Proceedings Chapter
Reference
Verb second in Hebrew
SHLONSKY, Ur, DORON, Edit
SHLONSKY, Ur, DORON, Edit. Verb second in Hebrew. In: Bates, D. The Proceedings of the Tenth West Coast Conference on Formal Linguistics . Stanford, CA : Center for the Study of Language and Information, 1992. p. 431-446
Available at:
http://archive-ouverte.unige.ch/unige:83458
Disclaimer: layout of this document may differ from the published version.
1 / 1
430 1 Rice, K.
References
Anderson, Stephen. 1982. Wbere's morphology? Linguistic lnquiry 13. 571-612.
Anderson, Stephen. 1988. Morphological theory. In F.J. Newmeyer (ed.) Linguistics: rhe Cambridge Series: volume/. Linguistic Theory: Foundations.
146-191.
Baker, Mark & Kenneth Hale. 1990. Relativized rninimality and pronoun incorporation. Linguistic /TU[Uiry 21,298-297.
Chen, Matthew. 1987. The syntax of phonology: Xiamen tone sandhi.
Phonology Yearbook 4. 109-149.
Chomsky, Noam. 1988. Sorne notes on economy of derivation and representation. Mrf Working Papers inlinguisrics 10.
Dresher, B.Elan. 1990. Anglian smoothing and lexical phonology. Paper presented at the workshop on Lexical Phonology, Univ~ity of WashingtOn.
Hale, Kenneth & Elisabeth Selkirk. 1987. Govemment and tonal phrasing in Papago. Plwnology Yearbook 4. 151-183.
Hargus, Sharon. 1988. The lexical phono/ogy of Sekani. New York: Garland.
Hayes Bruce. 1989. The prosodie hierarehy in meter. ln P. Kiparsky & G.
Youmans (eds.) Perspectives on mecer. Orlando, Florida: Academie Press.
lnkelas, Sharon. 1989. Prosodie constituency in the /exicon. PhD. dissenation, Sta.nford University.
Karl, James. 1976. Navajo verb prefix phon.ology. New York: Garland.
Kari, James. 1989. Affix positions and zones in the Atba_paskan verb eomplex:
Ahma and Navajo. Inrernational Journal of American Linguisrics 55. 424-452.
McDonough, Joyce. 1990. Topics in che phono/ogy and morpho/ogy of Navajo verbs. Ph.D. dissenation, University of Massachusetts, Amh~t.
Nespor, Marina & Irene Vogel. 1986. Prosodie phon.ology. Dordreche Foris.
Palma dos Santos, Ana 1991. Negative inflection in the Athapaskan verb. ms.
University of Toronto.
Pollock, Jean-Yves. 1989. Verb movement, universal grammar, and the structure of IP. Linguistic lnquiry 20. 365-424.
Randoja, Tiina 1989. The phono/ogy and morpho/ogy of Halfway River Beaver.
Ph.D. dissertation, University of Ottawa
Rice, Keren. 1989. A grammar of Slave. Berlin: Mouton de Gruyter.
!lice, Keren. 1990a. Tne structure of the Slave (Northern Athapaskan) verb.
presented at the workshop on Lexical Phonology, University of Washington.
!lice, Keren. 1990b. Predieting the order of me disjunct morphemes in the Athapaskan languages. Paper presented at the Athapaskan Lingcistics Conference.
Riee, Keren. 199Ia. Prosodie constitnency inHare (Athapaskan): evidence for the foot. LingutJ 82. 201-245.
Rice, Keren. 1991 b. Intransitives in Slave (Northern Athapaskan): Arguments for Unaccusatives. 1 nrernarional Journal
of
Ame rie an Linguistics 57. 51-69.Saxon Leslie. 1986. The synrax of pronou.ns in Dogrib (Athapaskan): some rheoretical consequences. Ph.D. dissenation, University of California, San Diego.
Selkirk., Elisabeth & Tong Shen. 1990. Proscxlic domains in Shanghai Chinese.
ln S. lnkelas Sharon & Draga Zee (eds.) The phonology-synrax connection.
Chicago: The University of Chicago Press. 313-338.
