modality ant- Potential modality 3rd person transitive subject reverts to its default Ergative (8) a. pu- nan- tay 3pl-Abs 2sg-Acc see erg abs They saw you. b. ka- mpu- Na- tput-n likely 3pl-erg 1sg-acc hit pres abs erg They are going to hit me. 2nd person subject reverts to its default Nominative (9) a. kapwa-Nkra- tay 2dl-Abs 1dl-Acc see nom abs You two saw us two. b. NarN ka- n- wa-kia-k 1-day-removed like 2sg-nom go NrFut irr abs nom It is likely that you will go tomorrow. 3rd person objects are unaffected their default is Absolutive (10) ta- pu- n- tay-c -um neg 3abs 3sg-erg see perf pl He didnt see them. Note: This shows that the changes in (8-9) are not the result of competition between the Complementizers and the Absolutive prefixes for a single position. The Complementizers appear to be doing the work of the EPP in (8-9). Assume they are nominal heads (most of them appear with nominal agreement at the end of the verb). (11) Extended Projection Principle Agr10 must agree with a nominal  Forms with complementizer prefixes allow us to see how arguments behave when they are not constrained by the EPP. Question: What is the default for intransitive subjects ergative or absolutive? Two positions in ergativity literature: (i) Special condition on intransitives determines their case Levin & Massam (1985); Campana (1992); Bobaljik (1993); Chomsky (1993); Laka (1993); Massam (1993); Sityar (1993) (ii) Independent condition determines case of intransitives (eg. EPP) Murasugi (1992, 1993); Bittner & Hale (1993); Phillips (1993, to appear a); Bittner (1994) Crossing vs. Nested Paths derivations (cf. Murasugi 1992, 1993; Campana 1992)  Answer: once the effects of the EPP are controlled for, intransitive subjects are marked Ergative just like transitive subjects. Ergative case pattern is an artifact of EPP Ergative is preferred over Absolutive because it is derivationally cheaper, as Nested Paths theory predicts. Its actually not very easy to determine how intransitive subjects behave when complementizers are added to the front of the verb. Some of the complementizers cause deletion of the left most agreement prefix, so that we can only infer the behaviour of the intransitive subject by backwards reasoning from the form of other morphemes in the verb. ka- is the one complementizer which does not regularly lead to blocking of the leftmost agreement prefix. ka- likely modality (13) balus-an ka- Nkl- ya -ka -arm -n airplane-Obl likely 3pl-erg come seq board pres Those few will board the plane now. ta- negation & ant- potential modality With these complementizer prefixes it is impossible to see directly whether an intransitive subject is marked ergative or absolutive, due to competition with the complementizer agreement suffixes which these prefixes introduce. For most numbers, the .i.complementizer agreement ;suffixes do not encode case differences, i.e. the form of agreement for absolutive dual is identical to the form for ergative dual, as in (14). (14) a. ta- mpu- tpul -c -rm Agreement with Absolutive neg 3pl-erg hit perf dl They didnt hit those two. b. ta- pu- nan- tpul -c -rm Agreement with Ergative neg 3erg 2pl-acc see perf dl Those two didnt hit you. Case differences are encoded for singulars, however: singular transitive objects (absolutive) are marked -ak, whereas singular transitive subjects (ergative) are marked by zero agreement (15a-b; Foley, p.252-256). In this respect, the intransitive subject in (15c) patterns just like the transitive subject in (15b). (15) a. ta- - mpu- tay -c -ak transitive object (ABS) neg 3sg-abs 3pl-erg see perf sing They didnt see him. b. ta- - kra- tpul - transitive subject (ERG) neg 3sg-erg 1pl-acc hit sing He didnt hit us. c. anan- - mal - intransitive subject (ERG) poss 3 die sing He almost died. m- relative clause complementizer In this case, the evidence comes from wh-question formation. A clefting strategy is used for extraction of transitive or intransitive subjects (16a-b), but is not required for object extraction (16c)(Foley, p.430). The clefting strategy, which supplies the C-prefix m-, is used as a last resort when there is no potential absolutive marker. Again the parallel between transitive and intransitive subject extraction indicates that 3rd person intransitive subjects are being treated as ergatives. (16) a. nawm m- - kul- cpul -um? transitive subject question who-pl comp 3pl-erg 2pl-acc hit pl Who hit you all? b. nawn m- - na- ya -n -? intransitive subject question who-sg comp 3 def come pres sing Who is coming? c. nawn impa- - tpul? transitive object question who-sg 3sg-abs 3sg-erg hit Who did he hit? Alternation III: Competition for Verbal Case Maximum of one nominative/accusative affix in any verb. give example of 2dlNom; is there evidence that 2nd persons are marked on vb. when theres a 1st person object? (17) kapwa- Nkra- tay 2dl-abs 1dl-acc see You two saw us two. (18) a. Why do 1st and 2nd person arguments incorporate into the verb? Incorporation into V0 satisfies an arguments Case requirement. See Ferguson (1993); Ferguson & Groat (1994) for similar conclusion, driven by independent considerations. b. Why do only 1st and 2nd person arguments incorporate? Yimas verbs are extremely picky Case assigners. Verbs that license any kind of argumentNoun Incorporation language (cf. Baker 1988) Verbs that license specific q-rolesinherent Case Verbs that license only indefinitespartitive Case (Italian: Belletti 1988) Verbs that license only f-featuresonly pronouns incorporate (eg. Chichewa: Bresnan & Mchombo 1987) Yimas: verbs are one step more picky than Chichewa verbsthey license only person & number features. 