RRL, LXIII, 4, p. 361–378, 2018
THE CASE OF THE ROMANIAN ETHICAL DATIVE
ALINA TIGĂU
1Abstract. This paper dwells on a number of syntactic and semantic properties of Ethical Datives (EDs) arguing that one should distinguish these expressions from other non-core, high datives with which they are usually ranged. Based on their semantics, we argue that EDs bear a [+Participant] feature which is checked by the logophoric operators inside the CP. Syntactically, EDs are argued to merge in a special Applicative projection, outside the lexical domain and higher than the (high) ApplP proposed for non-core datives. The proposed account rests on newly obtained experimental data uncovering the special status and behaviour of EDs.
Keywords: datives, applicatives, clitics.
1. INTRODUCTION
The literature on Romanian datives differentiates between core (subcategorized) (CD) and non-core datives (NCDs), also known as high datives (HDs) on account of their occupying a high applicative phrase above vP/VP (Pylkkänen 2002, 2008 a.o.). Among the latter, one distinguishes between datives of (inalienable) possession, benefactive/adversative, affected/experiental and ethical datives.
This paper aims at providing new insights into the syntactic characteristics of Romanian ethical dative clitics, which we distinguish from other types of NCDs (cf. Jouitteau and Rezac 2008 a.o.), and at paving the way towards an appropriate analysis of these elements. The proposed account rests on newly obtained experimental data uncovering the special status and behaviour of EDs. In particular, there exist a number of properties setting EDs aside from other HDs: EDs are ‘non-actantial’ datives, since they are not part of the valency of the verb but have an expressive function, grounding ‘the event structure in relation to the speech participants’ (Delbecque and Lamiroy 1996). As such, they do not affect the truth conditions of the sentence in which they occur, unlike the other HDs, but merely invoke the addressee or the speaker as a witness or as a vaguely affected party. Furthermore, EDs allow multiple clitic clusters, which is not possible with other instances of HDs. Note, moreover, that the multiple clusters of EDs may only contain a 1st and a 2nd person pronoun. When a third person dative clitic pronoun appears in the cluster, it may be interpreted as an argumentative, a possessive, beneficiary or affected dative.
Lastly, unlike other NCDs, EDs do not allow a full corresponding DP pronominal.
1 University of Bucharest and University of Köln, [email protected]. The research for this paper has been funded by the Alexander van Hunboldt Foundation, Germany.
The properties uncovered above urge one to draw several conclusions: a) co- occurrence with other HDs points to the existence of distinct hosting projections. b) the fact that EDs anchor the Speaker/Addresse leads us to conclude that EDs merge within a special Appl projection wherefrom they may be bound by abstract Speaker/Addresse operators from within CP (Sigurðsson 2012) and thus be interpreted as denoting the Speaker/Addresse. In line with Baker (2008), we propose person to be a derivative notion, the result of operator-variable agreement. Furthermore, in line with Michelioudakis (2016) we claim that ED clitics are specified as [+Participant,±author] and that these specifications amount to the interpretable, lexically valued features probed for by the Speaker/Addresse operators in CP. Moreover, given their featural make up, EDs merge into an ApplP carrying an uninterpretable [+Participant] feature, situated below T but above the HighApplP hosting other NCDs.
The analysis accounts for the co-occurrence between EDs with other HDs, as well as for their anchoring the Speaker/Addressee to the event denoted by the verb.
The paper has the following structure: section 2 dwells on the basic tenets regarding the syntax of datives in general distinguishing between Low vs. High/Core vs. Non-core datives and ranging EDs with the former class; in section 3 we discuss the main syntactic and semantic properties of EDs; section 4 is devoted to the two experiments proper, going through their motivation, their design, the results obtained and the discussion of these findings; section 5 contains a tentative syntactic account of Romanian ditransitives; in section 6 we find the conclusions to this article.
2. BACKGROUND
The literature on datives distinguishes between Core datives (CDs) and Non-core datives (NCDs): the former are c-selected arguments of the verb i.e., part of the verb’s argument structure. Such is the case of dative DPs within ditransitive configurations – in (1) below, the inflectional dative Mary functions as an internal argument of the ditransitive verb a da (give):
(1) (I)-am dat Mariei o carte.
(Her)-have.I given Mary.dat a book.
‘I gave Mary a book.’
