Little v workshop Leiden, October 25, 2013 Little v, Voice and phases in Participles Elena Anagnostopoulou University of Crete [email protected], [email protected] 1. Goals One question addressed in this workshop: Are v and Voice the same or different heads? [Recall that in e.g. Chomsky 1995, Marantz 1997, they were taken to be the same head]. Answer: v and Voice are different heads Empirical domain: A decomposition analysis of different types of adjectival participles in terms of the layers in (1):1 (1) [VoiceP [vP [Stative P ]]] The two verbal layers in (1) are required to account for differences between (i) participles without event implications (“lexical”: no v) vs. participles with (“phrasal”: they contain v). (ii) resultant state phrasal participles (they may contain Voice) vs. target state phrasal participles (they can’t contain Voice). 1 Cf proposals according to which, verbs and verb-based derivations are decomposed into a core lexical meaning (labeled root, in Distributed Morphology DM) and functional layers responsible for: (i) the introduction of an event variable (vP, following Marantz 2001) and (ii) the introduction of the external argument (VoiceP, following Kratzer 1996). See Alexiadou, Anagnostopoulou & Schäfer (2006, to appear), Marantz (2001) Ramchand (2010) and many others. 2. Decomposing adjectival participles Kratzer (1994, 1996, 2000): adjectival participles do not form a uniform category: 1) Phrasal vs. lexical participles (Embick 2004 calls them resultative vs. stative) 2) Resultant state vs. target state participles Note 1: Resultant state participles are always phrasal. Target state participles are either phrasal or lexical. I will only be concerned here with phrasal target state participles. Note 2: I will be using the terms “phrasal” vs. “lexical” in order to avoid terminological confusion between Embick’s (2004) resultative and Kratzer’s (2000) resultant state participles. Embick’s resultative participles subsume resultant state and phrasal target state participles. 2.1. Phrasal participles vs. lexical participles and v Adjectival participles (i) either denote states resulting from prior events (phrasal), (ii) or they lack event implications (lexical). See Kratzer (1994), (2001) for German, Anagnostopoulou (2003) for Greek, Embick (2004) for English, Meltzer- Asscher (2011) for Hebrew, and the literature building on them. In English, participles without vs. with event implications sometimes show differences in form: Regular, -ed participle for participles with event implications, irregular for participles without: Table 1: different allomorphy patterns in English (Embick’s 2004, (11)) Root Stative Resultative Passive BLESS bless-èd bless-ed bless-ed AGE ag- èd ag-ed ag-ed ROT rott-en rott-ed rott-ed SINK sunk-en sunk-0 sunk-0 SHAVE (clean)-shaven shav-ed shav-ed OPEN open-0 open-ed open-ed EMPTY empty-0 empti-ed empti-ed DRY dry-0 dri-ed dri-ed 2 For the root OPEN, “opened” has event implications and “open” not. With CLOSE, the same form “closed” is ambiguous. In Greek, participles always have a different form depending on whether they have event implications or not (Anagnostopoulou 2003). A different suffix: -menos in participles with event implications, -tos in participles without: (2) anig-men-os open-men-masc.sg.nom anix-t-os “opened”/ “open” open-t-masc.sg.nom (3) klis-men-os close-men-masc.sg.nom klis-t-os “closed”/ “closed” close-t-masc.sg.nom SEVERAL DIAGNOSTICS have been argued to differentiate participles with event implications from participles without: (a) Participles with event implications (vs. participles without) allow modification by manner adverbials (diagnostic proposed by Kratzer 1994): (4) The package remained carefully opened/ *open (5) To grama The letter miazi prosektika seems carefully English anig-meno/*anix-to Greek open-meno/ *open-to (b) Participles without event implications (vs. participles with) are disallowed after verbs of creation (diagnostic due to Embick 2004): (6) The door was built open/ *opened (7) To jiaurti ftiaxtike pix-to / *pig-meno The yoghourt was made thick-ti/ thick-meni ‘The yoghourt was made thick/*thickened’ English Greek (c) Participles with event implications show a repetitive-restitutive ambiguity with wieder ‘again’, unlike participles without which are only restitutive (due to von Stechow 1996): (8) a. I porta ine {ksana} anig-meni {ksana} Ambiguous: Repetitive (there was a previous The door is {again} open-ed {again} opening)/ Restitutive (the door was open before) b. I porta ine {ksana} anix-ti {ksana} The door is {again} open {again} Unambiguous : Only restitutive (no previous opening presupposed) 3 The above facts can be explained if the Adj(ectivizer)/ Stativizer embeds vP in participles with event implications and RootP in participles without (Embick 2004, Alexiadou & Anagnostopoulou 2008): Participles with event implications: attachment of Adj above v (phrasal) (9) Adj/StatP 3 Adj/Stat vP 3 v RootP 3 Root Participles without event implications: attachment of Adj to the Root (lexical) (10) Adj/StatP 3 Adj/Stat RootP 3 Root Explaining the diagnostics: (I) INTERPRETATIONS OF “AGAIN”. Following the structural theory of repetitive vs. restitutive ambiguity (von Stechow 1996), again attaches to vP (expressing the event) in the repetitive reading and to RootP (expressing the result) in the restitutive reading: (11) Repetitive – restitutive ambiguity in phrasal vP = repetitive: there was a previous opening 3 again vP 3 v RootP = restitutive: the door was open again 3 again RootP (12) Lexical: restitutive only 4 RootP = restitutive: the door was open again 3 again RootP (II) MANNER MODIFICATION is only licit when the v head is present: (13) Phrasal vP = 3 carefully vP 3 v RootP 3 RootP (14) carefully opened Lexical *RootP = *carefully open 3 carefully RootP 2.2. Phrasal participles may contain Voice Evidence: by-phrases, instruments, manner and agent-oriented adverbs. Two types of languages (Anagnostopoulou 2003, 2013, in progress; Paslawska & von Stechow 2003, Alexiadou, Anagnostopoulou & Schäfer, to appear). I) Greek, Russian: Voice-related modifiers may appear freely Agent PPs (15) Ta keftedakia ine tiganis-mena apo The meatballs are fried by 'The meatballs are fried by Mary' tin the Instrument PPs (16) Ta malia tis basilisas ine xtenismena The hair the queen-GEN are combed me xrisi xtena with golden comb ‘The hair of the queen is combed with a golden comb’ Agent-oriented, manner