3rd Person Absolutive Singular prefixes: Class Membership Criteria Absolutive Singular Prefix Class I Humans, unmarked na- Class II -ma, female humans na- Class III Higher animals na- Class IV -um, plants & trees mu- Class V -p, -k, -m, -n, -, -nt, -r, -l na- Class VI -k k- Class VII -mp p- Class VIII -i, -y i- Class IX -aw wa- Class X -k ku- etc. Class XVI contains just awi lime ya- Gender featureswhich encode semantic and phonological informationprevent 3rd person arguments from being licensed by V0. c. Why can only one argument incorporate? A Case assigner can license maximally one argument. This distinguishes my account from that in Baker (1988, 105ff), which assumed that incorporation exempts a noun from the Case Filter. On this account, multiple incorporation could exempt multiple arguments from the Case Filter. d. Why do patients win over agents? Derivations proceed cyclically (Chomsky 1993) objects get the first opportunity to satisfy their Case requirement. Incorporation is always preferred where possible, since it involves more economical movement. Why Subjects Cant Incorporate Generally Subject Incorporation i.e. Nominative only available in Yimas when either: i. Subject is sole argument ii. Lower arguments disqualified from receiving Verbal Case (19) * ipa- n- tay object should have incorporated 1pl-abs 2sg-nom see (20) a. ta- ka- wa -t subject is lowest argument neg 1sg-nom go perf I didnt go. b. pu- ka- tay object is not candidate for verbal Case 3pl-abs 1sg-nom see I saw them. Agent Incorporation unavailable in languages allowing Patient Incorporation (21) a. Volu nakai he tau fanau e fua niu? grate Q Erg-pl-children Abs-fruit coconut Are the children grating (the fruit of the) coconut? b. Volu niu nakai e tau fanau? Patient Incorporation grate-coconut Q Abs-pl-children Are the children grating coconut? (22) a. Fa totou he tau faiaoga e tau tohi Hab-read Erg-pl-teacher Abs-pl-book (The) teachers often read books. b.* Fa totou faiaoga e tau tohi *Agent Incorporation Hab-read-teacher Abs-pl-book Teachers often read books. (Niuean: Seiter 1980) Unaccusatives can incorporate, Unergatives cannot. (23) a. Ka-hi-hw-i neo-hsahet-a 3N-spill-caus-asp the Pre-bean-suf The beans spilled. b. Ka-hsahet-ahi-hw-i Patient Incorporation 3N-bean-spill-caus-asp The beans spilled. (24) a. Khwien-ide -teurawe-we dog-suf A-run-pres The dog is running. b.* -khwien-teurawe-we *Agent Incorporation A-dog-run-pres The dog is running. (Southern Tiwa: Allen et al. 1984) Bakers (1988) account of ban on subject incorporation: Lowering from [Spec,IP] violates ECP (25a) Under modified assumptions about the relevant structures, the structural asymmetry between subjects and objects evaporates: Agents start out in [Spec,VP] Spec-Head relation is core symmetrical relation of theory X0 = XP for non-branching phrases (Chomsky 1994; others)  Subject Incorporation is Unavailable because Objects get there first Note: In order to account for why subjects of unergative verbs do not incorporate I am forced to assume that all unergatives are underlyingly transitive (Hale & Keyser 1993). The Ergativity Conspiracy Economy without Greed 1st and 2nd person arguments can be Case-licensed in two ways: (i) Incorporate into V0 (ii) Raise to [Spec,AgrP] Why is the choice between (i) and (ii) never optional? (26) a. ka- n- wa-kia -k incorporate for Case likely 2sg-nom go NrFut irr You will probably go. b.* ka- ma- wa-kia -k raise to [Spec,AgrP] likely 2sg-abs go NrFut irr Preference for incorporation is due to the the economy condition in (15). (27) Relativized Economy Satisfy requirements using the shortest available movement. i. Requirement of source: move source to closest appropriate target. ii. Requirement of target: move closest appropriate source to target. I: If argument needs to move to satisfy its own requirement move to closest available position.  (28) [Agr1P Abs [Agr2P Erg [VP 2-sg V ]]]  II: Argument must move to satisfy requirement of target, eg. EPP. Neither argument in (29a) needs to move to [Spec,Agr1] (absolutive) for itself. (29) a. ka- mpu- Na- tput-n likely 3pl-erg 1sg-acc hit pres They are going to hit me. When one argument must move to satisfy EPP which one moves? b. pu- Na- tay 3pl-abs 1sg-acc see They saw me. Closest argument to target moves, as predicted by Relativized Economy.  c. [Agr1P Abs [ Agr2P Erg [VP Agent Theme V ]]]  Generalized Visibility Theory of Case Shlonsky (1987) & Baker (1991) Baker (1991) shows that Mohawk shows a lack of asymmetries between subject and object NPs for Condition C effects, extraction from NP (CED, Huang 1982), and weak crossover. Remarkably, the asymmetries found in English reemerge in Mohawk when clausal or phonologically empty arguments are involved. Baker therefore assumes that (i) clausal and phonologically null arguments do not require Case at S-structure; (ii) only arguments which do not require Case at S-structure may occupy A-positions in Mohawk. (30) PF Features (phonetic content) Case required at S-structure LF Features (theta-role) Case required at LF This is a unification of Rouveret & Vergnauds (1980) PF theory of Case, and the LF Visibility theory of Case (Aoun 1979; Lasnik & Freidin 1981). (31) Generalized Visibility X can be interpreted at level a only if X receives Case at level a. Assumption: Agreement morphemes require case in polysynthetic languages (modified from Baker 1991) Problem: how can Spec-Head relation Case-license both participants?  