Selkirk, Elisabeth O. 1982. The symax ofwords. Cambridge: MIT Press.
Selkirk, Elisabeth O. 1986. On derived domains in sentence phonology.
Phono/ogy Yearbook 3. 371-405.
Speas, Margaret. l990a. Functional heads and the Mirror Principle. ms.
University of MassachusettS, Amherst.
Speas, Margaret. 1990b. Phrase structure in naturallangULlge. Dordrecht: Kluwer.
Verb Second in Hebrew
Ur Shlonsky
Université du Québec à Montréal Edit Doron
Hebrew University, jerusalem
1. V2 and Inversion
subjects of Hebrew clauses can either appear preverbally, as in (1), or immediately after the verb, as
in (2).
(1) ha-miStara 9acra harbe pe9ilim ba-psita
(2)
the-police detained many activists in-the-raid ha-leilit.
the-nightly.
ba-psita in-the-raid pe9ilim.
activists.
ha-leilit 9acra ha-miStara harbe the-nightly detained the-police many
'The police detained many activists in the nightly raid. •
1 The authors wish to thank the participants of the workshop on Hebrew Syntax, Universite du Quebec a Montreal, 1990 and the audience at WCCFL, Tempe, AZ.
Shlonsky's work was supported by SSHRC grant # 411-85- 0012.
When subjects follow the verb, the verb must be preceded by sorne constituent, as shawn in the contrast between (2) and (3) .2
(3) *9acr~ ha-mist~ra harbe pe9ilim.
deta1ned the-poll.ce many activists.
'the police detained many activists. •
Althou~h ~e give the same gloss for (l) and (2), ther~:f adnu ~t1ng that the two have the same tru th con<;Ilt.Lons, 1t should be borne in. mind that the two
~ar1ant~ have different discourse functions and lllocut~onary force: The subject-initial clauses are pragmat.Lcally more neutral.
. _Follow~ng Shlonsky (1987), welabel the phenomenon l.llust:ra~ed ln (2) TRIGGERED INVERSION, tc distinguish it from a d1fferent phenomenon, mentioned in note 2, which resembles Romance FREE INVERSION.
Confining ou.rsel ves, th en, to the contrast displayed in (1) and (2), we see that Hebrew clauses cannat begin with a verb. This restriction is observationally similar ta the V2 constraint, familiar from the Germanie languages.
However, unlike, say, Standard German Hebrew V2 phenomena are not restricted tc root cla~ses or ta clauses emb~dded under verbs belonging to a particular class · SUbJ ects may follow the verb in clauses embedded under al! v7rbs which take sentential complements.
None~eless, 1n embedded clauses as in root clauses, sorne const1tuen~ .must preceed the verb, as shawn by the u~acce~t~1l1ty of (~b). In this respect Rebrew is more l.Lke Ylddl.s~ and Icelandic than German (see e.g., Vikner
(1990, sect1on 2.3) for a recent discussion).
2 However, presentational or optional, as in inversion aside.
if the verb is unaccusative, passive, the preverbal constituent is
( i) . We put the se cases of ' free • (i) ~ba-pSita . ha-leilit) ne9ecru harbe
1n-~h7 ra1d the-nightly PASSIVE-detain
pe9lllm. many
activists
'Many activists were detained (in the nightly raid.)'
(4) a. hicta9arnu Se-ba-psita ha-leilit
b.
(we) regretted that-in-the-raid the-nightly 9acra ha-mistara harbe pe9ilim.
detained the-police many activists.
'We regretted that the police detained many activists in the nightly raid. '
(we)
*
harbe many
hicta9arnu regretted pe9ilim activists
Se- 9acra ha-mistara that-detained the-police ba-psita ha-leilit.
in-the-raid the-nightly.
Sorne of the recent research into Yiddish and Icelandic converges on the idea that the V2 configuration in these languages has the structure in (5), where the preverbal constituent (XP) occupies [Specji], an A'
position and the verb is raised t o I (Diesing (1990), Rognvaldsson and Thrainsson (1990), Santorini (1989), Vikner (1990) among ethers). This explains the ward arder of embedded V2 and specifically the cooccurence of V2 and a lexical complementizer.