adverbs 5 Maria Mary (17) To thisavrofilakio itan prosektika anigmeno The safe was cautiously opened /skopima paraviasmeno /deliberately violated ‘The safe was cautiously opened/ deliberately opened’ Purpose clauses (18) Aftos o pinakas ine zografismenos apo mia This the painting is painted by a omadha aktiviston, gia na sokarun group activists-GEN for to shock-pl ‘This painting is painted by a group of activists in order to shock people’ II) English, German, Hebrew: Voice related modifiers are subject to restrictions Agents Two conditions on agents in English (McIntyre 2013): Condition 1 The referent must be “responsible for continuing the state expressed by the participle”: (19) a. b. The road remained blocked by the police/ supported by pylons Edeltraut seemed flattered by the report/ ??the journalist Condition 2 For fewer speakers, by-phrases are also possible with resultative participles if the external argument’s “sortal specifications are crucial for the nature of the state”: (20) a. b. %The text seems written by a genius/ foreigner/ ghostwriter *The text seems written by John. He wouldn’t have gone home without finishing it. McIntyre (2013, fn. 6) observes that there is a definiteness contrast displayed by by-phrases belonging to the “sortal” class. (21) receives worse judgments than (20a), even though Picasso is “detectable” from the state.2 (21) The door seems painted by Picasso This goes back to an observation made by Grimshaw (1990), namely that by-phrases in adjectival participles tend to be indefinite, generic, plural. 2 The judgments on (21) vary among speakers; in Bruening (to appear) comparable examples are reported to be attested on the internet. In the literature on German, Picasso examples are reported to be well-formed. 6 Similar conditions have long been identified in the literature for German. Rapp (1996: 257) proposes that by-phrases are licensed only when they continue to characterize the resultant state denoted by the adjective (examples in (22) from Litvinov & Nadjalkov 1988, chapter 5 via Maienborn 2007). (22) a. b. Die Zeichnung ist von einem Kind angefertigt The painting is by a child made ‘The painting is made by a child’ Die Unterlagen sind vom Chef korrigiert The documents are by the director corrected ‘The documents are corrected by the director’ Vs. (23) a. b. *Die Tür war von ihm geöffnet (Vaagland 1983: 194) The door was by him opened *Die Mutter war von ihrer Tochter besucht (Litinov & Nedjalkov 1988:149) The mother was by her daughter visited Instruments Instrument PPs in English and German are subject to similar conditions as by-phrases (reported examples are mostly subject to McIntyre’s Condition 2): English (Bruening, to appear, internet examples): (24) a….the round edge looks beaten with a hammer or sledge b….the stone looks cut with some kind of machine c….interior still looks carved with a dull spatula…. German (Litvinov & Nadjalkov 1988, chapter 5 via Maienborn 2007, Rapp 1996): (25) a. b. Der Brief war mit roter Tinte geschrieben The letter was with red ink written ‘The letter was written with red ink’ *Ihre Haare sind mit einem goldenen Kamm gekämmt Her hair is with a golden comb combed Result-oriented manner expressions English (Bruening, to appear, internet example, cf. Embick’s example (4) “The package remained carefully opened”): (26) …..for he seems dressed with more studied elegance than any body here. German 7 (27) a. b. Ihre Haare sind schlampig gekämmt Her hair are sloppily combed ‘Her hair is sloppily combed’ *Der Safe war vorsichtig/ vorsätzlich geöffnet The safe was cautiously/on purpose opened ‘The safe was cautiously opened/ opened on purpose’ Cf. Meltzer-Asscher (2011) on comparable restrictions in Hebrew. Conclusion: Greek, English and German phrasal participles may contain Voice. Greek nonnegated participles freely allow Voice (all kinds of modifiers are permitted). It must be explained (a) why English and German Voice-related modifiers are subject to restrictions 1 and 2 and (b) what differentiates English/German from Greek. [See Anagnostopoulou & Samioti (2012), Anagnostopoulou (2013, in progress) and Alexiadou, Anagnostopoulou & Schäfer (to appear) on negated participles in Greek, English and German. They behave identically in all three languages; they may contain Voice subject to restrictions. They can’t be discussed at all here due to time limitations.] 2.3. Voice in English and German vs. Greek adjectival participles Gehrke (2011, to appear) argues that the restrictions in German adjectival participles described above follow from the hypothesis that the underlying event cannot be an actual event but rather an event kind. Evidence that the underlying event is not an actual event in German: a) No temporal modification of the underlying event (von Stechow 1998, in contrast to the present perfect): Adjectival passive (28) a. *Der Computer The computer b. Der Computer The computer vs. ist is ist is vor 3 Tagen/am Mittwoch repariert before 3 days/ on Wednesday repaired seit 3 Tagen repariert since 3 days repaired Present Perfect (29) Ede hat den Computer vor 3 Tagen/am Mittwoch repariert Ede has the computer before 3 days/ am Mittwoch repaired “Ede repaired the computer three days ago/ on Wednesday” b) A modifier like recently can only modify the resultant state and not the underlying event: 8 (30) Die Tür war kürzlich geöffnet The door was recently opened “The door was in the opened state recently, but probably is no longer” NOT: The door is in the opened state, the opening took place recently c) No spatial modification of the underlying event: (31) ???Das Kind war The child was im Badezimmer in the bathroom gekämmt combed The unavailability of spatial and temporal modifiers provides evidence that the causal event in German participles has no spatial and temporal manifestation. Evidence that the underlying event is an actual event in Greek: a) Temporal modification: (32) a. b. To computer ?ine/ itan episkevasmeno The computer is/was repaired ‘The computer is/was repaired three days ago’ To spiti ine xtismeno to 1963 The house is built in 1963 prin apo tris meres before from three days b) Interpretation of “recently”: (33) I porta ine prosfata anig-meni The door is recently opened ‘The door is recently opened’ “The door is in the opened state, the opening took place recently” NOT: The door was in the opened state recently, but probably is no longer c) Spatial modification: (34) a. b. To forema