Solution: ensure that different elements require Case at different levels of representation (33) little pro theorem a. Overt Nominal arguments require Case licensing by Spec-Head agreement. b. Case absorption by X0s makes Case unavailable to specifiers. \ Only non-overt nominals may appear in specifier positions. The Ergative-Accusative Parameter Scenario I: Agreement heads do not require Case (eg. Icelandic) \ Shortest movement computed from perspective of overt arguments Accusative system arises  Scenario II: Agreement heads both require Case (Yimas) \ Arguments must be pro Shortest movement computed from perspective of agreement heads Ergative system arises  How do Objects Cross Subjects? Anti-Agreement affects subject and object extraction in Yimas: (36) a. nawm m- - kul- cpul -um? transitive subject question who-pl Comp 3pl-Erg 2pl-Acc hit PLUR Who hit you all? b. nawn impa- tpul? transitive object question who-sg 3dl-Abs hit Who did those two hit? Anti-Agreement with subject extraction familiar from Celtic, Romance, Arabic etc. (Ouhalla 1993). (37) a. Petore paotred a lenne (*lennent) al levrio which boys Comp read read:3pl the books Which boys read the books? (Breton: Borseley & Stephens 1989) b. Quante ragazze gli ha parlato con te? How many girls it has spoken to you? (Fiorentino: Brandi & Cordin 1989) Ouhalla (1993): Intervening A-bar specifier, eg. [Spec,NegP], blocks Anti-Agreement (38) a. Ar vugale ne (*lenne) lennent ket al levriou the children NEG read read-3pl NEG the books The children who did not read the books. (Breton: Hendrick 1988) b. tamghart nni ur t -ssn Mohand woman Comp Neg 3fs-know M. The woman who does not know Mohand. (Berber: Ouhalla 1993) Ouhallas explanation: (i) assume that empty categories are functionally determined; (ii) rich agreement licenses pro; (iii) pro is subject to A-bar Disjointness Condition (ABDR: Aoun & Li 1990, 1993), which requires that a pronominal not be bound by the closest A-bar binder. If rich agreement licenses pro as the tail of a subject extraction chain, the ABDR is violated. Impoverishing agreement forces subject extraction to leave a trace. When an A-bar specifier like negation intervenes between [Spec,IP] and [Spec,CP], the ABDR becomes irrelevant, so impoverishment of agreement is unnecessary. See Phillips (to appear a) for discussion of this account and an alternative analysis. Assume anti-anti-agreement is a diagnostic for an intervening A-bar position. Anti-anti-agreement found in Yimas when subject extracted across object in [Spec,Agr1]. (39) nawrm na- mpi -tpul? who-dl 3sg-Abs 3dl-Erg hit Who hit him? (Yimas: F431) Absolutive objects block Anti-Agreement for subject extraction Object is assigned Absolutive above Ergative Projection Absolutive is assigned to A-bar position  See Miyagawa (1993) for arguments from the interaction of quantifier scope and case marking that A-bar movement for Case is possible as a last resort. Bittner (1992) and Campana (1992) claimed that Absolutive is assigned to an A-bar position. Since binding conditions are generally assumed to be blind to A-bar movement, assuming that absolutive is assigned to an A-bar position answers some criticisms of the Nested Paths derivation of ergative/absolutive clauses. i. Subjects asymmetrically bind objects in ergative languages (Anderson 1976) ii. Weak crossover effects identical to accusative languages (Nishga: Belvin 1985, Urbanczyk 1992; Inuktitut: Bobaljik 1993) iii. A-movement skipping both [Spec,VP] and [Spec,Agr2] is impossible (Chomsky 1993) Answer: [Spec,Agr2] is the highest A-position in the clause Alternation IV: Effects of Complementizer Agreement Complementizer prefixes all satisfy EPP (41) a. ka- mpu- Na- tput-n likely 3pl-erg 1sg-acc hit pres abs erg They are going to hit me. b. NarN ka- n- wa-kia-k 1-day-removed like 2sg-nom go NrFut irr abs nom It is likely that you will go tomorrow. Not all complementizers lead to deletion of leftmost agreement. Generalization: only agreeing complementizers trigger deletion of agreement Agreement deletes Absolutive morpheme (42) a. na- mpi- tpul 3sg-Abs 3dl-Erg hit Those two hit him b. ta- mpi- tpul -c -ak NEG 3sg-Abs 3dl-Erg hit perf SING Those two didnt hit him Agreement deletes Ergative morpheme (43) a. ka- mpu- a- tput -n likely 3pl-Erg 1sg-Acc hit pres They are going to hit me b. ta- a- tay -c -um NEG 3pl-Erg 1sg-Acc see perf PLUR They didnt see me Agreement doesnt delete Nominative or Accusative morphemes (44) a. ka- nan- wa -kia -k likely 2pl-Nom go nr.fut. irr. You all will probably go b. ta- nan- wa -r -um NEG 2pl-Nom go perf PLUR You all didnt go (45) c. ipa kampan-tay 1pl 1/2sg-Acc see We saw you. d. ipa ta- mpan- tpul - 1pl Neg 1/2sg-Acc hit SG We didnt hit you two.  Differential Blocking is sensitive to the Agreement/Argument contrast I assume to underlie the split-ergative inflectional system (47) No Multiple Spell-Out In the domain of a word, do not insert two items which realize the same feature. Alternative Accounts Differential blocking effect not restricted to 3rd person agreement check (48) a. ipwa- kra-tay 2pl-Abs 1pl-Acc see You all saw us. b. ta- - kra- tpul-c -um Neg 2pl-Abs 1pl-Acc hit Perf PL You all didnt hit us. Differential blocking not reducible to structural contrast  Differential blocking not an effect of feature movement some features can still show up in the prefix position. (50) ta- pu- n- tay-c -um neg 3abs 3sg-erg see perf pl He didnt see them. ********************** ...popping up a level now... ********************** Why Alternations Matter In a theory without conditions on D-Structure, there should be no constraints on the choice of the initial array (Chomsky 1993) or enumeration (Chomsky 1994). Not all arrays will yield a possible