(5) CP
C IP
XP I '
I VP
NPsUBJ V'
v ... .
We believe that this analysis does not carry over directly ta Heb:r·ew, primarily because Hebrew embedded V2 clauses constitute strong Islands for extraction which they apparently do not in these two Germanie languages.
We discuss extraction out of embedded V2 structures in section 5. 3
Our view is that Hebrew triggered inversion is a case of CP recursion, along lines suggested also for Germanie by e.g., Platzack (1986), Holmberg (1986). We shall try ta defend the view that the structure of triggered inversion is as in ( 6) , where the verb has raised tc an empty C node and is preceded by the trigger.
3 The grammatical status of extraction out of an embedded V2 clause in Icelandic and Yiddish is bath complex and controversial. See Vikner (1990) for a lucid presentation of the various views on this matter.
(6) CP CP XP C'
C IP I'
I VP
V'
v
We argue that when the subject is clause-initial however, there is no neea to posit this extra layer of structure; preverbal subjects are in [Specji) at
s-
structure. We claint that the different position occupied by preverbal subjects and triggers for inversion is what accounts for their different syntactic properties.
2. What can Be a Triqqer?
Triggers can be sentential adverbs, adverbial PPs and clauses, direct and indirect objects of the verb, and
<:ll ~ypes _of clausal complements. A partial illustration 1s g1ven 1n (7).
(7) a. Temporal Adverb
'etmol 9acra ha-miStara harbe pe9ilim.
yesterday detained the-police many activists.
'The police detained many activists yesterday. •
b. PP Adjunct
ba-pSita ha-leilit 9acra ha-mistara harbe in-the-nightly raid detained the-police many pe9ilim.
activists.
'The police detained many activists in the nightly raid. •
c. Clausal Adjunct
mi-bli lekabel 'iSur mi-gavoha 9acra with-out to get permission from-high detained ha-mistara harbe pe9ilim.
the police many activists.
'The police detained many activists without getting authorization from higher up. •
d. Direct Object
harbe pe9ilim 9acra ha-mistara ba-psita many activists detained the-police in-the-raid ha-leilit.
the nightly.
'The police detained many activists in the nightly raid. '
e. Indirect Object
la-taxana hesi9a to-the-station drove 9acurim.
detainees.
'The police brought station.'
f. Clausal Complement
ha-miStara •et ha- the-police ace the-
the detainees to the
lo la9asot ra9aS be-meSex ha-nesi9a bikSa ha- neg to-make noise during the-ride asked the- mistara min ha-9acurim.
police from the-detainees.
1 The police asked the detainees not to make noise during the ride.'
There are several restrictions on what can serve as a trigger. First, VP (manner) adverbials may not be triggers, as shown by the ungrammaticality of (8).
(8) a. *le'at likek ha-xatul 'et Rina.
slowly licked the-cat ace. Rina 'The cat slowly licked Rina.' b. *maher nasax ha-xatul •et Rina.
quickly bit the-cat ace. Rina 'The cat quickly bit Rina.'
We assume that adverbs do not move. This implies that when sentential adverbs serve as triggers, they must be base-generated in trigger position. One syntactic difference between VP adverbs and sentential adverbs is that the former may be generated only in positions adjoined to sorne VP projection while sentential adverbs may appear in a variety of positions, as shown in the contrast between (9a) and (9b), where " indicates a possbile adverb position. 4 Note also that the possibility of an adverb intervening between V and the
4 Following Travis (1988), one might also argue that manner adverbs are heads and perforee, barred from trigger position, a position which can only host XPs.
direct obj ect is evidence that v rai ses tc I in the syntax of Hebrew.
(9) a. Positions of Sentential Adverbs
A Dani A 'axal A lexem A
Dani ate bread b. Positions of VP Adverbs
Dani 'axa! A lexem A
. ~econd, only an XP belonging tc the clause where 1nvers1on occurs ~àn serve as a.trigger. Thus, (10) is onl~ .acce,P't:able 1f the PP le-harbe pe9ilim •tc many act1v1sts 1s construed with the higher verb, as shawn in the glosses. We assume 1:hat long movement proceeds through a [SpecjC] marked [+WH]. Since triggers are[- WH], they cannet make use of [SpecjC] as an escape hatch.