The dress To vivlio The book ine sideromeno is ironed ine grameno is written sto katharistirio in the dry-cleaners stin Ellada ke stin Germania in Greece and in Germany Modifiers creating “new sub-kinds” in German Landman & Morzycki (2003) point out that temporal and locative adverbials cannot antecede adverbial ‘so’, unless they are creating a new or sub-kind: (35) a. *Maria hat am Dienstag getanzt, Mary has on Tuesday danced, 9 und Jan hat and Jan has auch also so getanzt so danced b. c. *M aria hat in Minnesota gegessen, und Jan hat auch so gegessen Mary has in Minnesota eaten, and John has also so eaten Maria schläft in einem Schlafsack, und Jan schläft auch so Mary sleeps in a sleeping bag and John sleeps also so ‘Mary sleeps in a sleeping bag and John does so, too’ In (35c) “in a sleeping-bag” does not specify the location of an actual sleeping event but rather it creates a new sub-kind of a sleeping event, this being the reason why it licenses “so” anaphora. Gehrke argues that the kinds of modifiers licensed in adjectival participles in German are precisely those that license so anaphora. In examples like “The letter is written with a pencil”, “The hair is sloppily combed”, “The drawing is made by a child”, the modifiers create new sub-kinds of the underlying events and are therefore licensed. By contrast, there are no established sub-kinds of e.g. “slow writings” (In *the letter is written slowly), or “Mary cookings” (In *the food is cooked by Mary), and the modifiers are ruled out. The latter are precisely the modifiers that are ruled in in Greek. Greek vs. German/English participles Greek participles with Voice: (36) Adj 3 Adj AspP 3 Asp VoiceP 3 Voice vP 3 v RootP The aspectual head in (36) has the semantics of Kratzer’s (2000) Perfect operator: (37) Stem + object: e [prove (the theorem) (e) ] Stativizer: Pte [P(e) & (e) t] Output after applying the stativizer to stem+object: te [prove (the theorem) (e) & (e) t] The Perfect operator leads to the interpretation that the run time of the verbal event took place before the time denoted by the stativized VoiceP (Kratzer 2000). 10 German/ English participles with Voice Gehrke (to appear), Alexiadou, Gehrke & Schäfer (2013): Verbal predicates are predicates of event kinds which get instantiated only when verbal structure is directly embedded under Tense/Aspect. If not they remain in the kind domain. Proposed Parameter: German and English resultant state participles lack the Asp layer hosting the Perf operator. VoiceP is directly embedded under Adj: (38) AdjP 3 Adj VoiceP 3 Voice vP 3 v RootP 3 As a result, resultant state participles are subject to the kind restriction. 2.4. Not all phrasal participles contain Voice: evidence from verb classes Uncontroversial cases of phrasal participles lacking Voice are based on internally caused (Levin & Rappaport Hovav 1995) non-alternating unaccusative verbs. In Greek: (39) a. b. c. d. To luludi The flower I Maria The Mary To fito The plant I limni The lake ine is ine is miazi looks ine is omorfa nicely apotoma abruptly prosfata recently epikindina dangerously anthis-meno blossomed geras-meni aged mara-meno wilted pago-meni frozen In (39) the ending is –menos signifying a phrasal participle. Manner adverbs are licit pointing to the presence of v. Crucially, these verbs never combine with Voice (see Alexiadou, Anagnostopoulou & Schäfer 2006, to appear, for discussion). Similarly in English (Bruening, to appear, mentions examples like “a gradually/dangerously frozen lake”, “suddenly fallen leaves”, “slowly melted snow”). For these, structures like (9) without Voice are needed: 11 (9) Adj/StatP 3 Adj/Stat vP 3 v RootP 3 Root 2.5. Resultant state participles vs. target state participles: presence vs. absence of Voice. 2.5.1. Target state participles do not tolerate Voice Kratzer (2000) based on Parsons (1990: 234-235). Two types of participles. (i) Target state participles describe states that are in principle reversible. Immer noch 'still' is compatible only with target state participles. They are related to telic verbs whose target state can modified by for-adverbials. Resultant state participles introduce states that hold forever after the event that brings them about. They do not tolerate immer noch. They are related to telic verbs that disallow for-adverbials. (ii) Anagnostopoulou’s (2003) observation: consider fouskono ‘pump up/ inflate’ in Greek, which can license a target state participle, and has a target state that can be modified by a durative adverb when used transitively: (40) a. b. Ta lasticha ine akoma fusko-mena The tires are still pumped up Tha fuskosume tin varka jia liges FUT pump-up the boat for few ‘We will inflate the boat for a few hours’ Implies: the boat will remain inflated for a few hours ores hours As soon as an agent or instrument PP or an agentive adverb is present, the target state participle is coerced into a resultant state participle (see Kratzer 2000: 11 that target state verbs normally also license resultant state participles “as long as a’job done’ or ‘that’s over’ reading is plausible.” ). By-phrases and instrument phrases are incompatible with akoma ‘still’: (41) a. b. Ta lastixa ine (*akoma) fuskomena apo The tires are (still) inflated ‘The tires are still inflated by Mary’ Ta lastixa ine (*akoma) fuskomena me The tires are (still) inflated with ‘The tires are still inflated with the pump’ 12 tin by Maria the Mary tin the tromba pump Manner adverbs modifying the visible result of an event such as schlampig ‘sloppily’ (resultoriented) are compatible with akoma, manner adverbs modifying the initiator of the action such as vorsichtig ‘carefully’ (agent-oriented) are not: (42) (43) To thisavrofilakio itan (*akoma) prosektika anigmeno The safe was (still) cautiously opened The safe was still cautiously opened’ Ta malia mu ine (akoma) atsala xtenismena The hair my is still sloppily combed My hair is still sloppily combed On the basis of the above facts, Anagnostopoulou (2003) concluded that Voice can only be present in resultant state participles. Unambiguous resultant state participles readily accept agent-oriented modification: (44) a. b. Afto to theorima ine diatipo-meno This the theorem is formulated ‘This theorem is formulated by Ferma’ Afto to theorima ine prosektika This the theorem is carefully ‘This theorem is carefully formulated’ apo ton Ferma by the Ferma diatipo-meno formulated More evidence from verb classes and coercion (Anagnostopoulou 2013, in progress): a) Target state participles are not possible with agentive verbs (verbs of killing and creation), while there is no such restriction with resultant state participles based on these verbs: (45) Verbs of killing (murder and poison verbs, torture verbs) Aftos o anthropos ine (*akomi) dolofonimenos/ straggalismenos This the man is (*still) murdered/ strangled/poisoned dhilitiriasmenos/ pirovolimenos/ mexeromenos poisoned/ shot/ knived (46) Verbs of creation a. To vivlio mu ine (*akomi) grameno The book my is (*still) written b. O kiklos ine (*akomi) sxediasmenos The circle is (*still) drawn These restrictions can be explained if agentive verbs necessarily include Voice (Alexiadou, Anagnostopoulou & Schäfer 2006, to appear), and target state participles are not allowed to have Voice. b) Similarly with manner verbs implicating an agent: 13 (47) a. b. ?Ta magula tis ine kokina epidi ine xastukismena The cheeks her are red because are-3pl slapped “Her cheeks are red because they are slapped’ *Ta magula tis ine akomi xastukismena, gi’afto ponai The cheeks her are still slapped, for this hurts-3sg “*Her cheeks are still slapped, that’s why she hurts” Result state Target state c) In Greek (though not universally), participles based on denominal verbs overtly encoding an instrument, for example participles based on verbs belonging to Levin’s (2003) wipe-class, tolerate still under coercion leading to a result interpretation of the participle. In the presence of still, the participles are understood to mean “clean”, “un-wrinkled” or “tidy”: Wipe verbs incorporating an instrument (48) a. b. c. To patoma ine (akomi) sfuggarismeno Coercion into a result-reading The floor is (still) sponge-meno ‘The floor is still swabbed (i.e. clean)” To pukamiso ine (akomi) sideromeno The shirt is (still) iron-meno ‘The shirt is still ironed (i.e. unwrinkled)” Ta mustaki tu ine (akomi) psalidismeno The mustache his is (still) scissors-meno ‘His mustache is still trimmed (i.e. tidy)’ Adding the agent-oriented adverb epaggelmatika ‘professionally’, which is licensed by Voice, brings out the manner/ instrument interpretation of the participles, making the presence of “still” outright ungrammatical: (49) a. b. To patoma ine (*akomi) sfuggarismeno epaggelmatika Coercion impossible The floor is (still) sponge-meno professionally when Voice is present ‘*The floor is still professionally swabed” To pukamiso ine (*akomi) sideromeno epaggelmatika The shirt is (still) iron-meno professionally ‘*The shirt is still professionally ironed” With the instrument reading, the clean-ness/ unwrinkled-ness etc. of the result can be denied (cf. Rappaport Hovav & Levin 2010 on manner – result complementarity): Voice→ manner and no coercion (no result, no contradictions when the result is denied): (50) a. b. To patoma ine sfuggarismeno epaggelmatika ki omos ine vromiko The floor is sponge-meno professionally and however is dirty ‘The floor is professionally swabbed but still dirty” To pukamiso ine sideromeno epaggelmatika ki omos ine tsalakomeno The shirt is iron-meno professionally and however is wrinkled ‘The shirt is professionally ironed but still wrinkled” 14 In the presence of “still”, denying the cleanness/tidyness leads to a contradiction: Target state: manner coerced to result (evidenced by contradictions): (51) a. b. *To patoma ine akomi sfuggarismeno ki omos ine vromiko The floor is still sponge-meno and however is dirty ‘The floor is still swabbed but still dirty” *To pukamiso ine akomi sideromeno ki omos ine tsalakomeno The shirt is (still) iron-meno and however is wrinkled ‘The shirt is still ironed but still wrinkled” Voice forces a manner interpretation and “still” is impossible (target state participles construal disallowed). Absence of Voice allows for coercion of manner into a result interpretation and “still” is possible (target state construal allowed).3 Conclusion: The absence of Voice in target state participles explains why Voice modifiers to the underlying causing event are ruled out in all languages under consideration, even in languages like Greek which has been seen to be very liberal when it comes to Voice in resultant state participles. 2.5.2. Analysis of target state participles We saw three conditions on target state participles: 1) Voice impossible with target state partici[les 2) Result verbs and coerced manner verbs yield target state participles (see Rappaport Hovav & Levin 1998, 2010 on result vs. manner) 3) Manner verbs can be coerced into result verbs, but only in the absence of Voice These facts can be explained if target state participles represent outcomes of scalar changes. See Baglini (2011) for a semantic proposal,4 Gehrke (to appear), Anagnostopoulou (2013, in progress). See Rappaport Hovav (to appear) and appendix 1 for more on scalar changes. Verbs classified as ‘result verbs’, among them change of state verbs, encode the scale in their lexical meaning: -darken describes changes in illumination, -cool describes changes in temperature, -lengthen describes changes in length, -deepen describes changes in depth. 3See Anagnostopoulou (in progress) for implications concerning the manner-result complementarity hypothesis (Rappaport Hovav & Levin 2010). 4 Baglini and Gehrke claim that all adjectival participles represent outcomes of scalar changes and they argue against the resultant vs. target state classification of participles; I crucially differ in limiting this claim to target state participles which sharply differ from resultant state participles as far as Voice is concerned. 