expression, of coursebut bad arrays should be bad because: (i) no convergent derivation possible (ii) derivation yields semantic gibberish (iii) derivation yields phonological gibberish An array should never be bad for no other reason than that it is a bad array. If this does happen, we are forced to impose conditions on arrays this amounts to a reintroduction of D-Structure conditions. Core of Checking Theory a. Heads are inserted into derivations fully specified with phonological form and inflectional material already attached. b. Derivations merely check that these heads are appropriate, by matching features with the features of other heads. Checking Account of Alternation II Generalization: intransitive subjects are marked by ergative or absolutive agreementall other things being equal ergative is the preferred choice, since it involves a less costly derivation. Two possibilities in an array in which the C-prefix ka- is chosen.  a. Choose pro with absolutive features as well as f-features; choose Agr1 with f-features. Subject moves to [Spec,Agr1], matches features with Agr1. Derivation converges. b. Choose pro with ergative features as well as f-features; choose Agr2 with f-features. Subject moves to [Spec,Agr2], matches features with Agr2. Derivation converges. But: array (a) doesnt yield a possible expression it is blocked by array (b). Checking Account of Alternation III Generalization: 2nd person subjects either incorporate into V, in which case they are overt, or they agree with a functional head, in which case they are phonologically null pro. Incorporation is chosen where possible, since it involves a less costly derivation. Two possibilities in an array in which a 3rd person object is chosen.  a. Choose /----/ as the phonological form of the second person subject, i.e. pro, and f-features for Agr2. Subject moves to [Spec,Agr2], matches features with Agr2. Derivation converges. b. Choose /blah/ as the phonological form of the second person subject. Subject incorporates, satisfying its own case requirement at PF. Derivation converges. But: array (b) doesnt yield a possible expression it is blocked by array (a). Checking Account of Alternation IV Generalization: 3rd person object agreement absolutive realizes either (a) [person, number, gender] features or (b) [person, gender (?)] features. Number features are realized wherever possible (i.e. when not blocked by number agreement suffix). Two possibilities in an array in which a C-prefix is not chosen. a. Choose [person, gender] features on the object and on Agr1. Object moves to [Spec,Agr1], matches features with Agr1. Derivation converges. b. Choose [person, number gender] features on the object and on Agr1. Object moves to [Spec,Agr1], matches features with Agr1. Derivation converges. But: array (a) doesnt yield a possible expressionit is blocked by array (b). Note: We could not account for this by assuming that derivation (a) crashes because number features must always be realizedbecause number features dont always have to be realized Why Checking Theory? Each of the alternations appears to lead to the same conclusion, when coupled with the Checking Theory of Chomsky (1993). Different arrays of lexical items compete with one another i.e. there are conditions on D-Structure. 2 possible motivations for Checking theory: (i) Feature matching ensures that elements move to the right target. (ii) Pre-assembly of complex heads ensures compatibility between PF & LF, in cases where the crucial movements take place after the derivation splits. Its hard to find cases which rely on (i). Appropriate movements are usually independently forced by the theory of movement. eg. a. Icelandic is an accusative language rather than an ergativelanguage, because of how case requirements are satisfied by PF. b. Head movement example??? We do find fairly convincing cases which rely on (ii). These are cases where a certain form is licensed by a configuration which does not hold at spell-out (or earlier). eg. a. English ECMsubject of tenseless clause case licensed in higher clause. b. Japanese ga/no conversion subject of tensed complement clause may be case licensed in higher clause (but does not have to be). c. Mohawk agreementsubject and object agreement is overtly realized, but arguments do not raise out of VP overtly (Baker 1991, 1993). The only motivation for checking theory, then, is to ensure LF/PF compatibility without imposing complex S-Structure conditions. Checking Theory does not play a role in the derivation before spell-out. If We Really Dont Want D-Structure... The account of alternations II-IV is intuitively as follows: in each case, syntactic conditions should choose between the two possible arrays. We can allow this by making less initial committments than Checking Theory would have us make. Avoid competition between different arrays by neutralizaing differences between arrays Alternation II: Generalization: intransitive subjects are marked by ergative or absolutive agreementall other things being equal ergative is the preferred choice, since it involves a less costly derivation. a. NP, {3dual, class VI, absolutive} Agr1{3dual, class VI} b. NP, {3dual, class VI, ergative} Agr2{3dual, class VI} Both Agr1 and Agr2 are chosen. They have case-licensing properties, but their f-features are initially underdetermined. Derivation: the subject pro moves overtly to either [Spec,Agr1] or [Spec,Agr2], where it copies its f-features onto the head no matching of features takes place. The head bearing phi-features is realized as overt agreement. Since the initial array is the same for both derivations, the