Morphemes as fseq zones. Illustration of the theory on the example of Slavic L>T>N participles∗ Bartosz Wiland Advances in Minimalism and the Structure of Slavic Languages, Poznań 20th December 2016

1

Morphemes as fseq zones and the unaccusativity puzzle

The long-standing puzzle in the syntax of Polish and Czech: While all types of verbs form active L-participles, adjectival passive L-participles are only formed from unaccusativs; unergatives and transitives can only form N- or T-participles (e.g. Cetnarowska (2002)): (1)

unaccusative V – X L-passive a. blednąć – po-blad-Ł-y (Pol) get pale pale b. vlhnout – z-vlh-L-ý (Cz) get wet wet

(2)

unergative V – *L-passive a. ziewnąć – *ziew-Ł-y (Pol) yawn b. dupnout – *dup-L-ý (Cz) stamp

(3)

transitive V – *L-passive a. kopnąć – *kop-L-ý (Pol) kick b. koupit – *koupi-L-ý (Cz) buy

A structural solution: • The lexical entry for L includes case peels left by the movement of the unaccusative NP argument of the verb stem as part of what constitutes the participle zone in Slavic syntax. • A ‘morpheme’ is a lexicalized individual zone of functional sequence in syntax, or an ‘fseq zone’. Under this view, morphemes which compete for insertion with each other form the same fseq zone, while morphemes which co-occur together form different fseq zones in a syntactic representation.1 • Claim: the participle zone is projected on top verb stems in Slavic languages and spells out as L, T, or N, depending on its size and internal constituent structure The format of the Slavic verb: (4)

(prefix) – root – theme – participle – agr a. u – děl – a – l – a (active: L-participle) pref – do – AJ – L – f.sg ‘(she) did’ (Cz) b. u – děl – á – n – o (passive: N/T-participle) pref – do – AJ – N – n.sg ‘(it was) done’ (Cz)

7 themes in Polish and Czech: Ø, E, EJ, NU, AJ, OVA, and I, which together with the root they merge with encode the verbal argument structure: (5)

My nie chcemy o tym słysz-e-ć. (Pol) we not want about it hear-E-inf ‘We don’t want to hear about it.’

stative E

∗

This talk is based on joint work with Lucie Taraldsen Medová, (Taraldsen Medová and Wiland (In press)). In other words, if a particular sequence can be lexicalized by two morphemes, it means that there are two syntactic fseq zones. For example, if Tense is lexicalized as a fusional morpheme with phi-agreement features in a language, then temporal and phi features form a singleton fseq zone in that language. 1

1

(6)

Jan po-sadz-i-ł-Ø dziecko na stole. (Pol) Jan po-sit-I-L-3.m.sg child on table ‘Jan made the child sit on the table.’

(7)

a.

b.

causative I

Petr kop-nu-l-Ø psa. (Cz) Petr kick-NU-L-m.sg dog-acc ‘Petr kicked the dog (once).’ Petr hloup-nu-l-Ø. Petr stupid-NU-L-m.sg ‘Petr was getting more and more stupid’.

semelfactive NU

degree achievement NU

A consequence of the claim that morphemes are fseq zones: • L- (cf. (4-a)) and T-/N-suffixes (cf. (4-b)) form different types of participles but they form a single fseq zone since all three compete for lexical insertion • The resulting containment relation is (8)

active L-participle > adjectival L-passive > T-/N-passive

as in: (9)

active (non-present) L-participle a. Wczoraj kop-a-Ł-em piłkę z kolegami. (Pol) yesterday kick-AJ-L-1.msc.sg ball with friends ‘Yesterday I played some soccer with friends.’ b. Karel hod-i-L-Ø boty do kouta. (Cz) Karel throw-I-L-msc.sg shoes into corner ‘Karel threw the shoes into the corner.’

(10)

L-passive a. Król jest zmar-Ł-y i nic tego nie zmieni. (Pol) king-nom is die-L-msc.nom and nothing this not change ‘The king is dead and nothing will change that.’ b. Ta treska je z-mrz-L-á na kost. (Cz) this codfish-f.sg is pref-freeze-L-f.sg to bone ‘The codfish is frozen solid.’