(10) a. le-harbe pe9ilim hodi9a ha-mistara se-
3. Topics
t~-many.activists announced the-police that- hl tag1S tvi9a.
she presses charge.
i. 'The police told many activists that it will press charges.'
ii. *'The police announced that i t will press charges against many activists.'
. Triggers ~ust be dist inguished from topicalized co~t1tuents~ l.e. ~ constituents which precede the sub]ect. Wh1le top1cs may also appear in beth root and em.J;>edded clauses, they ohey a number of restrictions Whl~h set them apart from triggers. First, in Hebrew, top1c;:; must be definite while triggers may be indefinite.
The differe~c~ ~etwe~n (7d) above and (lla) below is that the cla~se-lnltlal d1rect abject in (lla) is fol lowed by
th7
sub]ect; hence it is not a trigger but a tepic, since trlggers are fol lowed by the verb. The sentence is unacceptable. because tapies cannet be indefini te. In(llb), .the d1rect abject is definite and thus licensed as a genu1ne topic.5
(11) a. *harbe pe9ilim ha-miStara 9acra many activists the-police detained ba-psita ha-leilit.
in-the-raid the nightly.
The police detained many activists in the nightly raid. •
It should be noted, however, that contrastive stress on the tepic in (lla) renders i t more acceptable.
b. 'et ha-pe9ilim ha-politiyim ha-miStara ace the-activists the-political the-police 9acra ba-pSita ha-leilit.
detained in-the-raid the nightly.
'The police detained the political activists in the nightly raid.'
c. noda9 la-nu Se-'et ha-pe9ilim ha- was known to-us that-acc the-activists the- politiyim ha-mistara 9acra ba-psita ha- political the-police detained in the-raid the leilit.
nightly.
' I t became known tc us that the police detained the political activists in the nightly raid.' Second, tapies, unlike triggers, are not clause- bound. Contrast (10) above with (12) below and note the ambiguity in the interpretation of the tapie, absent in
( 10) .
(12) a. la-pe9ilim ha-politiyim tc-the activists the-political hodi9a Se- hi tidroS announced that-she will demand ba-po9al.
sentence.
ha-mistara the police ma'asar
jail
i. 'The police told the political activists i i .
that i t will demanda jail sentence.' 'The police announced that it will demand a jail sentence for the political
activists.'
We assume that tapies are adjoined tc IP, as in (13) below.
(13) ... [1p the activists [1p the police arrested ... ] ]
4. Trigqers and Topics
Consider, now, the fact that triggers and tapies cannet cooccur. (14a) shows a case of a a tepic preceding a trigger and (14b) of trigger preceding a tepic. Beth are bad.
(14) a. Tepic > Trigger
leilit
*'et ha-pe9ilim ha-polityim ba-pSita ace the-activists the-political in-the-raid
a
9acra ha-mistara.
the-nightly detained the police.
'The police detained many Arab activists in the nightly raid.'
b. Trigger > Tepic
*harbe pe9ilim ba-pSita many activists in-the-raid 9acra ha-miStara.
detained the-police.
'the police detained many nightly raid. •
ha-leilit the-nightly
activists in the
I I t riggers were in (Specjl ], nothing should prevent (14a), where a tepic precedes a trigger, since triggers are in (Specj i], while tapies are adjoined tc IP. However, if triggers were in [SpecjCJ, then the ward arder displayed in (14a) could only mean that the tqpic is adjoined tc CP. Adjunction tc CP is independently ruled out , as shawn by the unacceptabi_Lity of (15).
(15) *le yada9ti 'et ha-tapuax ha-ze le-mi
ru led shows (16)
neg (I) knew ace the-apple the-this tc whom Dani natàn.
Dani gave.
'I didn't know tc whom Dani gave this apple.' (14b), where the trigger precedes the tepic, is out on par with wh-extraction over a tepic: (16) that tapies create islands for wh-movement.
*'eize pe9ilim ba-pSita ha-leilit ha- which activists in-the-raid the-nightly the- mistara 9acra?
police detained?'
Which activists did the police detain in the nightly raid?'
s.