Romanian also possesses a class of unaccusative configurations selecting dative DPs as obligatory constituents with quirky subject characteristics. Consider (2) below2:
(2) a. La prăjituri aş mai putea renunţa, dar ciocolata to cakes would more could give up but chocolate
îmi place la nebunie.
me.dat like to madness
‘I could give up cakes but I am extremely fond of chocolate.’
2 As shown in Cornilescu (2016), these configurations rest on several classes of light unaccusatives such as the verb a fi ‘to be’, anticausative reflexive verbs (a se face ‘become’) or change of location verbs (a veni ‘come’, cădea ‘fall’).
b. Îmi este dor de Maria.
Me.dat is yearning of Mary ‘I miss Mary.’
Unlike, these dative DPs, NCDs are not c-selected arguments of the verb. This is the case of datives of (inalienable) possession, benefactive/adversative, affected/experiential datives or ethical datives a.o.:
(3) a. I-am admirat casa cea nouă.
him.dat-have.I admired house the new
‘I admired his new house.’ possessive dative b. Paul i-a furat toate cireşele din grădină.
Paul him.dat-has stolen all cherries from garden Lit. ‘Paul stole all the cherries in the garden on him.’
benefactive/adversative c. Nu ştiu ce să mă fac cu el: nu îmi
not know what să me do with him not me.dat
mai doarme bine noaptea.
more sleep well night.the
Lit. ‘I don´t know what to do with him: he no longer sleeps well on me at
night.’ affected/experiental dative
d. Şi când mi-l luă zmeul odată pe Greucean and when me.dat-him.acc grabbed ogre.the once pe Greucean
de mijloc de-i pârâiră toate oasele, crezu by waist so that-him.dat cracked all bones thought voinicul că i-a sosit ceasul în clipa aceea.
lad.the that him.dat-has arrived time in moment that
‘And when the ogre grabbed Greauceanu by the waist so that all his bones cracked, the lad thought his time to die had come.’
ethical dative Pylkkänen (2002) argues in favour of another classification dividing datives into High and Low. Drawing on Marantz3 (1984, 1993), Pylkkänen (2002, 2008) distinguishes between core arguments of the verb (direct objects, subject) and non-core arguments. While languages tend to function on a par with respect to the former, they differ with respect to the range of non-core arguments which they may allow. In Bantu, languages, for instance, non-core arguments are introduced into the argument structure of the verb by means of verbal functional heads which one may identify as affixes present in the verb´s morphology. Such morphemes bear the name of applicative arguments and they are said to introduce applied arguments. Pylkkänen proposes an analysis of dative DP as applied arguments introduced by means of two Applicative projections:
3 Marantz (1984, 1993) argues that verbs may only have one internal object i.e., the accusative one and posits that the subject (the external argument) is introduced by means of a Voice P and is not a true argument of the verb. This idea has become standard in the Minimalist Program, where the external argument is introduced by a small v.
Low applicatives merge inside the VP and introduce arguments which bear no semantic relation to the verb but which point to a transfer-of-possesson relation (Pylkkänen 2008: 14). This is the case of dative objects in ditransitive configurations which are associated with the idea of a transfer of possession between the direct object and the indirect (=applied) argument. In (4) we may see this at work: (4a) shows that the indirect object John is introduced into the argument structure by the low Appl and (4c) shows the semantics of this argument, which is not related to the event quantification by means of a thematic relation e.g., Goal but only bears a direct relation with the direct object DP to-the- possession(x, y):
(4) a. Mary baked John a pie.
b. VoiceP
3
DP Voice’
Mary 3
Voice vP
3
bake ApplP
3
John Appl’
3
Appl a pie
c. Semantics for low applicatives (recipient applicative)
λxλy, λf <e <s,t>>.λe.f(e,x) & theme(e,x) & to-the-possession(x, y)
High applicatives merge above the VP/vP and point to a thematic relation holding between an individual and the event denoted by the VP. Pylkkäenen (2008) exemplifies this type of applicative by means of a construction in Chaga where a benefactive participant is added to an unergative verb:
(5) a. N-ä-ï-lyì-í-à m-kà k-élyá.