(11)

T-passive a. Piłka zosta-ł-a kop-nię-T-a. (Pol) ball become-L-f.sg kick-NU-T-f.sg ‘The ball was kicked.’ b. Karel by-l kop-nu-T-Ø do břicha (Petrem). (Cz) Karel be-L kick-NU-T-msg.sg in stomach (Petr-ins) ‘Karel was kicked in the stomach (by Petr).’

(12)

N-passive a. Ta dziura zosta-ł-a wczoraj wy-kop-a-N-a. (Pol) this hole become-L-f.sg yesterday out-dig-AJ-N-f.sg ‘This hole was dug out yesterday.’ hoze-N-y do kouta (Petrem). (Cz) b. Boty by-l-y shoes be-L-msc.pl throw-N-msc.pl into corner Petr-ins ‘The shoes were thrown into the corner (by Petr).’

2

Passive participles

2.1

Verbal and adjectival passives

Two kinds of passive participles: verbal (eventive) and adjectival (stative). • In German: verbal passives occur with werden ‘get’/‘become’, adjectival passives with sein ‘be’ (cf. Kratzer (2000), Maienborn (2007)), as in: (13)

Die Reifen werden aufgepumpt. the tires get/become up-pumped ‘The tires are being inflated.’

eventive passive

2

(14)

Die Reifen sind aufgepumpt. the tires are up-pumped ‘The tires are inflated.’

stative passive

• Kratzer (2000) on adjectival passives: Target States (can be modified by a temporal adverbial immer noch or ‘still’ in English) and Resultant States (resist such a modification): (15)

Das Gebäude ist (immer noch) geräumt. the building is (still) evacuated ‘The building is (still) evacuated.’

(16)

Das Theorem ist (*immer noch) bewiesen. the theorem is (*still) proven ‘The theorem (*still) is proven.’

T(arget) S(tate)

R(esultant) S(tate)

• Morphological distinction in Swedish: TS but not RS passives can come out as underived adjectives (examples from Lundquist (2008: 148)):2 (17)

a.

b.

(18)

a.

b.

dörren är fortfarande öppen/??öppnad door-def is still open-adj/??open-de ‘The door is still open.’ dörr en redan öppnad/??öppen an already open-de/??open-adj door ‘an already opened/??open door’ Dörren är fortfarande stängd. door-def is still close-de ‘The door is still closed.’ en redan stängd dörr an already close-de door ‘an already closed door’

TS

RS

TS

RS

• by-phrase modification – verbal passives denote initiated events and can be modified by agentive by-phrases (not available for adjectival passives, as states do not have agentive implication):3 (19)

2.2

a. b.

The door was (recently) opened (by John). The door is (recently) opened (*by John).

eventive stative

The English -ed suffix

English verbal and adjectival passives are all formed with the -ed suffix (and its allomorphs) on the root.4 In other words, the -ed morpheme is an exponent for Past Tense, eventive, Resultant-State, and Target-State passives alike: (20)

a. b. c. d.

The The The The

gardener mow-ed the grass at 3pm yesterday. grass {gets, is being} mow-ed by the gardener. mailbox is (*still) empti-ed. building is (still) evacuat-ed.

Past Tense eventive RS TS

• Starke’s (2006) analysis of participles (cf. ‘the more you do the bigger you are’ logic): 2 In (18), there is no morphological distinction between the Target State and Resultant State: the example provides a purely contextual (with still- modification) example for the validity of the Target State vs. Resultant State distinction in Swedish on top of the morphological distinction given in (17). 3 Among other well-known modification tests that distinguish between eventive and stative passives is the compatibility with degree-modifying adverbials, which is impossible with eventive passives, as in (i), but possible with statives, as in (ii):

(i) (ii)

a. ?*The cart is completely pushed. b. ?*The bottle is half emptied. a. The equipment is completely damaged. b. The window is half opened.

4 Suffice it to say, the formation of different types of passives may require different categories of roots. We see this, for instance, in (20-c) and (20-d), where the Resultant State passive is based on an underived adjective empty and the Target State participle is based on an underived verb evacuate.