Triggered Inversion and Wh-movementOne major difference between Hebrew and the Germanie languages with embedded V2 has tc do with the interaction of V2 with wh-movement, as mentioned briefly above.
In this section we show that clause-initial subjects and clause-initial triggers diverge in their interaction with wh-movement.
The data in (17a)-(20a) demonstrate that both long and short wh-movement is grammatical over a preverbal subject while similar movement over a trigger is ungrammatical (17b)- (20b).
(17) Short Wh-Movement
a. le-mi Dani natan harbe sfarim 'etmol?
tc whom Dani gave many books yesterday?
'To whom did Dani give many books yesterday?'
(18)
b. *le-mi harbe sfarim natan Dani 'etmol?
tc whom many books gave Dani yesterday Long Wh-Movement
a. le-mi Rina xoSevet to whom Rina thinks sfarim?
se- Dani natan harbe that Dani gave many
b.
books?
1 To whom does Rina think that Dani gave many books?'
*le-mi Rina xoSevet Se- tc whom Rina thinks that- natan Dani?
gave Dani
harbe sfarim many books
(19) Relativization
a. Ze ha-sefer Se- Dan natan le-Rina.
this the-book that-Dan gave to-Rina.
'This is the book that Dan gave to Rina,' b. *Ze ha-sefer Se- le-Rina natan Dani.
this the-book that-to-Rina gave Dani.
(20) Adjunct Movement
a. 'eix. xasavta se-Dani tiken 'et ha- how 1(you) thought that-Dani fixed ace the- mexonit ti?
car?
'Howi did you think that Dani fixed the car t;?
b. **eix;
how Dani?
Dani?
xaSavta Se-'et ha-mexonit tiken (yeu) thought that-acc the-car fixed
In fact, we note a three-way divergence in judgements. Extraction over a subject is perfect, as expected. Extraction of an argument over a trigger is bad but adjunct extraction over a trigger is worse.
These data can explained if we adopt the structure proposed in (6) for triggered inversion. The additional CP node created in these cases of triggered inversion is not an L-marked category and hence forms a barrier to extraction. The structure in ( 6) also predicts that extraction of arguments over a trigger should constitute a milder violation than extraction of adjuncts. This is confirmed by the data in (20). Extraction over a subject, on the ether hand, is predictably perfect, since
subjects are in [Specji] and there is no additional CP node barring wh-movement directly to [SpecjC].
6. Locality of Triggers
A further prediction entailed by the adoption of (6) is that the trigger itself should not be able to move, since i t would have to cross CP which is a non L- marked category. This prediction is also borne out by the data in (2la) which shows that the dative abject cannat serve as a trigger in a lower clause and then move and topicalize, that is, adjoin to an IP of a higher clause. Notice that long topicalization is in general not ruled out in Hebrew, as shawn by (2lb).
(21) a. *le-Dan~; MoSe siper Se- kanta Ruti to-Dan1 Moshe told that-bought Ruti matana.
present
'Moshe told that Ruti bought Dani a present.' b. le-Dani. MoSe siper Se-Ruti kanta
to-Dani1 Moshe told that-Ruti bought matana.
present
'Moshe told that Ruti bought Dani a present.'
7. Wh-Movement and Inversion
Wh-words seem ta be able ta also serve as triggers for inversion, as shawn in (22) below. Although preferred by purists, inversion under a wh-phrase is optional in spoken Hebrew so that (23) is also perfectly acceptable.
(22) a. 'et ma 'axal Dani?
ace what ate Dani?
'What did Dani eat?'
b. le-mi natna Ruti matana?
to-whom gave Ruti present?
'ta whom did Ruti give a present?' (23) a. 'et ma Dani 'axal?
ace what Dani ate?
'What did Dani eat?'
b. le-mi Ruti natna matana?
to-whom Ruti gave present?
'ta whom did Ruti give a present?'
In this section we discuss sorne differences between WH triggers and non-WH triggers and account for them by arguing that inversion under a WH ward does not involve the creation of an additional CP category, but rather
movement of the inflected verb to a c0 which contains either a WH-ward or its trace.