shorter derivation movement to [Spec,Agr2] can block the longer derivation in the appropriate contexts. Alternation III: Generalization: 2nd person subjects either incorporate into V, in which case they are overt, or they agree with a functional head, in which case they are phonologically null pro. Incorporation is chosen where possible, since it involves a less costly derivation. a. NP, {2sing, ergative}, /empty/ b. NP, {2sing, nominative}, /blah/ At the stage where the array is chosen, there is no difference between overt and null 2nd person arguments. Derivation: the 2nd person subject either incorporates into the verb or moves to [Spec,Agr2]. The shorter derivation blocks the longer one when it is available. Phonological features are assigned post-syntactically (Halle & Marantz 1993, others...). Alternation IV: Generalization: 3rd person object agreementabsolutiverealizes either (a) [person, number, gender] features or (b) [person, gender (?)] features. Number features are realized wherever possible (i.e. when not blocked by number agreement suffix). a. Agr1, {3rd person, class I, dual} b. Agr1, {3rd person, class I} Derivation: [person, number, gender] features are always selected at the start of the derivation. The ultimate realization of these features has no effect on the course of the syntactic derivation. When there is no complementizer agreement suffix, two phonological forms are candidates for insertion in Agr1. Competition principles as in Halle & Marantz (1993) apply, ensuring that the more specific phonological form is used. How are the Benefits of Checking Retained? We already assume that items may not be inserted into the derivation between Spell-Out and LF, in order to prevent: Bill likes Hillary = Bill likes Hillary and Socks does too. If the ban on insertion also applies to feature copying between heads and specifiers, we ensure that all features relevant to both LF and PF are present by Spell-Out. Checking may still occur between Spell-Out and LF, which is the only part of the derivation where it is required. A Prediction Alternations like Alternation II in Yimas are always resolved by overt syntax. i.e. a language Yimas-prime is impossible, in which Alternation II is found, but arguments do not move until LF.  Feature checking is involved in the mapping to LF, and the two derivations involve different sets of features therefore, they should not be in competition. Note: This does not mean that there will be no language in which similar pairs of possible derivations are possible at LFwhat is impossible is that one of the derivations blocks the other, unlike in Yimas. eg. Japanese ga/no conversion subject of tensed complement of N can be marked nominative (remains inside complement clause) or genitive (raises to higher clause). Movement for genitive takes place at LF (Miyagawa 1993, Ura 1993). It is less economical than the derivation in which nominative is assigned, but it is not blocked by the shorter derivation. Conclusion: What does the Yimas Child Need to Know? Knows incorporation satisfies case-requirement Learns verbal case is very picky in Yimas Knows economy takes account of what movement achieves Learns agreement heads want case in Yimas Knows lexical items are not inserted into syntax fully inflected References Allen, B, D. Gardiner & D. Frantz (1984) Noun Incorporation in Southern Tiwa. IJAL 50. Baker, Mark (1988) Incorporation. Chicago: University of Chicago Press Baker, Mark (1991) On some subject-object non-asymmetries in Mohawk. NLLT 9. Baker, Mark (1993) The Polysynthesis Parameter. ms. McGill U. Bobaljik, Jonathan. 1993. Ergativity and Ergative Unergatives. In: C. Phillips (ed), Papers on Case & Agreement II. MITWPL 19. Borseley, Robert & Janig Stephens (1989) Agreement and the Position of Subjects in Breton. NLLT 7. Brandi, Luciana, & Patrizia Cordin (1989) Two Italian Dialects and the Null Subject Parameter. In: Jaeggli & Safir (eds.) The Null Subject Parameter. Kluwer. Chomsky, Noam (1993) A Minimalist Program for Linguistic Theory. In: Hale & Keyser (eds), The View from Building 20. Cambridge, MA: MIT Press. Chomsky, Noam. 1994. Bare Phrase Structure. MIT Occasional Papers in Linguistics #4. Diesing, Molly (1990) Verb Movement and the Subject Position in Yiddish. NLLT 8. Foley, William (1991) The Yimas Language of New Guinea. Stanford: Stanford UP. Hale, Ken & Jay Keyser (1991) The Syntax of Argument Structure. ms, MIT Hendrick, Randall (1988) Anaphora in Celtic and Universal Grammar. Kluwer. Jelinek, Eloise (1984) Empty Categories, Case and Configurationality. NLLT 2. Miyagawa, Shigeru. 1993. LF Case-checking and Minimal Link Condition. In: C. Phillips (ed), Papers on Case & Agreement II. MITWPL 19. Murasugi, Kumiko (1992) Crossing & Nested Paths: NP Movement in Accusative and Ergative Languages. PhD thesis, MIT. Ouhalla, Jamal (1993) Subject extraction, Negation, Anti-Agreement. NLLT 11. Phillips, Colin (1993) Conditions on Agreement in Yimas. In: MITWPL #18, Papers on Case & Agreement I, 173-213. Phillips, Colin (to appear a) Ergative Subjects. 6th Biennial Grammaticsl Relations Conference. Stanford: CSLI. Phillips, Colin (to appear b) Verbal Case and the Nature of Polysynthetic Inflection. Proceedings of CONSOLE II. Leiden: SOLE. Seiter, W. (1980) Studies in Niuean Syntax. Garland Press. Shlonsky, Ur (1987) Null & Displaced Subjects. PhD thesis, MIT. Ura, Hiroyuki. 1993. L-Relatedness and its Parametric Variation. In: C. Phillips (ed), Papers on Case & Agreement II. MITWPL 19. Department of Linguistics & Philosophy 20D-219, MIT Cambridge, Mass., 02139 cphill@mit.edu U. Maryland, 5/12/94 Phillips/Agreement Alternations  uvwx aced WORD  ,Times .