FOC-1SG-PRES-eat-APPL-FV 1-wife 7-food
‘He is eating food for his wife.’ Pylkkänen (2008): 11 (2a) b. VoiceP
3
DP Voice’
He 3
Voice ApplBen
3 wife A Appl´
3 Appl A vP
3
bake food
Unlike, Benefactives in English, which merge low and only bear a relation with the direct object DP, the Benefactive in Chaga is related with the event described by the VP through the high applicative head: the DP wife thus stands in a benefactive relation with the event of eating but entertains no relation to the object of eating i.e., the food:
(6) [λx.λe [Beneficiary (e,x)].
Note that the difference regarding their semantics plays an important part in the derivation of these Applicative configurations: High applicatives may merge relatively high since they only necessitate access to the event variable supplied by the verb, while low applicatives have to merge at an earlier stage so as to surface close to the direct object to which they need to be related.
Whereas High applicatives have remained unchallenged up to date, Low Applicatives have been argued to present a number of problems which cast doubt upon their legitimacy. Thus, Larson (2010) shows that the account proposed for low applicatives leads to undesirable inferences, given that the indirect argument only bears a relation to the direct object and no relation to the event introduced by the verb phrase. More specifically, the semantic account proposed by Pylkkänen for low Applicatives allows incorrect references such as John baked the cake and Mark brought Alice the cake → John baked Alice the cake, which the regular neo-Davidsonian semantics correctly discards.
Larson (2010) thus shows that in the so-called low applicative configurations both the direct object and the indirect object are part of the thematic structure of the verb. The applied argument is introduced by the lexical verb itself, composing inside the VP along the lines of Larson (1988, 2014).
Georgala, Paul and Whitman (2008) identify a morphological problem with low applicatives: while they show that the Applicative morpheme may only occupy a suffixal position with respect to the verb, the point out that the two types of applicatives proposed by Pylkkänen enable this morpheme to surface both as prefixes (low Applicatives) and as suffixes (high Applicatives): if according to the mirror image principle proposed in Baker (1988), head movement presupposes raising and adjunction to the left, then with high Applicatives, the applicative morpheme will surface as a suffix to the verb moves to Appl.
On the other hand, with low Applicatives, the applicative morpheme raising onto the verb ends up as a prefix, by way of left-adjunction. This latter prediction does not seem to have coverage cross-linguistically, as all applicative morphemes are shown to surface as suffixes.
In view of the problems that low applicatives seem to face both semantically as well as morphologically, the notion of a low functional projection introducing non-core arguments into the argument structure of the verb was given up in favour of a more refined typology of ´high´ applicatives (see Harada and Larson, 2009, Georgala 2011 a.o. for a distinction between thematic and expletive applicatives)4 i.e., Applicative projections merging above the vP/VP.
4 According to Georgala (2011) a.o., thematic applicatives are supplied with θ-features which are valued by the non-core arguments they introduce. Thematic applicatives also case-license the non- core arguments. Expletive (raising) applicatives, on the other hand, only case-license an argument that is introduced by the lexical verb. The lexical verb also assigns a θ-role to this argument. The expletive applicative head thus only case-licenses the indirect object (introduced by the lexical verb), attracting it to its specifier.
In the following sections we will maintain the distinction between high and low datives without retaining the idea that there is a low functional projection inside the VP introducing datives. Rather, we tend to favour Larson (2014) in allowing core datives to be part of the verb´s theta-grid, introduced by the lexical verb. Note also that all non-core datives will be analysed as high applicatives in Pylkkänen´s terms and express relations between individuals and (sub)events. As shown, non-core datives are conventionally considered to merge in the same position irrespective of their diverse interpretation as there is no more than one (High)ApplP in the functional domanin of the verb. In the following sections we will, however, endeavor to distinguish among the representatives of the High datives class by setting ethical datives appart and arguing in favour of an even higher Applicative projection, in line with the one proposed by Boneh and Nash (2010).
3. THE ETHICAL DATIVE – RELEVANT PROPERTIES
EDs are weak/clitic pronominal forms whose referents are not participants in the event denoted by the verb. They necessarily refer to a discourse participant i.e., either the speaker or the hearer. Michelioudakis (2016) also points out that EDs may also point to a reported speaker in certain languages e.g., Greek, but this is not the case in Romanian where EDs are only restricted to the 1st and 2nd person use.
The meaning that EDs evince is entirely non at issue and has to do with some sort of relation holding between a discourse participant (either the logophoric agent or the logophoric patient) and the event introduced by the lexical verb, of which this discourse participant does not form a part. In both examples (6) and (7), the contribution of the ED instances could not be captured by the translation, but one understands the narrator (and the reader in (7)) as somehow taking sides with the good character and positively assessing his/her deeds.