One diference between fronted WH words and triggers is that WH phrases can licence inversion in an embedded clause and then raise up to a higher clause. This is impossible for non-WH triggers, as we saw in (21a).
Contrast the sentences in (24) with that in (21a). Note that (24a) is ambiguous: The fronted wh-ward can either be interpreted as the goal of the higher verb or the benefactor of the lower verb. Crucial ta our analysis, however, is that i t may be interpreted as the benefactor of the lower verb kanta/kana 'bought'.
(24) a. le-mi siper Dani Se- Ruti kanta matana.
to-whom told Dani that Ruti bought present i. 'to whom; did Dani tell t; that Rina bought
a present?'
11. 'ta whom; did Dani tell that Rina bought t;
a present?'
b. ['eize matana]. siper Yosi Se- MoSe which present 1 tell Yossi that-Moshe kana le-Rina t.? • 1
bought to-R1na?
'[which present]; did Yossi tell that Moshe bought Rina t;? '
The ability of wh-words fronted from a lower clause ta trigger inversion in a higher clause is not the only property which sets them apart from ether triggers. A clearly related characteristic of wh-words is that they can licence inversion cyclically, i.e., they can licence inversion in a lower clause and then move up to a higher clause and licence inversion there as well. This is illustrated in (25).
(25) a. le-mi siper Dani Se- kanta Ruti to-whom told Dani that-bought Ruti ma tana?
present?
'ta whom did Dani recount that Ruti bought a present?'
b. 'eize matana siper Yosi Se- kana MoSe which present tell Yossi that-bought Moshe le-Rina?
to-Rina?
'which present did Yosi tell that Moshe bought Rina?'
Moreover, wh-words, but not other triggers, can licence inversion in a lower clause and then proceed upwards to [Spec/C] in a higher clause. Triggers, as we
clauses. We argued that positing an extra CP layer which hasts the verb and the XP preceding it, allows us ta straighforwardly explain the interaction of V2 and WH movement in Rebrew. Finally, we argued for a distinction between inversion due to the fronting of V and an XP into an extra CP category and inversion triggered by a Camp 111arked (+WH).
References
Diesing, Molly. 1990. Verb-Second and the Subject Position in Yiddish. Natural Language and Linguistic Theory 8:41-80.
Holmberg, Anders. 1986. Word Order ans svntactic Features in the Scandinavian Languages and English. Stockholm:
Department of General Linguistics, University of Stockhollll.
Kayne, Richard and Jean-Yves Pollock. 1978. Stylistic Inversion, Successive cyclicity and Move NP in French.
Linguistic Inguiry 9: 595-621.
Platzack, Christer. 1986. COMP, INFL and Germanie Ward Order. Tapies in Scandinavian Syntax, ed. Lars Hellan and Kristi Koch Christensen, 185-234. Dordrecht:
Kluwer.
Rognvaldsson, Eirikur and Hoskuldur Thrainsson. 1990. on Icelandic Ward arder Once More. Modern Tcelandic Syntax, ed. Joan Maling and Annie Zaenen, 3-40.
(Syntax and Semantics 24.) San Diego: Academie Press.
Santorini, Beatrice. 1989. The Generalization of the Verb-Second Constraint in the History of Yiddish.
Ph.D dissertation, Philadelphia: university of Pennsylvania.
Shlonsky, Ur. 1987. Null and Displaced Subjects. Ph.D.
dissertation. Cambridge: Massachusetts Institute of Technology. Distributed by the Indiana University Linguistics Club, 1989.
Shlonsky, Ur. 1988. Complementizer-Cliticization in Hebrew and the Empty category Principle. Natural Language and Linguistic Theory 6:191-205.
Torrego, Esther. 1984. On Inversion in Spanish and Seme of its Efects. Linguistic Inguiry 15: 103-130.
Travis, Lisa. 1988. The syntax of Adverbs. McGill Working papers in Linguistics, 280-310. Montreal: Department of Linguistics, McGill University.
vikner, sten. 1990. Verb Movement and The Lic~nsing o~ NP Postions in the Germanie Languages. Ph.D d1s~ertat1on.
Geneva: Departement de langue et 11tterature anglaises, Faculte des lettres.