15 Change of state verbs lexically encode (“lexicalize”) the scalar component on the root (Rappaport Hovav & Levin 2010, Rappaport Hovav, to appear). Phrasal target state participles involve stativization of a verbal structure expressing the result. Embick (2009) proposes two structures for change of state verbs depending on whether they are based on roots expressing states (e.g. dark-darken) and roots expressing events (e.g. break-broken): (52) √DARK-type: Predicate of states √BREAK-type: Predicate of events; must co-occur with ST (vs. √POUND-type: Predicate of events; can’t co-occur with ST) (53) a. √DARK-type: complement of v v 3 v √DARK b. √BREAK-type: modifier of v selecting for a “Proxy ST” v 3 v 3 √BREAK v ST In (53b), ST is an ‘empty’ state defined by the modifier of v, √BREAK; Embick calls it a “Proxy ST”. Its meaning is filled by the content of √BREAK expressing a “state caused by a breaking event” = broken (cf. Marantz 2013b who generalizes this to all change of state verbs in English). By contrast, an ‘adjectival’ root of the √DARK type fills the ST position selected by v. To these, the following structure must be added for verbs with non-compositional prefixes like auf-pumpen (Alexiadou & Anagnostopoulou 2013): (54) v 3 v PP 3 6 √PUMPv auf 16 Local determination of the meaning of “auf” in the context of √PUMP. In line with Marantz’s (2013a) theory of local contextual allosemy (see Anagnostopoulou & Samioti 2012, 2013 and below for an analysis of participles in terms of local allosemy). In target state participles, (52)-(54) are embedded under the Adjectivizer/Stativizer: (55) Adj 3 Adj/Stat Asp 3 Asp v 3 v ResultP (= RootP, ST or PP) The presence of Voice leads to a complex interaction between the agent, the instrument (if present) and the theme, cumulatively rendering the change too complex to qualify as scalar. See Rappaport Hovav (to appear: 36-41) for a discussion of incremental theme verbs along these lines.5 [See Appendix II for some potential counterexamples to the claim that target state participles can’t contain Voice, and how they must be dealt with.] 3. Two domains in participles: v and Voice 3.1. Two meaning domains 3.1.1. The domain for allosemy Greek “lexical” –tos participles often have a special meaning, a meaning not related to the corresponding verbs. By contrast, “phrasal” –menos forms that always have a compositional meaning identical to that of the corresponding verbs; Anagnostopoulou & Samioti 2012, 2013): (56) a. spas-tos break-tos “folding” vs. b. spas-menos break-menos “broken” The literal meaning of “break” is retained in the -menos participle in (56b), while (56a) has a distinct interpretation: it means “folding” and applies to objects consisting of parts that can “break into” smaller pieces: 5 Rappaport Hovav (to appear) argues that even with mow used by Dowty as an example of an incremental theme verb the change is non-scalar: “…if mowing involves changing the grass in the lawn from being tall to being short, there is no relation between the parts of the event of mowing and transition from (any of the grass) being tall to being short. That is, the grass does not get shorter as the event of mowing progresses”. 17 (57) spas-ti ombrella/ break-ti-fem umbrella / ‘folding umbrella’ (58) spas-meni break-meni-fem ‘broken umbrella” spas-to break-to-neut ‘folding table’ ombrella/ umbrella/ trapezi table spas-meno break-meno-neut ‘broken table’ trapezi table The two meanings are closely related but sufficiently distinct to be expressed by different roots in a language like English. Another pair illustrating a clear meaning difference between the two forms is based on the verbal roots FTIAX (‘make’), PLAS- (‘form/create’): (59) a. (60) a. ftiax-tos make-tos ‘made up/ fake’ plas-tos create-tos ‘false/forged’ b. b. ftiag-menos make-menos ‘made’ plas-menos create-menos ‘created’ The –menos forms in (b) have the regular meanings of the corresponding verbs “make” and “create”. In the –tos forms in (a) the root has the interpretation “made up”, i.e. something “made”/ “created” and therefore unreal/ false/fake. Analysis Marantz’s (2013a) local theory for contextual allosemy: (61) Contextual allosemy = the choice of the meaning of a morpheme – the appropriate alloseme- within a local environment. Local contextual allosemy is sensitive to phases: once a designated phase-head is merged with a complement, it triggers its spell-out (62a), and a higher head in (62b) can no longer serve as context for the choice of an alloseme located below that phase head. (62) a. Phase 3 Phase-head Complement = Spelled Out b. XP 3 X Phase-head Phase 3 Complement = Spelled Out 18 In (62) X cannot determine allosemic choice of Complement across a Phase-boundary (Complement is spelled out once the phase head is merged). Assuming categorizing heads to be phase heads, two category heads are relevant for adjectival participles: v: a: present in phrasal, but not in lexical participles present in lexical participles (and in phrasal ones, but attaching outside v). In phrasal –menos participles, v attaches to RootP prior to adjectivization: Phrasal participles: they retain the verbal meaning (as we saw above for –menos participles) (63) aP 3 a vP = Spelled Out in the context of a = verbal meaning 3 v RootP = Spelled Out in the context of v= choice of an alloseme in the context of v (the “verbal” meaning of the root) Necessary assumption: Root-to-v movement (necessary to bring the Root together with the head that types it as a category, see Marantz 2013a, 2013b for discussion): (64) aP 3 a vP = Spelled Out when a is merged (a=phase head) 3 √ROOT - v RootP = Spelled Out when v is merged (v=phase head) 3 √ROOT In (64) a can no longer access the √ROOT: a semantically contentful v (introducing an event variable) intervenes. Lexical participles: they can have a different meaning from the verbal one (as we saw above for –tos participles) (65) tosP 3 tos RootP = 3 choice of an alloseme the context of -tos (potentially different alloseme than in the context of v). 19 3.1.2. The domain for idioms Allosemy should not be conflated with idioms (see Anagnostopoulou & Samioti 2012, 2013; Marantz 2013a for discussion). The boundary for idioms in participles is Voice (as proposed by Marantz 1997; contra Marantz 2001, 2007). Both –tos and –menos forms may form idioms. “Spas-tos” and “spas-menos” can form closely related (though not identical) idioms, a fact showing that the idiomatic interpretation of participles is a separate issue: (66) a. b. spas-ta Ellinika break-ta-pl.neut Greek ‘broken Greek (not very good Greek)’ spas-meni foni broke-meni-fem.sg voice ‘broken voice (voice that does not sound clear)’ Only the –menos participle in (67b) can have an idiomatic reading (it means stoned/ drugged or angry), while the -tos form not: (67) a. ftiax-tos make-tos ‘made up/ fake’ b. ftiag-menos make-menos ‘made’ Menos idioms do not necessarily inherit the idiomatic interpretation of the corresponding verb, while they necessarily inherit the compositional interpretation of the corresponding verb: A) Idiomatic verb kathar-iz-o lit. ‘clean’ idiom. ‘kill’ Non-idiomatic participle kathar-iz-menos only lit. ‘cleaned’ B) Non-idiomatic verb trav-a-o only lit. ‘pull’ Idiomatic participle trav-ig-menos lit. pulled idiom ‘far fetched’ C) Idiomatic verb: one meaning