(6) Fata, însă, nu se lăsă înduplecată de vorbele girl.the however not refl. let convinced by words.the
mieroase ale bătrânei şi mi-o luă
sugary of old woman and me.dat-her.acc chased la goană din faţa porţii, ba încă mai puse şi câinii pe ea.
away from gate.the even more set.she and dogs.the on her
‘The girl did not allow herself to be convinced by the sugary words of the old lady, even more she chased her away and set the dogs on her.’
(7) Dar nici Greuceanul nu se lăsă mai prejos: unde but neither Greuceanu not refl. gave up: where
nu se încordă odată şi când mi ţi-l
not refl. strained once and when me.dat you.dat-him.acc
apucă pe zmeu de umeri şi dădu cu el de pământ, grabbed pe ogre by shoulders and threw with him to ground praf şi pulbere se făcu netrebnicul.
dust and ashes refl. became wicked.the
‘But Greuceanu rose up to the challenge grabbed the ogre and threw him tot he ground so that the wocked one perished.’
Some cross-linguistic studies on EDs have tried to capture the contribution of these expressions by adding an explanatory note to the proposed translation. In all the examples below, the discourse participant takes on an evaluative stance5 with respect to the event described. Note also the diversity of attitudes which may arise as the referent of the ED may show delight, disappointment, surprise, disapproval etc.
(8) Que buena se nos puso Maria!
how nice refl. 1.pl.dat.cl became Maria
‘How nice Maria became, to our delight.’ Spanish
(9) I Maria ine kali kopela, ala mu psifizi LAOS The Mary is nice girl, but 1sg.dat.cl votes LAOS
‘Mary is a nice girl but, to my disappointment, she votes for the Nationalist Party.’
Greek, Kapogiannni and Michelioudakis (2013: 3)
(10) Da hat er mir ihm etwas zugeflüstert und…
then has he me.dat him.dat something to-whispered and ‘Then to my surprise, he whispered something to him and...’
German, Draye (1996 :184) (11) Je te lui ai donné un de ces gifles!
I you.dat him.dat have.I given one of these smacks ‘(I´m telling you) I smacked him well!’
French, Boneh and Nash (2010: 15)
(12) Ez meg mi-t csnál itt nekem?
this and what.acc does here me.dat ‘And what the hell is this one doing here?’
Hungarian, Rákosi (2008: 413)
EDs thus have an expressive function, grounding ´the event structure in relation to the speech participants´ (Delbecque and Lamiroy 1996: 24).
Furthermore, the semantic contribution of EDs is entirely non-truth functional, unlike that of other dative DPs. In (13) below, (a) and (b) may not both describe the same state of affairs i.e., in order for the two variants to be true, they need to be assessed against different models of reality. Similarly, the variants in (14) where a non-core Benefactive dative is employed may not both hold as true at the same time. Example (14), on the other hand, differentiates itself from (12) and (13) in that the addition of a further ED instance ţi in (14b) does not trigger a change in the truth conditions. In fact, both (14a) and (14b) may be uttered felicitously to describe the same state of affairs without contradicting each other.
Furthermore, the negation of one of the two variants entails the falsity of the other.
(13) a. Mi-a dat cartea.
me.dat-has.he given book.the ‘He has given me the book.’
5 As Michelioudakis and Kapogianni (2013) point out, in other languages such non-truth functional forms may also encode other types of relation holding between a participant and an event or a proposition e.g., evidentiality Speas and Tenny (2003).
b. Ţi-a dat cartea.
you.dat-has.he given book.the
‘He has given you the book.’ (Goal) (14) a. Îmi spală o cămaşă.
me.dat whashes.he a shirt ‘(S)he is washing me a shirt.’
b. Îţi spală o cămaşă.
you.dat whashes.he a shirt
‘(S)he is washing you a shirt’ (Beneficiary) (15) a. Mi-l luă de braţ şi îl conduse la uşă.
me.dat-him.acc took by arm and him.acc led to door ‘He took him by the arm and led him to the door.’
b. Mi ţi-l luă de braţ şi îl conduse la uşă.
me.dat you.dat-him.acc took by arm and him.acc led to door
‘He took him by the arm and led him to the door.’ (ED) Note that their behaviour as non-truth functional expressions distinguishes EDs from other types of datives in Romanian which always seem to contribute to the truth conditions of a proposition. Truth functionality seems to be the criterion separating ED from the rest of dative expressions.