3

(21)

-ed ⇔

   PastTP          Eventive Passive   Fn+3 Fn+2

            

Resultant State

Fn+1

Target State Fn

...

The fact that (20-a)-(20-d) are all spelled out as the -ed morpheme follows from the Superset Principle, the major tenet of Nanosyntax, which regulates the insertion of the lexical-phonological material into syntactic nodes.5 (22)

a.

b.

A phonological exponent is inserted into a node in syntax if its lexical entry has a (sub-)constituent which is identical to that node (ignoring traces for insertion). (The Superset Principle, Starke (2009)) Where several items meet the conditions for insertion, the item containing fewer features unspecified in the node must be chosen (The Elsewhere Condition)

• The representations that match the lexical entry in (21) under the Superset Principle: (23)

a. b. c. d.

-ed -ed -ed -ed

⇔ ⇔ ⇔ ⇔

[ [ [ [

PastTP [ Eventive Passive [ RS [ TS ]]]] Eventive Passive [ RS [ TS ]]] RS [ TS ]] TS ]

English vs. Polish/Czech: Unlike in English, in Polish and Czech the participle zone is lexicalized by three different exponents: L, T, and N.

3

Participle fseq in Polish and Czech

3.1

Fseq zones in Slavic

Given the format of a Slavic verb in (4) and the bottom-to-top derivation, L, T, and N spell out the participle fseq zone which is projected on top of a zone which spells out as theme vowels. (24)

Fseq zones in the Slavic verb participle zone ⇔ {L, T, N}

F2 F1

theme zone ⇔ {Ø, E, EJ, NU, AJ, OVA, I}

F2 F1

root zone ⇔ {adj, noun, verb} F2

3.1.1

F1

Excursus: the structural complexity of the root zone

‘NU’ as two morphemes: N + U Taraldsen Medová and Wiland (2016): a (traditional) theme vowel NU (as in (7)) is made of two separate morphemes N + U, where N spells out the light verb GIVE or GET, depending on the amount of syntactic structure it lexicalizes: 5 The Superset Principle has been applied to the domain of case (Caha (2009)), directional adpositions (Pantcheva (2011)), Bantu class-markers (Taraldsen (2010b) and Taraldsen (in press)), Slavic prefixes (Wiland (2012)), Czech numerals (Caha (2013)), as well as to an extended theory of feature lexicalization in paradigms (Taraldsen (2012)), among others. See also Baunaz and Lander (in press) for the illustration of how the structural subset-superset relation works in the lexicalization patterns of strong, weak, and clitic pronouns in French in Cardinaletti and Starke’s (1999) system of the tripartition of Romance pronouns.

4

(25)

The light verb theory of N GIVE N3

GET N2

HAVE N1

• N spells out the light GET in degree achievements, which explains their GET-readings, e.g.: (26)

a.

Ziemniaki już mięk-n-ą. (Pol) potatoes-nom already soft-N-m.pl ‘The potatoes are getting soft.’

b.

Petr slep-n-u-l-Ø. (Cz) Petr blind-N-U-L-m.sg ‘Petr was getting blind.’

• N spells out the light GIVE in semelfactives, which defines their GIVE-readings: (27)

a.

Jan krzyk-n-ą-ł-Ø. (Pol) Jan shout-N-ThV-L-m.sg ‘Jan shouted.’ cf. Jan gave a shout.

b.

Petr kop-n-u-l-Ø psa. (Cz) Petr kick-N-U-L-m.sg dog-acc ‘Petr kicked the dog (once).’ cf. Petr give the dog a kick.

• Morpheme U, which in Czech surfaces either as /-u-/ or as /-ou-/ due to a structurally defined phonological lengthening, is an actual theme vowel and is part of the higher fseq zone. Containment theory of lexical categories Starke (2009), Lundquist (2008) – lexical categories are not primitive but structurally complex: (28)

Verb Cat3

Noun Cat2

Adjective ...