xon-ev-o lit. ‘digest’ idiom. ‘like’ Idiomatic participle: another xon-e-menos lit. ‘digested’ idiom. ‘understood’ 20 Agentive Voice is a boundary for Idioms. Whenever agentive Voice is present, participles can only be interpreted compositionally. Extensive evidence for this discussed in Anagnostopoulou & Samioti (2012, 2013) for Greek: Agentive adverbs like ‘deliberately’ and agent-oriented manner adverbs like ´carefully’ systematically block idiomatic interpretations: (68) a. b. trav-ig-menos lit. pulled idiom ‘far fetched’ stri-menos lit. twisted prosektika / skopima travigmenos only lit.: carefully/ deliberately pulled prosektika/ skopima strimenos only lit: carefully/ deliberately twisted idiom. ‘crotchety’ The same effect is triggered by agentive and instrument PPs: (69) stri-menos jeros BUT stri-menos apo kapion/ me kati idiom. ‘crotchety man’ only lit. ‘twisted by someone / with something’ The effects of agentivity illustrated above are consistent with Marantz’s (1996, 1997) generalization that the syntactic head that projects an agent defines a locality domain for special meanings: Just as for phrasal idioms: see Harley & Stone (2013) who reanalyze all counterexamples to the agentivity generalization (cases like “a little bird told him”, “vom Teufel geritten sein” etc.) and show that it is never the case that a true agent is in the fixed part of an idiom, to the exclusion of an object. Phases potentially provide an explanation for this: if agentive Voice is an inescapable phase head, sending its complement for spell-out, and if “Encyclopedia Search Domains” (Borer 2008, 2009) for idioms coincide with (LF) spell-out domains, an Agent will never be able to be included in an idiom. 3.2. Two form domains Embick (2003): Two points for Vocabulary Insertion (VI) in English participles: VI is divided into (i) the Root Cycle for Root-attached ASP and (ii) an Outer Cycle (above v) for the non-Root attached ASP: (70) a. Spell-Out of ASP: Root Cycle ASP ↔ -en / ______ {√ROT, √SHRINK,.....} ASP ↔ -0 / _______ {√OPEN, √EMPTY, ...} ASP ↔ -t /________ {√BEND,.......} ASP ↔ -èd/ _______ {√BLESS, ALLEGE, AGE, .....} ASP ↔ -ed/ ______ {√CLOSE, OBSTRUCT,.....} 21 b. Spell-Out of ASP: Outer Cycle (above v) ASP ↔ -en / _____ { √BREAK, √SPEAK,.....} ASP ↔ -0 / ______ { √HIT, √SING, √SHRINK, ....} ASP ↔ -t / ______ {√BEND, √BUY,.......} ASP ↔ -ed As a result: -√ROT is on a list in the Root Cycle for –en, but not in the Outer Cycle (where default –ed surfaces): rotten vs. rotted. -√ SHRINK is on a list for –en in the Root Cycle and on a list for 0 in the Outer Cycle. -etc. We can think of the distribution of –tos vs. –menos in Greek participles along similar lines, treating –menos as the default exponent of ASP at the Outer Cycle, similarly to English –ed in (70b), and –tos as the default exponent of ASP at the Root. See Anagnostopoulou (in progress) for details. 4. Summary In this talk I argued for the following points: Participles split into the following Adjectivizer/Stativizer attaches: [Adj/Stat [Adj/Stat [Adj [Stat [Adj [ Root ]] [ v [STATE/ST/PP ]]] [Voice [v [ ]]]]] [Voice [v [ ]]]] types, depending on where the = lexical = target state = Greek resultant state =English/ German resultant state There are two meaning domains in participles, the Root domain (domain for allosemy) and the Outer domain (above v). The boundary for idiomatic meanings in participles is Voice. VI can take place at the Root Level and the Outer level (above v), giving rise to different spell out options for the adjectivizer/stativizer attaching directly to roots vs. above v. Some references Alexiadou, A. & E. Anagnostopoulou. 2008. Structuring participles. Proceedings of WCCFL 26. Alexiadou, A & E. Anagnostopoulou 2013. Manner vs. Result Complementarity in Verbal alternations: a view from the Clear-alternation. Proceedings of the North East Linguistic Society 42, University of Toronto, November 11-13 2011, 39-52. Alexiadou, A., E. Anagnostopoulou & F. Schäfer. 2006. The Properties of Anticausatives Crosslinguistically. In M. Frascarelli (ed.), Phases of Interpretation. Mouton de Gruyter, 187-212. Alexiadou, A., E. Anagnostopoulou & F. Schäfer. To appear. External arguments in transitivity alternations. A layering approach. Oxford: Oxford University Press. 22 Alaexiadou, A., B. Gehrke & F. Schäfer. 2013. The argument structure of adjectival participles revisited. Ms. University of Stuttgart & CNRS. Anagnostopoulou, E. 2003. Participles and Voice. In A. Alexiadou, M. Rathert & A. von Stechow (eds.) Perfect Explorations. Mouton de Gruyter. Anagnostopoulou, E. 2013. Voice related properties in different types of adjectival participles. Paper presented at the International workshop on Participles, University of Konstanz, October 8-9. Anagnostopoulou, E. In progress. Domains in Participles. Ms. University of Crete. Anagnostopoulou, E. & Y. Samioti. 2012. Domains within Words and their meanings : a case study. To appear in A. Alexiadou,, H. Borer & F. Schäfer (Eds.) The syntax of Roots and the roots of Syntax. Oxford: Oxford University Press. Anagnostopoulou, E. & Y. Samioti. 2013. Allosemy, idioms and their domains : Evidence from adjectival participles. To appear in Folli, R., Sevdali, C. and Truswell, R., Eds., On Syntax and its Limits. Oxford: Oxford University Press. Baglini, R. 2011. The scalar source of stative passives. Sinn und Bedeutung 2011. Bruening, B. To appear. Word Formation is Syntactic: Adjectival Passives in English. Ms. University of Delaware. Natural Language and Linguistic Theory. Chomsky, N. 2001. Derivation by phase. Ken Hale: a life in language. Cambridge, MA: MIT Press. Embick, D. 2003. Locality, Listedness and Morphological Identity. Studia Linguistica 57: 143-169. Embick, D . 