Another property which distinguishes EDs from other datives is that they are exclusively realized as clitics and do not allow a full corresponding DP. Moreover, no co-indexed DP associate is possible in an A-position (see also Cuervo 2003, Roberge and Troberg 2009).
This is, however, an option that both core (17) and other non-core datives (16) have:
(16) Părinţii i-au construit (fetei/ei) o casă ca-n poveşti.
parents.the her.dat-have.they build girl.dat/her a house as in stories.
‘Her parents have build Mary/her a house as one can only see in fairy tales.’
(Beneficiary) (17) De când stă la oraş, Maria le telefonează (părinţilor)
ever since lives in town Mary them.dat phones parents.dat
în fiecare zi.
in every day.
‘Ever since she moved to town, Mary phones her parents in the countryside every
day.’ (Goal)
(15) Pe unde mi-ai fost (*mie)?
‘Where have you been this time (to my annoyance)?’ (ED) Furthermore, EDs may not undergo A´-movement: in (16) both the core dative (Goal) as well as the non-core datives (Beneficiaries, Maleficiaries) may undergo wh- movement, while this is not possible for the ED in (17):
(16) Cui i-ai arătat/distrus/zugrăvit casa?
who.dat him.dat-have.you shown/destroyed/painted house.the ‘Who did you show/destroy/paint his house to/for.’
(17) *Cui i te-ai făcut aşa frumoasă?
who.dat him.dat you.acc-have become so beautiful? (ED)
Another interesting property concerning EDs has to do with the fact that they allow multiple clitic clusters, with a 1st person ED always preceding a 2nd person ED. This is not possible with other datives which may never enter such clusters of same-type datives:
(18) Iedul cel mic se ascunse într-un cotlon aşa că lupul goat.the the youngest refl. hid in a recess so that wolf.the
nu-l găsi. Pe cei doi iezişori mai mari, not-him.acc found pe the two kid goats elder,
însă, care îi deschiseseră uşa, mi ţi-i
however who him.dat had opened door.the me.dat you.dat-them.acc mâncă cu tot cu blană.
ate with everything and fur
One final property which seems to distinguish EDs from other dative counterparts has to do with their co-occurrence possibilities: while EDs seem to be able to co-occur with any other type of dative (including with another ED as shown above), other high (and low) datives seem to be more restricted in this respect.
(19) Să nu mă faci că, dacă îmi pun mintea cu tine, să.subj. not me.acc challenge because if me.dat set mind with you
mi te trimit una-două plocon mumă-tii.
me.ED you.acc send quickly gift mother-your.dat
‘Don´t upset me or else I´ll immediately send you back to your mother.’
(20) Şi când se dezlănţuiră ghiorlanii odată şi mi-i and when refl. went wild children.the once and me.dat-her.dat întoarseră toată casa mătuşii cu susul în jos, nu-i
turned entire house.the aunt.dat upside down not-her.dat mai trebui femeii nici copii şi nici nepoţi.
more need woman.dat neither children and nor grandchildren
‘And when the children suddenly ran wild and turned the woman´s house upside down on her, she stopped wanting children or grandchildren.’
(21) *Or să mă omoare ai mei: cred că tocmai will să.subj. me.acc kill mine think.I that just li le-am distrus maşina.
them.dat them.dat-have.I destroyed car.the
‘My parents will kill me: I think I have just destroyed their car (on them).’
(19) shows an ED felicitously co-occurring with a Goal DP. This state of affairs would be expected, given that core datives such as Goal DPs in ditransitive configurations are merged low within the VP, leaving the (high) ApplP available for the ED to occupy.
Example (20) is more interesting in this respect as it shows an ED co-occurring with a possessive dative. As already extended upon above, these datives have both been classified as HDs occupying the (high) ApplP. The expectation in this case would be for ungrammaticality to arise, given that there is only one projection available for the two dative DPs to occupy. Contrary to expectations such examples have been rated as acceptable by native speakers of Romanian. Finally, example (21) shows that when no ED is involved, two other HDs may not co-occur.