• Czech/Polish semelfactives = nominal root + the light N with the GIVE-reading, as in (29-a) • Czech/Polish degree achievements = adjectival root + the light N with the GET-reading, as in (29-b): (29)

a.

root zone of a semelfactive stem kop-N‘give a kick’

b.

root zone of a degree achievement stem slep-N- ‘get blind’

rootN -NGIV E rootA -NGET Noun ⇒ kop-

GIVE ⇒ -n-

Cat2

N3

Adjective ⇒ slepAdjective ...

GET N2

N1

3.2

...

HAVE

GET ⇒ -nN2

HAVE N1

t

t

T/N-passives

Polish and Czech T and N morphemes form eventive and stative passives (cf. (11)-(12)), Czech distinguishes between them morphosyntactically: • eventive passives in Czech have a short form (SF) and can be modified by an agentive by-phrase, which in Czech involves an agent marked with an instrumental case, as in (30)

5

• adjectival passives in Czech have a long form (LF) and resist the agentive modification, as in (31):6 (30)

Ten článek je přelože-n do italštiny (Karlem). this article-m.sg is translate-N-m.sg.sf into Italian Karel-inst ‘This article is translated into Italian by Karel’.

(31)

Ten článek je přelože-n-ej do italštiny (*Karlem). this article-m.sg is translate-N-m.sg.lf into Italian Karel-inst ‘This article is translated into Italian (*by Karel)’.

Medová and Taraldsen (2007) – Czech additionally distinguishes between RS and TS passives in the following way: • RS takes a locative PP instead of an expected directional PP • TS does not. Certain verbs of induced motion require a directional PP, as in (32), which is a property of periphrastic verbal passives, as in (33): (32)

Jan hodil boty [do kouta]DIR /*[v koutě].LOC Jan-nom threw-3.sg boots-acc into corner-gen / in corner-loc ‘Jan threw boots into the corner.’

(33)

Boty byly hoze-n-y [do kouta]DIR /*[v koutě].LOC . boots-acc were throw-N-sf.pl into corner-gen / in corner-loc ‘The boots were thrown into the corner.’

An adjectival passive, which stands out as the one which uses the long form of the participle in (34), however, takes the locative instead of the directional PP: (34)

3.3

Boty byly hoze-n-ý [v koutě]LOC /??[do kouta]DIR boots-nom.pl were-3.pl throw-N-lf.pl in corner-loc/ into corner-gen ‘The boots were thrown in the corner.’

L-syncretism and the unaccusativity puzzle

All verb stems produce active non-present L-participles, verb stems with different argument structure properties produce different types of T- and N-passives. For instance:7 (35)

Accusative semelfactive NU-stems forming T-passives: a. Maria ścis-nę-ł-a cytrynę. (Pol) Maria-nom squeeze-N+U-LT ense -f.sg lemon-acc ‘Maria squeezed a lemon.’ b. Cytryna zosta-ł-a ścis-nię-t-a. lemon-f.sg.nom become-LT ense -f.sg squeeze-NU-T-f.sg ‘The lemon got squeezed.’ c. Marie si skříp-(nu)-l-a prst do dveří. (Cz) Marie-f.sg SELF-dat squeeze-(N+U)-LT ense -f.sg finger-m.sg in door ‘Marie pinched her finger in the door .’ d. Prst byl skříp-nu-t-ý ve dveřích. finger-m.sg.nom be-LT ense -m.sg squeeze-N+U-T-m.sg in door ‘The finger got pinched in the door.’

6 Polish also exhibits sensitivity to agentive by-phrase modification with eventive and stative passives, as in (i) and (ii) respectively. Polish differs from Czech in that the agentive phrase is closer to its English equivalent in that it includes a preposition przez ‘by’ and an accusative agent.

(i)

(ii)

Piłka została kopnię-t-a (przez bramkarza). ball-nom became kick-T-f.sg (by goalkeeper-acc) ‘The ball was kicked by the goalkeeper.’ Maria jest przebra-n-a za klauna (*przez swoją mamę). Maria-nom is dress-N-f.sg for clown-gen (*by her mom-acc) ‘Mary is dressed up as a clown (*by her mom).’