2004. On the structure of resultative participles in English, Linguistic Inquiry 35: 355392. Embick, D. 2009. Roots, States and Stative Passives. Paper presented at the Root Workshop, University of Stuttgart, June 2009. Gehrke, B. 2011. Stative passives and event kinds. In Reich, I., Horch, E. & Pauly D, Eds., Proceedings of Sinn und Bedeutung 15, Saarbrücken, Universaar-Saarland University Press: 241257. Gehrke, B. To appear. Different ways to be passive. Natural Language and Linguistic Theory. Grimshaw, J. 1990. Argument Structure. Cambridge MA: MIT Press. Kratzer, A. 1994. The Event Argument and the Semantics of Voice. Ms., University of Massachusetts at Amherst. Kratzer, A. 1996. Severing the external argument from its verb. In J. Rooryck & L. Zaring. Dordrecht: Kluwer, 109-137. Kratzer, A. 2000. Building statives. Berkeley Linguistic Society, 26. Maienborn, C. 2007. Das Zustandspassiv: Grammatische Einordnung – BildungsbeschränkungInterpretationsspielraum. Zeitschrift für germanistische Linguistik 35: 83-44. Marantz, A. 1997. No escape from Syntax: Dont' try morphological analysis in the privacy of your own lexicon. In: Dimitriadis, A, Siegel, L (eds) University of Pennsylvania Working Papers in Linguistics, Philadelphia:University of Pennsylvania, pp 201-225. Marantz, A. 2001. Words and Things. Ms., MIT & NYU. Marantz, A. 2007. Phases and words. S. H. Choe et al, (eds.) Phases in the theory of grammar. Seoul: Dong In Publisher. Marantz, A. 2013a. Locality Domains for Contextual Allomorphy Across the Interfaces. To appear in Distributed Morphology Today, Festshrift for Morris Halle. MIT Press. Marantz, A. 2013b. Locating the Verbal Root. Handout of talk presented at the Scandinavian Conference in Linguistics 2013. McIntyre, A. 2012. Adjectival Passives and Adjectival Participles in English. To appear in A. Alexiadou & F. Schaefer (Eds.) Non-Canonical Passives. Amsterdam: John Benjamins. Meltzer-Asscher, A. 2011. Adjectival passives in Hebrew: evidence for parallelism between the adjectival and verbal systems. Natural Language and Linguistic Theory 29: 815-855. Parsons, T. 1990. Events in the Semantics of English. A Study in Subatomic Semantics. Cambridge Mass: MIT Press. Paslawska, A. & A. von Stechow. 2003. Perfect readings in Russian. In A. Alexiadou, M. Rathert & A. von Stechow (eds.) Perfect Explorations. Mouton de Gruyter. Rapp, I. 1996. Zustand ? Passiv ? Überlegungen sum sogenannten “Zustandspassiv”. Zeitschrift für Sprachwissenschaft 15.2 : 231-265. 23 Rappaport Hovav, M. & B. Levin. 1998. Building verb meanings. In Butt, M. & W. Geuder, Eds., The Projection of Arguments : Lexical and Compositional Factors. Stanford : CSLI Publications, 97-134. Rappaport Hovav, M. & B. Levin, 2010. Reflections on manner/result complementarity. In Doron, E, Rappaport Hovav, M & I. Sichel, Eds., Syntax, Lexical Semantics and Event Structure. Oxford : Oxford University Press, 21-38. Rappaport Hovav, M. To appear. Building Scalar Changes. In A. Alexiadou,, H. Borer & F. Schäfer, Eds., The syntax of Roots and the roots of Syntax. Oxford: Oxford University Press. Schlücker, B. 2005. Event-related modifiers in German adjectival passives. In Maier, E., Bary, C. & J. Huitink, Eds., Proceedings of Sinn und Bedeutung 9, Radboud University Nijmegen : 417-430. Von Stechow, A. 1998. German participles in Distributed Morphology. Ms. University of Tübingen. Appendix I : A bit more on scalar changes Scalar change of state verbs are further distinguished into the subclasses in (i) proposed by Rappaport Hovav (to appear; she builds on work that has distinguished a number of different subtypes of property scales Cruse 1980; Hay, Kennedy & Levin 1999, Winter and Rothstein 2004, Kennedy & McNally 2005, Kearns 2007, Kennedy & Levin 2008): (i) Scale type 2 valued Multi-point upper bound Multi-point lower bound Multi-point open Adjectival property base De-adjectival verb dead flat wet long die flatten wet lengthen De-verbal Property base Verb cracked thawed streched crack thaw all –ed participles all verbs strech Table (i) includes the following distinctions: 1) Verbs expressing scales associated with multi-valued attributes vs. verbs that lexicalize a twopoint scale (Beavers 2008). 2) Multi valued scalar verbs are further divided into three subclasses: a) Open scale verbs associated with predicates like long, wide, or stretch, shrivel with no maximal value. b) Upper bound close scale verbs with a maximal degree based on predicates such as flat, empty, clear, dry, straight, freeze, thaw. c) Close scales with a lower bound which represents a non-zero degree of the measured property, e.g. wet, impure. As argued for in Anagnostopoulou (in progress), only certain subtypes of scalar verbs in table (i) yield states that can be conceptualized as transitory and reversible. class 2c verbs are optimal inputs for target state participle formation, class 2a verbs are impossible inputs, 24 class 1 and 2b verbs are sometimes licit and sometimes excluded, depending on various factors, among them also Embick’s (2009) “competition for use” effect (i.e. dry blocking dried, empty blocking emptied). Appendix II: Apparent counterexamples to the claim that target states cannot include Voice It is possible to allow Voice modifiers in target state participles when the referents of by phrases and instruments are present in the state and the adverbs modify the state directly. 6 The Greek data showing this (Anagnostopoulou 2013, in progress, Alexiadou, Anagnostopoulou & Schäfer, to appear): (1) By-phrases introducing referents present in the state a. To stadio ine akomi perikiklomeno apo tin astinomia The stadium is still surrounded by the police ‘The stadium is still surrounded by the police’ Instruments introducing referents present in the state b. O skilos ine akomi demenos me skini The dog is still tied with leash ‘The dog is still tied with a leash’ Manner adverbs modifying the state directly c. O skilos ine akomi demenos sfixta The dog is still tied tight ‘The dog is still tight tied’ d. To stadio ine akomi filagmeno prosektika The stadium is still guarded carefully ‘The stadium is still carefully guarded’ Proposal In these examples, Voice attaches only after Asp has applied to the v+STAT complex, hence introducing the holder of the target state. (In (2) the adjectivizing head attaches above ASP; it is possible that the two heads are conflated): (2) target state participles with Voice Voice 3 in Greek, German, English Voiceholder Adj 3 Adj Asp 3 Asp v 3 v 6 ResultP (= RootP, ST or PP) This is exactly the condition identified by Meltzer-Asscher (2011) for Hebrew. Meltzer-Asscher argues that Hebrew only has target state participles. At the same time, though, she argues that target state participles based on transitive verbs always entail an external argument. This is unexpected in the present approach and, if true, it would be problematic. The issue requires further investigation, especially in view of Doron’s (2012) considerably different description of the state of affairs in Hebrew. 