+]:Abs +<0Erg )=VP +V' +V (agent +theme (R~pro (#Dpro"0q## #"H## #"`## #"w## #"## #"T## # +Nom0" +Acc0"0/[=q#0_o# ( _Agr1P +Agr1' +Agr2P +Agr2'"7##3#####"#",##0#####"# ( (5)U!P"P"{d WORD{ {">R"##"#"$"##"# ,Times .+9 Agr1P +0Agr1' (H>Agr1  + (Abs)  (*proa%9%0"47##.#aO?cS0## +MEPP"\j"##"#"=<"##"# (*QAgr1P +0Agr1' (I Agr1 (qEPP"?"##"# (+C (JC +t ( C'  +CaAgr1" 0##1#aEY0"T##U#aE Y 0##  (b. (a. G\ 9\ )d WORD  ,Times .+jVP +V' +V"0R## #"Hn## #"`## #"w## #"## #"5## #aG[[o0"Ve#a#.70",-# ( @Agr1P + Agr1' +Agr2P +Agr2'  ( 12a.  (9Abs +<0Erg )=VP +V' +V  (oagent  (theme  (Q3pro ("pro"/&## #"GB## #"_^## #"vy## #"## #" ## #`+<}#a-0"(#aG3[G0"Z=# ( Agr1P + Agr1' +Agr2P +Agr2'`"Q *.t (t +t +t  (agent  +theme  (:AAgr1 +;0Agr2 (#Nom +=.Acc`d"Qm`+|# +aCrossing Paths (c Nested Paths  ( b.}ooo_d WORD  ,Times .+=VP +V' +V">## #"3Z## #"Kt## #"c## # ( QIP +I' (=SI + t (V (N"6#"y ## #  + i (Mi  (3IV' +V")## #"A4## # +V (aON"W[ ## #  + i  (,VP (;1t (%,NP"(2#  + i  ( (25) +b.  + &NP/N  ( a.] O% ?% d WORD%  % "##/#####" # ,Times .+ short , Zapf Dingbats (+C5%d WORD "#! 8#####"8# ,Times .+olong"r_Q%A%d WORD% %"#! 8#####"8# ,Times .+v short , Zapf Dingbats (+* d WORD " #! 8#####"8# ,Times .+ulong\\z d WORDz  z "="##"#""##"# ,Times .+ AgrP" +0Agr'") (GAgr"G ()xpro")a$u80"37##.#aCW0##  +LCase"u  ( (32):d WORD:  :,Times .+kVP +V' +V"1## #"I## #"a## #"x## #"## #"q## #aH\0"W#a_/s0"-i# ( |Agr1P + Agr1' +Agr2P +Agr2'  ( (34)  +theme  (t +t  (agent  (;}Agr1 +;0Agr2 ($ZNom +=.Acc`&"R`h.# +aCrossing Pathsad WORDa  a,Times .+{:Abs +<0Erg )=VP +V' +V (agent +theme (Rpro (#bpro"0# "H# "`# "w# "# "r# `,~`l`a`1z0"*maE_0"[ ( }Agr1P + Agr1' +Agr2P +Agr2' ( (35) + Nested Paths m_Od WORD "z#"<#"Z#"y#"#":# ,Times .+ CP +#C' (FC +EAgr1P +'Agr1' (Agr1 +RAgr2P +*Agr2' ( Agr2 +UVP (subject (dobjectam90 z2z"!z (no anti-agreement  ( (40)mVH8d WORD "nr###"V# ,Times .+ Neg (x0Neg )# (0Abs (w# +Gender"lBm##\_#"#"I##28# (xTPerson )V (SAbs" #ap0##  +8G Agreement  ( (46)ap0" #BBSYd WORDSY SY"%7#5" ,Times .+/Neg )#"'47#5"k (1Neg )# (2cAbs (0Nom"4Q##"8##  (Osblocking (L no blocking  (bNeg ( Neg ( (49) +b. ( a.BwB4wB$d WORDҡ  ,Times .+jVP +V' +V"0R## #"Hn## #"`## #"w## #"## #"5## #a#.70",-# ( @Agr1P + Agr1' +Agr2P +Agr2'  ( 12a.  +=.Abs (iNErg )=VP +V' +V  (oagent  (Q3pro ("#pro"/&## #"GB## #"_^## #"vy## #"## #" ## #`+<}#aG3[G0"Z=# ( Agr1P + Agr1' +Agr2P +Agr2' *.t (t  +agent`+|"  ( b.sU2eU2Uqd WORDq  q,Times .(;@VP +V' +V"## #"1## #"H.## #"`G## # ( Agr2P +Agr2'  ( 12a.  +V/Erg )=VP +V' +V  (Sxagent  (#<pro"K## #"1g## #"H## #"`## #`EV#a<-P0",F# ( WAgr2P +Agr2' *.t )t  +agent  ( b.`G$m" aalu0"kq#a'wBwB d WORDҡ  ,Times .+jVP +V' +V"0R## #"Hn## #"`## #"w## #"## #"5## # """"a2<0")+ ( @Agr1P + Agr1' +Agr2P +Agr2'  ( 12a.  +=.Abs (iNErg )=VP +V' +V  (oagent  (Q3pro ("#pro "/&## #"GB## #"_^## #"vy## #"## #" ## # """"`+<}aB.bN0"Z= ( Agr1P + Agr1' +Agr2P +Agr2' *.t (t  +agent`*|  ( b.  (i0LF (y3LF s of grammar (PF, LF, convergenceeasy8[\scdMw     " $ + F H X w z 2 E G ] |    9 : ; LľľľIJľľĸľľIJĸľľIJĸľľľĬ @  D @  @ @      $$ D L ] d g j q    2 O R Y Z a f j l p   X a ()UXq$';?S   @     D   @J")16?ADe<AHLZ[\_56w~  [\з H  @     D  @  H\]KMOdop  #9t|58QR %(4;@BV_NQ#^`f}ýýýɱë     D  @ @   @  @  HC#+6:i !!""""6"<"J"Q"Y"a"c"""""""""#E#_#f#|#####$$w$x$~$$$$$$$%%%&%S%%ܾܾ     @  @    @ D  F%%&2&E&H&&'''i'(O(Q(](`(a(r(t((((((((((((((((((((((((((((((()) )))))!)#)')()4)7);)=)H)I)J)K)O)Q)r)u)))))*c**+H+,G,J,t-#    @r  @ @T-#-G-f-j-y------..,./.>.V.......///G/O/l///////00 000S0c0o00001111,1-1/14151618191@1E1W1[11111111111112222;2I2M2a2g2n23 @r r   @ T33334+41454645.515J5L5d556666>6R6]6d6h6q6666666767;7N777788J8M8888888888899,929c9}9~99999:: :!ϩϩ  H  H  @    H4    @B:!:3:B:[:::::::::::::::::;;B;C<6<==b=c====>E>F>K>c>>>>>>>?T?U???@.@/@W@Y@@@@@AW   H @  @   H @   HX@  =AWAXAAAABBB B'BiBBBBC1CKCCE#EzFFGGG!G$GGHWH\IUIJbJJKKKNINLNNNNNNNNNNNNNNOOOO(O)O.O/O1¼¼¼¼ԯԩԩԯ  D   H @   @   Hl @   H>O1O2OLOMOqOvO}OOOOOOPP"PgPqP|PPQ:Q?QEQLQlQpQQQQQQRRRbRcRRRRRRRRRRSSSSSSSTTTT8T<TTTTTTTTTUUUUUUַ H| r   @   D  @  HUUUUV VRVWVYV{VVVVVVWWTWUWWWWXXX XXX2XXXYY YY)Y:ZZ[ [[[[\\#\\]]]*]+].]:]=]b]c]]]]]]^   @ H  @  D  H @      @>^^^.^/^w^^^^^___`X`Y`\````aaaaafagaaaaabb:ckcncxcd ddddddeeegiijj jkkkkm*m:mmnn!n)n8nVn^nfnunzn~ný     @ H4 @     Gnnno ooMoSoXoYpppqDqGqqqqqqqqrWsGsMsSsVsetvWvZvwxSxWxdxxy*y+yzz|||:}H}S}T@B @ H     HĨ   @   @   >(M[\s*+WXJKMNOvw3pWX#N  1 B x 1 2»䶶ު P  hP@P PPPP 8P P P P PPPC 2 }  ) ^ z   P p : #UVWXrs!8STUEef嫫垞P P   l@ P   l@ P  p@P P P P@ P  P    P   P P  P    P   2fg)`56<}~:;h   [\^ǾǸǬǦǟǘvNJo P P8 P8PP P PPhP P PPP hP  P   l@ P   l@ P    P   P +^KLMcd   89&Cb#$~;Qߺߨ٢zzoo P  hP@ PhhP!@@ P hP!@@ P hP!@@P  P  hP@PP  P @PPP  P 8P P +!!"7"d"y""# #`#######$$$w$x$$$$%%%S%%%%%󖉉~rrbTrrrr P  l@P 8  l@ P   l@ P l@ P   l@ P  l@P  P  hP@ P  hP@ P hP!@@ P  hP@ P  hP@ P hP!@@P Ph %&1&2&&''i'j''''(3(S(v(((())*)?)S)Y)))***-*.*b*c+H+I+J+p+q+󽽴󛏏th P h  P  l@ P   l@ P   l@P 8  l@P 8P8PP 8  l@P8  l@P8  l@ P   l@&++,G,H,I,J,u,v,,,-"-#-g-~----.-.D.V.W.X.../ /C////00(0N00000111:1`1x1ǽ⯯v P h!@@ P hP@@PhhP@@ P h!@@ P h !@@ P P  P   @P P  Ph  P h .111112242b2t222223333334344454748494~445.5/50515K5L5d5e55555666S6r6666675ϴϮϢϖσP  P   l@ P   @P PPP 88@@@@ @@!@@PoP P  P hP@@PhhP@@ P h!@@375767O778888888888999b9c9999:::0:G:[:\:::::::;B;C===>===>E>F>c>>>>¶¶鰪鞞阑P P P PP PP P  P   @ P   @PP PP P  P P 1>???S?T?V?W?X???@@W@@@@@@@A;AUAVAWAYAZA[AAAABBBBBB B(B)BhBiBBCC CLCnCCCCC񲨲P P  P hP  P h P  hP@PPP P 8P 8P P P\P P P@3CD,D^D~DDDDE#E$E%EyEzEEF#F$FFFcFFGH^HIUIVIWIIIIJ$JCJbJcJdujddP  P h Ph Ph Ph P h P  hP@ P hP P  P  @ P  h!@ P !@ P P@P #JdJJKKKKKKKKL LLM:MN N NINJNKNLNNNNNOO(O)O^OOOOPPPRPSPTP{P|PPPQ: Phh P hP P P   l@ P   l@ P  P!