Polish does not distinguish morphologically between eventive and stative participles as Czech often does. 7 There exists a degree of variation in the formation of T- and N-passives between Czech and Polish in that certain closely related stems which form T-passives in Czech, as for instance hřát (‘warm up’) – vy-hřá-t-ý (‘warmed up’), will form N-passives in Polish, as grzać (‘warm up’) – wy-grza-n-y (‘warmed up’). Also, there is at least one reflexive verb in Polish, namely bać się (‘to be afraid + SE-reflexive’), which does not produce any passive participle (*wy-ba-n-y, *wy-ba-t-y), while it does in Czech, as in bát se – ? vy-bá-t-ý. The nature of this variation is not explored in this talk. What is essential, however, is that there is no variation between Polish and Czech with respect to formation of active L-participles only by unaccusative verb stems.

6

(36)

Accusative causative I-stems forming N-passives: a. Mama kro-i-ł-a warzywa. (Pol) mom cut-I-LT ense -f.sg vegetables-acc ‘Mom was chopping the vegetables.’ b. Warzywa są już po-kro-jo-n-e. vegetables are already pref-cut-I-N-n.pl ‘The vegetables are already chopped.’ c. Trenérka od-stran-i-l-a překážky. (Cz) coach-f.sg pref-side-I-LT ense -f.sg hurdle-acc.pl ‘The coach put aside hurdles.’ d. Překážky už jsou od-stran-ě-n-é. hurdles already are pref-side-I-N-pl ‘The hurdles are already put aside.’

By contrast, unaccusative verb stems produce only adjectival L-passives:8 (37)

a.

b.

(38)

a.

b.

Król zmar-ł-Ø. king-nom die-LT ense -m.sg ‘The king died.’ zmar-ł-y król / Król jest zmar-ł-y. die-L-m.nom king-nom / king-nom is die-L-m.nom. ‘dead king / the king is dead.’ Pies wark-ną-ł-Ø. dog-nom growl-NU-LT ense -m.sg ‘The dog growled.’ *wark-ł-y pies / Pies jest *wark-ł-y. *growl-L-m.nom dog-nom / dog-nom is *growl-L-m.nom ‘*growled dog / *The dog is grawled.’

unaccusative V X L-passive

unergative V

*L-passive

The picture: the adjectival L-participle is syncretic with the active L-participle rather than T- or N-passive and is only produced by unaccusative verb stems Explanation: within the participle zone, the adjectival L-participle is syntactically smaller that the active Lparticiple (cf. (21)) and includes the accusative case peels in its lexical entry Assumption: Caha’s (2009) theory of case: • Case representation – layered case: (39)

InsP K6

(40) DatP

K5

LocP K4

Lexical entries for cases for pan ‘man, sir’ (msc.sg) a. /∅/ ⇔ [ K1 ] b. /a/ ⇔ [ K3 [ K2 [ K1 ]]] c. /u/ ⇔ [ K5 [ K4 [ K3 [ K2 [ K1 ]]]]] d. /em/ ⇔ [ K6 [ K5 [ K4 [ K3 [ K2 [ K1 ]]]]]]

GenP K3

AccP K2

NomP K1

NP ...

• Case derivation – example of movement triggered by the shape of the lexical entries in the paradigm of the Polish pan ‘man’ in (40), which has syncretic exponents for acc–nom and dat–loc (cf. the Superset Principle): 8 The fact that only unaccusatives, but not unergatives, can form adjectival L-participle has been used as a diagnostic to distinguish between these two verb classes in Polish and Czech (e.g. Cetnarowska (2000), Medová (2012)).

7

(41) InstP ⇒ em

NP pan-

K6 DatP ⇒ u

...

K5 ...

LocP K4 GenP ⇒ a

...

K3 ...

AccP K2 ...