25 By contrast, recall that resultant state participles include an agentive Voice under Adj: Greek resultant state participles German/ English resultant state participles (3) a. Adj 3 Adj b. AspP 3 Asp Adj 3 Adj VoiceP 3 Voiceagent VoiceP 3 Voiceagent vP 5 vP 5 Evidence for different height of attachment of Voice in target state participles German provides evidence in favor of the different height of Voice in target state participles, as opposed to resultant state participles, relative to the stativizer/adjectivizer. Two types of by-phrases in German adjectival participles, which are intonationally, semantically and syntactically distinct: (i) kind by-phrases of the type discussed in Gehrke (2011, to appear) and many others (see above for discussion) (ii) by-phrases present in the state discussed in Rapp (1996), Schlücker (2005) and Gehrke (2011). The latter type is seen in (4): (4) a. b. Er ist von der Musik beindruckt He is by the music impressed ‘He is impressed by the music’ Das Haus ist von Studenten bewohnt The house is by students in-lived According to Gehrke (2011), these modifiers can only combine with stative predicates, i.e. the following is degraded, for her: (5) weil der Saal von der Heinrich-Böll-Stiftung gemietet ist because the hall by the Heinrich-Böll-foundation rented is Schlücker (2005), however, takes (5) to be well-formed, and other native speakers’s judgements side with Schlücker’s (Florian Schäfer, p.c.): “angemietet” makes the example even better. These examples are compatible with “immer noch” (still), i.e. they qualify as target state participles. 26 Intonational differences By-phrases are intonationally distinct, depending on whether they modify the underlying event-kind or directly the state (Schlücker 2005). When they modify the state they do not form a prosodic unit with the participle. Neutral stress falls on the participle, secondary stress on the modifier: (6) weil Peter von dem GeJÁmmer geNÈRVT ist because Peter by the lamentation annoyed is When they modify the event-kind, they form a prosodic unit with the participle, with neutral stress on the modifier: (7) weil die Wände von FEUer geschwärzt sind because the walls by the fire blackened are Semantic differences By-phrases are semantically distinct, depending on whether they modify the underlying event-kind or directly the state (Schlücker 2005, Gehrke 2011, to appear). Animates can be expressed through proper names or members of a group denoted by a collective noun, e.g. von dem Polizist ‘policeman’ and inanimates can surface as definite mass nouns, e.g. vom Feuer ‘by the fire’ or von der Bombe ‘by the bomb’. By-phrases modifying event kinds are expressed through collective nouns, e.g. von der Polizei and inanimates are expressed by generic mass nouns, e.g. von Feuer and indefinites von einer Bombe/ von Bomben. Syntactic differences The two types of von-PPs in German behave differently with respect to movement operations. No scrambling of kind by-phrases Von-phrases creating ‘well-established sub-kinds’ are not allowed to scramble across adverbs, negation and floated quantifiers, as shown below, even when they are definite, as in (108), or proper names, as in (109), which usually have to undergo scrambling in German: (8) a. b. (9) a. b. (10) a. b. c. d. Die Wände sind nicht von Feuer geschwärzt *Die Wände sind von Feuer nicht geschwärzt ‘The walls are not blackened by fire’ (Shlücker 2005: 423, ex. (22)) Seine Töchter waren wahrscheinlich von der Sangesmuse geküßt *Seine Töchter waren von der Sangesmuse wahrscheinlich geküßt ‘His daughters are probably kissed by the muse of singing’ (Shlücker 2005: 424, ex. (23)) ?Das Bild ist wahrscheinlich von Picasso gemalt *Das Bild ist von Picasso wahrscheinlich gemalt ‘The painting is probably painted by Picasso’ ?Die Bilder sind alle von Picasso gemalt *Die Bilder sind von Picasso alle gemalt ‘The paintings are all painted by Picasso’ (Winfried Lechner, p.c., based on a Picasso example by Rapp 1996) 27 Scrambling of target-state by-phrases allowed Von-phrases introducing a referent present in the target state7 are allowed to undergo scrambling: (11) a. Peter ist von dem Gejammer nicht genervt b. Peter ist nicht von dem Gejammer genervt ‘Peter is not irritated by the lamentation’ (Shlücker 2005: 423, ex. (18c,d)) (12) a. Das Haus ist von den Demonstranten wahrscheinlich schon tagelang umstellt b. Das Haus ist wahrscheinlich von den Demonstranten schon tagelang umstellt ‘The house is probably surrounded by the demonstrators many days now’ (Winfried Lechner, p.c.) Rightward movement across the participle A similar contrast is found with rightward movement across the participle (Gehrke 2011): (13) a. b. (14) a. b. Der Brief war von einem Experten geschrieben *Der Brief war geschrieben von einem Experten ‘The letter was written by an expert’ (Winfried Lechner, p.c., and Gehrke 2011, ex. (19c) ) Er ist von der Musik beeindruckt Er ist beeindruckt von der Musik ‘He is impressed by the music’ (Gehrke 2011, ex. (21a) and Winfried Lechner, p.c.) In the present analysis, their different syntactic behavior can be attributed to the fact that Voice holder is attached above the adjectivizing head in target state participles while in resultatnt state participles, Voice attaches below the adjectivizing head. Under the hypothesis that the adjectivizing head creates a boundary for movement across it (e.g. by being a phase head, Chomsky 2001), then the vonphrases cannot escape that boundary in resultant state participles explaining their immobility; by contrast, the von-phrases in target state participles never have to cross the aP boundary since they are merged above it. 7 Schlücker (2005, ex. 20c,d; 21c,d) also provides the following examples showing that when the von-phrase relates to a specific causal event in a resultant state participle (as in Greek resultant state participles) it can undergo scrambling: (i) a. Die Mannschaft ist doch vom Gegner geschlagen b. Die Mannschaft ist vom Gegner doch geschlangen ‘The team is particle beaten by the opposing team’ (ii) a. Die Hasen sind alle von dem Jäger getroffen Die Hasen sind von dem Jäger alle getroffen ‘The rabbits are all hit by the hunter’ Many speakers, however, do not accept these examples to begin with, since they do not accept a specific causal event in German participles. Apparently, Schlücker’s (2005) grammar is more liberal, resembling the grammar of Greek. She also accepts (iii) (her example (1b)) which is rejected by many others: (iii) Die Suppe ist von Maja gewürzt The soup is spiced by Maja 28
© Copyright 2026 Paperzz