@P8P8PPPPP  P h-Q:QqQQQQQQR"R?RRRRRSSS`SSST=TSTTTqTTTTTUUUUUUUUUUUV V VXVYVyVVVVŶŰP hP hP P P P PP Phh!@ P h!@ Phh P h P h P h1VVWWWWSWTWVWWWXWWWXX2X3X4XKXhXXXXXY9Y:YYYYZZZZZZ[ [ [[[[\\#\$\\](])]*],]-].徾嬬坝PwP P 8PPP P P P hP hPBP hP  P  h!@P h4].]]]^:^^^^^^^__```W`X`Z`[`\`aaagaaabbbbb:b;c4c5c6cwcxcd d dTdddddddeeeeeefCfDfEffff⽷⢢P  !P PP P0hPUP P P  P 0h P0h>fffg gggh8h9hhhhhiiij^jjjkkkkkkkklnlolllm'm(m)m*m:m;mmmn;n<nxnynznnoppp¶PP hP P P  !P  P8 P8 P8PPP  P 8P 4ppppqqqqqqqqrVrWsSsTsUsVsfsgt^t_t`tttttvWvXvYvZvvvvw8w9w:wwxTxUxVxWxdxexxy(y)y*y,y-yyyyzz˺PPPwP P P hP P P PP P hP P P ;z|||||:|;|m|||}}}F}G}H}S}T}}~D~k J2 D:x?@ABiv ( ) E N Q˼аP P7!7(@PP P hP hP  P8B) 18ex-star18body18ex>    Th P@@  hP@Most importantly: i 3rd person contrasts with 1st & 2nd person. Anticipating conclusions... (3 (cf. (5 Alternation II: More EPP Effects [Assume they are nominal heads reement at the end of the verb.] (second try) strategy, which supplies the complementizer If inflections can be either arguments or agreement, theories of polysynthetic languages of both Jelinek (1984) and Baker (1991) turn out to be correct. *Unaccusatives can incorporate, uranching phrases (Muysken 1982; Chomsky 1994 Note: The desire to avoid competition between different arrays here is not merely driven by a dogmatic desire to get rid of D-structure conditions. The D-structure conditions we are trying to avoid are likely to be extremely complex since inflectional marking for an argument is affected by other arguments/inflections, an inflectional paradigm for Yimas would need to be at least 6-dimensional! It is questionable whether such an object could be learned. n, B, D. Gardiner & D. Frantz. 1984. Anderson, Stephen. 1976. On the Notion of Subject in Ergative Languages. In: C.N. Li (ed.), Subject and Topic. New York: Academic Press. Aoun, Joseph & Yen-hui Audrey Li. 1990. Minimal Disjointness. Linguistics 28, 189-204. Aoun, Joseph & Yen-hui Audrey Li. 1993. Syntax of Scope. Cambridge, Mass.: MIT Press.Baker, Mark. 1988.k. 1991. -asymmetries in Mohawk. Natural Language and Linguistic Theory 9, 537-76.Baker, Mark. 1993. Belvin, Robert. 1985. Nishga Syntax and the Ergativity Hypothesis: An Analysis Using the Government and Binding Model. In: Proceedings of the 20th International Conference on Salish and Neighbouring Languages. Bittner, Maria. 1992. earlier version of Bittner (1994). Bittner, Maria. 1994. On Case and Scope. Dordrecht: Kluwer. Bittner, Maria & Ken Hale. 1993. Ergativity: Towards a Theory of Heterogeneous Class. Ms., Rutgers U. & MIT..Case & Agreement II. MITWPL 19, 45-88.seley, Robert & Janig Stephens. 1989. of Subjects in Breton. Natural Language and Linguistic Theory 7, 407-428.ciana, & Patrizia Cordin. 1989. Campana, Mark. 1992. A Movement Theory of Ergativity. PhD thesis, McGill University. Chomsky, Noam. 1986. Knowledge of Language: Its Nature, Origin and Use. New York: Praeger.Chomsky, Noam. 1993.K. S.J. . Cambridge, Mass.asional Papers in Linguistics #5 Ferguson, K. Scott. 1993. A Note on Object Checking and the Shortest Step Metric. Harvard Working Papers in Linguistics #3. Ferguson, K. Scott & Erich Groat. 1994. Defining Shortest Move. Paper presented at GLOW Colloquium, Vienna.Foley, William. 1991.Hale, Ken & Jay Keyser. 1991. and the Lexical Expression of Syntactic Relations K. S.J. . Cambridge, Mass. Halle, Morris & Alec Marantz. 1993. Distributed Morphology and the Pieces of Inflection. K. S.J. . Cambridge, Mass.Hendrick, Randall. 1988.Jelinek, Eloise. 1984.and Configurationality. Natural Language and Linguistic Theory 2, 39-76. Kitagawa, Yoshihisa. 1986. Subjects in Japanese and English. PhD thesis, UMass Amherst. Amherst, Mass.: GLSA. Koster, Jan. 1978. Locality Principles in Syntax. Dordrecht: Foris. Lasnik, Howard & Robert Freidin. 1981. Core Grammar, Case Theory and Markedness. In: A. Belletti, L. Brandi & L. Rizzi (eds.) Theory of Markedness in Generative Grammar. Pisa: Scuola Normale. Laka, Itziar. 1993. Unergatives that Assign Ergative, Unaccusatives that Assign Accusative. In: J.D. Bobaljik & C. Phillips (eds.), Papers on Case & Agreement I. MITWPL 18, 149-172. Levin, Juliette & Diane Massam. 1985. Surface Ergativity: Case/Theta Relations Reexamined. In: S. Berman (ed.) Proceedings of NELS XV. Amherst, Mass.: GLSA. Massam, Diane. 1993. Localizing Case Systems. Paper presented at 6th Biennial Conference on Grammatical Relations, Vancouver, BC.Case & Agreement II. MITWPL 19, 213-254Murasugi, Kumiko. 1992. Murasugi, Kumiko. 1993.. Ms. McGill University.Ouhalla, Jamal. 1993.ation, Anti-Agreement. Natural Language and Linguistic Theory 11, 477-518.Phillips, Colin. 1993. J.D. Bobaljik & C. Phillips (eds.), Papers on Case & Agreement I. MITWPL 18, Phillips, Colin. to appear a.Phillips, Colin. to appear b.Seiter, William. 1980.Shlonsky, Ur. 1987. Cambridge, Mass.: MITWPL. Sityar, Emily. 1993. Voice and the Licensing of Subjects in Cebuano. Paper presented at CONSOLE 2, Tbongen, Germany.. MITWPL 19, 377-399. Urbanczyk, Suzanne. 1992. Ergativity and AgrO in Nishga. Ms., UMass Amherst. Vd WORD  ,Times .+]:Abs +<0Erg )=VP +V' +V (agent +theme (R~pro (#Dpro"0q## #"H## #"`## #"w## #"## #"T## # +Nom0" +Acc0"0/[=q#0_o# ( _Agr1P +Agr1' +Agr2P +Agr2'"7##3#####"#",##0#####"# ( (4)k c\ U\ Ed WORD  ,Times .+jVP +V' +V"0R## #"Hn## #"`## #"w## #"## #"5## #aG[[o0"Ve#a#.70",-# ( @Agr1P + Agr1' +Agr2P +Agr2'  ( (12)  (9Abs +<0Erg )=VP +V' +V  (oagent  (theme  (Q3pro ("pro"/&## #"GB## #"_^## #"vy## #"## #" ## #`+<}#a-0"(#aG3[G0"Z=# ( Agr1P + Agr1' +Agr2P +Agr2'`"Q *.t (t +t +t  (agent  +theme  (:AAgr1 +;0Agr2 (#Nom +=.Acc`d"Qm`+|# +aCrossing Paths (c Nested Paths  ( b. ( a. * Shlonsky (1987), eg. John, there eg. pro, wh-tracethe     ' ( ) D E G L M N P Q a f i q s v z |             +48Ajn=#+J[ *= @  @ H$\   @   @ J=AFG`$jkyzKdey5Tu#.19y)Gyz} !