NomP ⇒∅ K1

4

t

Case peels inside a passive participle

Consider the step-by-step derivation of L-passive participle based on the unaccusative verb stem, as in zmarły król ‘dead king’ of (37-b): • the verb stem (the verbal root zone and a prefix z- and the Ø-theme) selects an accusative NP: (42)

KaseP

VP(verb

AccP K2

NomP K1

NP

stem)

root zone

theme vowel

z-mar-

Ø

król • the layers of the higher participle fseq zone are merged – recall that the lower layers of the participle fseq build the T/N-passives: F2 Ppass ⇔T/N

(43) F2

F1 Ppass F1

KaseP AccP K2

VP(verb

NomP K1

stem)

z-mar-

NP król

• at this point, neither the participle fseq nor the case fseq can be lexicalized, as none of them form a constituent; in order to facilitate the spell out, the KaseP constituent raises to the top of the tree as in (44)(cf. Caha (2011) and Baunaz and Lander (in press) on spell out driven movement):

8

(44)

F3 P

F2 Ppass ⇔T/N

KaseP AccP K2

F2

VP

F1

z-mar-

NomP K1

F1 Ppass t

NP król

• Given the hierarchy in (21), the Polish/Czech exponent L spells out bigger structures than T and N since it spells out the biggest active L-participle • Since L also spells out smaller L-passives in unaccusatives, its lexical entry must include a lower layer of structure but exclude the verb stem and the nominative NP (which have separate lexical entries). • Thus, L-passives have the following shape of lexical entry: (45)

Lpass /ł/ ⇔ [[ K2 ][ F2 [ F1 ]]]

• spell out driven movements: 1. evacuation of the verb stem (VP) in (46) and 2. evacuation of NomP, which peels the AccP-layer in (47): (46)

F4 P

F3 P

VP z-mar-

F2 Ppass ⇔T/N

KaseP AccP K2

F2

t

F1 Ppass F1

NomP K1

t

NP król

(47) NomP K1

F4 P

NP król

F3 Ppass ⇔L

VP z-mar-

F2 Ppass ⇔T/N

KaseP AccP K2

tV P

F2

t

F1 Ppass F1

t

The extraction of NomP derives a constituent that matches a lexical entry in (45), which overrides the insertion of T and N (cf. the Elsewhere clause of (22)).

9

4.1

Adjectival Agr is outside the participle zone

While (bigger) active non-present L-participles merge with verbal agreement morphemes (cf. (10)-(12)), (smaller) L-passives merge with agreement morphemes which shows the adjectival declension pattern (cf. (9)). • Agr merges directly with a constituent whose head (msc.nom in the example below) it agrees with: (48) AgrP y

F4 P

NomP K1

NP król

F3 Ppass ⇔L

VP z-mar-

F2 Ppass ⇔T/N

KaseP AccP K2

tV P

F2

t

F1 Ppass F1

t

• spell out driven movement of F4 P to the top of the tree makes the Agr surface as a suffix on the participle zone:9 (49)

F4 P AgrP NomP K1

z-marF2 Ppass ⇔T/N

KaseP AccP K2

tV P

F2

t

t

y

F3 Ppass ⇔L

VP

NP król

F1 Ppass F1

t

• the evacuation movement of the constituent comprising the verb stem (VP) and the subset of the participle zone F3 P>. . . >F1 P (without yet a higher layer of structure building active non-present L-participles, as predicted by the hierarchy in (21)) explains why NPs are placed after participles (as long as there is no copular be present in the sequence, as already indicated in (37-b)):10 (50)

5

a.

zmar-Ł-y król dead-L-nom.3sg.msc king-nom ‘(the) dead king’ b. *król zmar-Ł-y king-nom dead-L-nom.3sg.msc

Why unergatives do not build L-passives

Taraldsen Medová and Wiland (2016): both unaccusative verbs and unergative verbs select accusative objects, that is NPs with projected AccP>NomP sequences on top, which subsequently raise by case peeling movement.11 The difference between Slavic unaccusatives and unergatives: 9 Note that the derivation in (49) resembles the proposal in Leu (2015) for deriving adjectival agreement in German. More generally, Leu’s work relies on subextractions from complex specifiers in the derivation of the German was für ‘what for’ split, which is similar to the derivational steps in (46)-(47) in that they also involve subextractions from complex constituents. 10 Note that Polish allows for both pre- and post-nominal placement of adjectives, hence the L-participle > NPnom order does not simply follow from a general constraint on adjective placement. 11 Such an analysis of Czech and Polish unergatives is in line with Taraldsen (2010a), who shows that Norwegian unergative participles have agentive get-passive readings.

10

• unaccusatives are syntactically smaller than unergatives and merge the accusative NP on top of the theme vowel zone • the accusative NP is merged as a layer of structure inside the theme vowel zone which builds unergatives • the stranded AccP-peel is spelled out as part of the unergative semelfactive theme vowel NU or activity AJ • the AccP-peel is part of a lexical entry of unergative stems, such stems do not produce L-passives as they include the head of the AccP-peel as part of their own lexical entry, as in (45). Consider the derivation of an unergative stem syknout ‘hiss’ as in the Czech sentence below. (51)

Karel syk-nu-l-Ø. (Cz) Karel-nom hiss-NU-L-msc.sg ‘Karel hissed.’

• Taraldsen Medová and Wiland (2016): nominal roots of semelfactives select the GIVE superset of the light N morpheme, the adjectival roots of degree achievements select its GET subset: (52)

GIVE N3

99K + nominal root (in semelfactives: kop-N-ą-ć)

GET N2

99K + adjectival root (in DAs: bled-N-ą-ć)

HAVE N1

• popular assumption about the VP: unergative > transitive/accusative > unaccusative • in terms of morphemes as fseq zones: verb stems which spell out as unergatives form a bigger fseq zone than verb stems which spell out as unaccusatives • consequently: a bigger unergative stem includes the structure of an unaccusative stem, which comprises the AccP argument as in (42) • in (53), the root zone merges with an unergative theme vowel U, which is bigger than (any) theme vowel which builds unaccusatives: (53)

F3 P

rootN - NGIV E

Noun ⇒ sykCat2

Adjective ...

GIVE ⇒ -nN3

F3

GET N2

AccP

HAVE N1

K2

t

F2 P

NomP K1

NP

F2

F1 P F1

t

Karel

• spell out driven movement of the NP facilitates the lexicalization of the theme vowel zone – the NP raising takes place exactly as in the derivation of the unaccusatives in (47), in that the movement of the NP strands the AccP peel and surfaces as a nominative-marked subject, as in (54) • the difference: is that in unergative stems, the AccP peel becomes spelled out as a theme vowel (the U-theme), in the case at hand:

11

(54)

NomP K1

NP . . .

F3 P ⇒ -u-

rootN - NGIV E

Noun ⇒ syk-

F3

GIVE ⇒ -n-

AccP Cat2

Adjective ...

N3

GET N2

K2

t

F2 P F2

F1 P

HAVE F1 N1

troot−N give

tNoun

A final note on the theme U present in the light N+U sequence: • it builds not only unergatives as in (51)/(54) but also unaccusatives • in accusatives, as argued in Taraldsen Medová and Wiland (2016), U is structurally smaller in that it spells out only a subset of projections of the unergative U to the exclusion of the AccP layer This makes a correct prediction about the fact that not only unaccusative Ø-stems as in (37-b) but essentially also degree achievement unaccusative NU-stems build adjectival L-passives: (55)

a.

b.

6

Moje koty z wiekiem będą głuch-N-ą-ć. (Pol) my cats-nom with age be.fut deaf-N-ThV-inf ‘My cat will be getting deaf with age.’ Moje koty będą ogłuch-Ł-e. my cats-nom be.fut deaf-Lpass -agr. ‘My cats will be deaf.’

Summary

Features of morphemes as fseq zones: • each zone of the fseq occupies its own position with respect to other fseq zones • the boundary of one zone marks the beginning of another, higher zone The participle zone in Polish and Czech – two contrasting examples: 1. the lexical entry of the L-passive includes accusative case peel left by the movement of the nominative NP argument of the verb stem 2. the accusative case peels become spelled out as part of the verb stem, a constituent lower than the participle zone This presence (in the case of 1.) and absence (in the case of 2.) of accusative peels in the lexical entries explains the puzzle why only unaccusative verb stems can build L-passives.