&=V|+?0F^_jpv #$7; @ `;<Ufk:LXjo-Jg}~----------.6666 6 6/6Q6^6o66666666677 7 7 7797_7h7iⴭ  HQH HI  HEH @ H? H: @   @ @ Bx{v$@_$.~---7 7i78&889O9:VVWWaMaNaOaPaQaRaSaTaUĵਝ P   l@P P  P l@ P   l@P  !P P P 8  l@PhPhPh/ ; [Spec,Agr1] is an A-bar position.s trigger deletion of agreement ta-, ant-, m- Complementizer a Complementizer aComplementizer aAssumptions ties in an array in which the complementizer lities in an array in which a complementizer (53) property be allowed to s by making less initial committo the ultimate form of lexical items Strategy: (54) f(55) ctically (Beard 1966, 1991; Anderson 1992; Aronoff 1976, 1994; Halle & Marantz 1993 Anderson, Stephen. 1992. A-Morphous Morphology. Cambridge, UK: Cambridge University Press. Aronoff, Mark. 1976. Word-formation in Generative Grammar. Cambridge, Mass.: MIT Press. Aronoff, Mark. 1992. Morphology by Itself. Cambridge, Mass.: MIT Press. Beard, Robert. 1966. The Suffixation of Adjectives in Contemporary Literary Serbo-Croatian. PhD thesis, University of Michigan. Beard, Robert. 1991. Lexeme-morpheme Base Morphology. Ms. Bucknell University. Muysken, Pieter. 1982. Parametrizing the Notion Head. Journal of Linguistic Research 2, 57-75. Pad WORDa  a,Times .+{:Abs": +<0Erg"j )=VP"j +V'" +V"3 (agent" +theme" (Rpro"R (#bpro"#r"0## #"H## #"`## #"w## #"## #"r## #`,~#`l`#ac.w0")m#aH\0"[# ( }Agr1P + Agr1'"" +Agr2P": +Agr2'"R ( (35) + Nested Paths~p`d WORD "z###"<###"Z###"y###"###":### ,Times .+ CP" +#C'") (FC"F +EAgr1P"G +'Agr1'"e (Agr1" +RAgr2P"B +*Agr2'"i ( Agr2"8 +UVP" (subject" (dobject"da"p60 z1z#"#!z# (no anti-agreement"  ( (40) LDDVad WORDVa Va"%7##5#"# ,Times .+/Neg"/ )#"/U"'47##5#"k# (1Neg"13 )#"1 (2cAbs"2v (0Nom"0'"4Q#####"8#####  (Osblocking"O (L no blocking"LK  (bNeg"v ( Neg" % ( (49) +b." ( a.^wBPwB@d WORDҡ  ,Times .+jVP +V' +V"0R## #"Hn## #"`## #"w## #"## #"5## #a#.70",-# ( @Agr1P + Agr1' +Agr2P +Agr2'  ( (51)  +=.Abs (iNErg )=VP +V' +V  (oagent  (Q3pro ("#pro"/&## #"GB## #"_^## #"vy## #"## #" ## #`+<}#aG3[G0"Z=# ( Agr1P + Agr1' +Agr2P +Agr2' *.t (t  +agent`+|"  ( b. ( a. CU2U2qqd WORDq  q,Times .(;@VP +V' +V"## #"1## #"H.## #"`G## # ( Agr2P +Agr2'  ( (52)  +V/Erg )=VP +V' +V  (Sxagent  (#<pro"K## #"1g## #"H## #"`## #`EV#a<-P0",F# ( WAgr2P +Agr2' *.t )t  +agent  ( b.`G$m" aalu0"kq# ( a. a- tpul-c -r2d 1sgDL b. Head movement usually driven by Lasniks stray affix filtertures or (b) [person, gender(56) apply lle & Marantz (1993)(57)(59) a.b. c. l The grammatical gender system is even more elaborate - 16 noun classes are distinguished in all, and there 7i7j7l7m7n7o7t7~77778%89O9]9d999::VVVVVVVVWW$W.W@WAW]WbWhW|WWWWWWWWWZaKaLaQaRaSaTaUaVaWaXaYaZa[a\a]a^a_a`aaabafag    HZ @ @  r  @ @   FKwB=wB-d WORDҡ  ,Times .+jVP +V' +V"0R## #"Hn## #"`## #"w## #"## #"5## # """"a ,>0"%, ( @Agr1P + Agr1' +Agr2P +Agr2'  ( (58)  +=.Abs (iNErg )=VP +V' +V  (oagent  (Q3pro ("pro "/&## #"GB## #"_^## #"vy## #"## #" ## # """"`+<}aB-bM0"Z< ( Agr1P + Agr1' +Agr2P +Agr2' *.t (t  +agent`*|  ( b.  (i0LF (y3LF  ( a. Some : array (aocked by array (b  agalamanapaqara|aaaaaaaaaaaaa @   aUaVaWaXaZa[a\a^a_a`aaabacadaeafagamanaoapaqaraaaaaaaaaaaaaaaP h P hPP  P h !@@P  P   l@%j j@  Y,#5?^F/QYbp"{j0#+`5  #   w@X | (   m^  |}(Y}!!"#"y"# #I#v###$$J$((():);)<***+++,, ,",N,.2.30223444`4a5/50516t9:=#=f=w=>?\?^@@@@AgAhBBB4F0K+KKN'NNPBPPQQQQQSMSVqVrW%WTWWWWWXXXXXXYYY[t]t]__```babbbbbcd'fdfeg_g`hohphi~jklPlQlllmfmmmmnOnPnQnWpp p!p"pqs~stuVvwwyyz%{{{{{|]9mn?|43}8 \W@6&w8<(o3gFQu}jP P P P P PP P P P P P P P PP P P P P P P P P P P P P P P P P P P P P P P PP P PP P P P P P P P P P PP P P P PP P P P P P P P P P P P P P P P P P P P PP P P P P P P P P P P P P P P P P P P P P P P P P PNP P P P P P P PP P P PPPPP P PP P P P P P P PP P P P P P P PDP P P P P P P P P P P P P PwP P P P PUP P P P P P P PP P P P PPPPPP P P P P P P P P P PP P PPP P P P P PP P P PP P PP P P P P P P P PwP P P P P P P P P PPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPP 7:= L\%-#3:!AWO1U^n=;7iagafghijklmnopqrs 2f^%+175>CJdQ:V].fpzaUatuvwxyz{|}~ r +|+}> HH(FE(HH(d'@/=8`8`R@H -:LaserWriter ChicagoNew YorkGenevaMonaco CairoPalatinoTimes HelveticaCourierSymbolMobile Zapf DingbatsrPalPhon PalPiRomanWPL-LogoIPATimes MT ExtraSibir1+|InsigniaLQmono+}InsigniaLQprop> KellsFLFEX?E^_8  !"#$8h l 8  ! ^_'.h=E != hhP=PH}~+L | ( *   "%#+-  -:{} "*+34?@'(Y d !!!!"3"4""# # #I#v###$$$(()9):);)<****+++++,, ,",#,N- -!.2.300223344444_4`4a555/5051669;;=e=v==>?\?]?^?t?x@@@@AgAhBBBB4FAFCFFFJFLFOFTFXK+K,KKNbNNNPBPPPnPPPPQQRSMS]SSVqVrVwW3W>W?WRW[W]WbWfWtWvWWWWWWXXXXXYYY[y[]]]]__````bdbnbbbbcdd'd,fdfeg_g`hohphhi:i;i<i~iiiijjlOlPlQlRlTllmmm/m=mvmmmmmmmmnOnPnQnno]o}pppp p"ppppprrs~sttv&v0vBvQvevyvwwwy!y'y/yCyyz%z){{{{{{{{{|||]|^9mnu@C/N_{&^m23FHU '7 5F[!"%/CVZ e*?@PU 45@CK%&<>^w+7 1<Sl'=Zn2_ur|/346Mf8EF^>Qco+-FYEFIg|,-.hijE}~G E  ' (aK ) D  E  G  L  M  N  P  UaL QE aE fE iEqEr q s v z |   & R  9 |aQ~ aR  ZaS[ ۀ]^M      !=!"D="H"=""="#_#`#E#$wVV$-%RaT)))aU*aV+I+=+aW,Gjn,HaX,JE,u-GaY.X0aZ34142434}a[4~a\51E5Ka]5L-69-;4-=-=>?T?U?X-?na^@a_@a`A[-BB EB(-F56F86F:6F?6 KaaK6 NHNLENPEP6/PQPR6^PVab6oQ6RS SafagUVVVVVVVVVValW6WUamWXanX3X4LXKXhLXXY6Z6]6]+ap^aq`6`Yaraa|bab6ca6cyaefag6g7h{h|WW$W.W@hj j 7 kjA7 kkkEkl*=l37 lF7l79lal7_7hla=#m(am*7i7j7l7mm7noap7oq7t7~777sPWAsW]taWbuWhvavZW|vaaw=xS xTaxWWxWxWy+|E|:W|;W||Ӏ |W}}E }FW}G}T}}}~}}+J[7w 8%}*}}}΀=A~F~ G`~C~V~W~r89O9]9d$jkyzK~d~؀~ۀey.j5Tu !T[iˀ#.19y_a)ȀGڀyPzT}[iPT[i !'&H=V|+?0F^_jpv  $&o#$7;$&<999UfkҀ;bx:LՀ)XjLNfo-Jg>?a%Maryland.handoutMITWPLMITWPL