References Baunaz, Lena, and Eric Lander. in press. Nanosyntax: the basics. In Exploring Nanosyntax, ed. Lena Baunaz, Karen De Clercq, Liliane Haegeman, and Eric Lander. Oxford: Oxford University Press. Caha, Pavel. 2009. The nanosyntax of case. Doctoral Dissertation, CASTL, University of Tromsø. Caha, Pavel. 2011. The parameters of case marking and spell out driven movement. Linguistic Variation Yearbook 2010 32–77. Caha, Pavel. 2013. Czech numerals and no bundling. Ms. CASTL/University of Tromsø. Cardinaletti, Anna, and Michal Starke. 1999. The typology of structural deficiency. A case study of the three classes of pronouns. In Clitics in the languages of Europe, ed. Henk van Riemsdijk, 145–233. Berlin: Mouton de Gruyter. 12

Cetnarowska, Bożena. 2000. The unaccusative/unergative split and the derivation of resultative adjectives in Polish. In Formal Approaches to Slavic Linguistics 8 , ed. Tracy H. King and Irina Sekerina, 78–96. Ann Arbor: Michigan Slavic Publications. Cetnarowska, Bożena. 2002. Unaccusativity mismatches and unaccusativity diagnostics from derivational morphology. In Many morphologies, ed. Paul Boucher, 48–81. Somerville, MA: Cascadilla Press. Kratzer, Angelika. 2000. Building statives. University of Massachusetts, Amherst. Leu, Thomas. 2015. The architecture of determiners. Oxford: Oxford University Press. Lundquist, Björn. 2008. Nominalizations and participles in Swedish. Doctoral Dissertation, University of Tromsø. Maienborn, Claudia. 2007. Das Zustandspassiv: grammatische Einordnung - Bildungsbeschränkung - Interpretationsspielraum. Zeitschrift für germanistische Linguistik 35:83–144. Medová, Lucie. 2012. Anticausatives are derived unergatives. In Slavic Languages in Formal Grammar, ed. Marketa Ziková and Mojmir Dočekal, 291–306. Frankfurt am Main: Peter Lang. Medová, Lucie, and Tarald Taraldsen. 2007. 1, 2, se. In Czech in generative grammar, ed. Mojmir Dočekal, Marketa Ziková, and Jana Zmrzlíková, 119–138. LINCOM: Munich. Pantcheva, Marina. 2011. Decomposing path: the nanosyntax of directional expressions. Doctoral Dissertation, CASTL/University of Tromsø. Starke, Michal. 2006. The nanosyntax of participles. Lectures at the 13th EGG summer school, Olomouc. Starke, Michal. 2009. Nanosyntax: a short primer to a new approach to language. Nordlyd 36:1–6. Taraldsen, Knut Tarald. 2010a. Unintentionally out of control. In Argument structure and syntactic relations: A cross-linguistic perspective, ed. Maia Duguine, Susana Huidobro, and Nerea Madariaga, 283–302. Amsterdam: John Benjamins. Taraldsen, Tarald. 2010b. The nanosyntax of Nguni noun class prefixes and concords. Lingua 120:1522–1548. Taraldsen, Tarald. 2012. Modeling the Neighborhood Hypothesis for syncretisms. Handout for a talk given at the 43rd annual meeting of the North East Linguistic Society (NELS), CUNY, October 2012. Taraldsen, Tarald. in press. Spanning vs. constituent lexicalization: The case of portmanteau prefixes. In Exploring Nanosyntax, ed. Lena Baunaz, Karen De Clercq, Liliane Haegeman, and Eric Lander. Oxford: Oxford University Press. Taraldsen Medová, Lucie, and Bartosz Wiland. 2016. Semelfactives are bigger than degree achievements. Ms. Taraldsen Medová, Lucie, and Bartosz Wiland. In press. fseq zones and Slavic L>T>N participles. In Exploring Nanosyntax, ed. Lena Baunaz, Karen De Clercq, Liliane Haegeman, and Eric Lander. Oxford: Oxford University Press. Wiland, Bartosz. 2012. Prefix stacking, syncretism, and the syntactic hierarchy. In Slavic Languages in Formal Grammar, ed. Mojmir Dočekal and Marketa Ziková, 307–324. Berlin/Frankfurt: Peter Lang.

13