The properties discussed in this section seem to set EDs apart from other types of datives in general and from the class of high datives in particular. One important aspect which needs to be settled has to do with their unexpected syntactic behaviour regarding co- occurrence with other HDs, which seems to justify the idea that ED have a different, higher merge position than the (high) ApplP which has been posited so far. This difference of behaviour distinguishing between EDs and other HD along these lines constitutes the prime motivation for our experimental undertaking which will be extended upon in the next section.
4. THE EXPERIMENT(S)
This section extends upon a two-stage experiment we unfolded with Romanian datives in an attempt to capture the exact co-occurrence possibilities with these DPs. We initially started from the assumption that datives should be grouped into two classes i.e., that of low datives (including Goal DPs in ditransitives) and that of high datives (comprised of possessive datives, beneficiaries, maleficiaries, ethical datives). The initial results pointed out that a further split should be operated within the latter group, setting EDs apart from other high datives on account of their different distributional characteristics. A second experimental round (focusing only on high datives) confirmed that such a split is desirable.
In the following subsections we consider each experiment in part.
4.1. Experiment 1 – high and low datives
4.1.1. Design
The first experiment focused on the co-occurrence possibilities of various types of datives as shown below:
(22) a. Ethical datives and Core datives (Goal DPs)
b. Ethical datives and Non-core datives (only possessive datives) c. Non-core datives (possessives) and core datives (Goal DPs) d. Two Non-core datives (benefactive/possessive)
We also checked acceptability differences depending on the possibility of using dative clitics vs full DPs to express the respective dative as well as clitic doubling of a dative full DP. As shown above, ED may only surface a clitics and never allow an associate DP (23); the other non-core datives (possessives) always surface as clitics but may allow a full associate DP (24). Finally, core datives may be used either as clitics or as full DPs and the option in which the full DP Goal is clitic doubled is also available (25):
(23) În loc să se ocupe de copil aşa cum promisese, mi-l încredinţă instead of taking care of child as how had promised me.dat-him.acc entrusted (*mie) unei bone fără suflet care îl ţinu toată ziua închis
me.dat one.dat nanny without soul who him.acc kept all day locked într-o cameră
in a room
‘Instead of taking care of the child as he had promised the judge, he entrusted him to a heartless nanny who kept him locked in a room.’
(24) cum ei nu puteau sa iasă din casă din pricina bolii, le-am how they not could to get out of house because of illness them.dat-have.I
dus eu manuscrisul (lor) la editură.
sent I manuscript.the (their) at publishing house.
‘As they could not get out of the house due to their illness, I took their manuscript to the publishing house.’
(25) (I-)am trimis (Mariei) o carte poştală.
her.dat-have.I sent Mary.dat a postcard ‘I sent Mary a postcard.’
We thus ended up having 9 possible co-occurrence patterns in need of checking:
1. ED & CD (only full DP) 2. ED & CD (clitic + full DP) 3. ED & CD (only clitic) 4. ED & NCD (cl + full DP) 5. ED & NCD (only clitic) 6. NCD & CD (only full DP) 7. NCD & CD (cl+full DP) 8. NCD & CD (only clitic) 9. NCD1 & NCD2
Due to experimental constraints (the length of the questionnaire, the number of conditions a.o.), we only selected the 6 patterns for the actual investigation as shown below:
1. ED & CD (only full DP) 2. ED & CD (clitic + full DP) 3. ED & NCD (cl + full DP) 4. NCD & CD (only full DP) 5. NCD & CD (cl+full DP) 6. NCD1 & NCD2
For each of these conditions we constructed 6 sentences such that 36 sentences were thus obtained.
The items were then distributed into 6 different questionnaire using the Latin square method for an even distribution. To each questionnaire thus formed, 24 fillers were added, grouped into 8 expectedly inacceptable items, 8 completely acceptable items and 8 average items with respect to acceptability (the fillers were separately checked for acceptability in a smaller, informal experiment). Each questionnaire thus ended up having a number of 30 items.
The questionnaires were formatted as google online forms in such a way that the potential respondent could only access one item at a time, without having the possibility of going back of forth. Each questionnaire was further assessed by at least 10 native speakers of Romanian such that more than 50 people took part in the initial experiment.
The results thus obtained were then verified and outliers were removed i.e., questionnaires in which more than 7 fillers had been wrongly assessed. Remaining questionnaires entered statistical analysis.
4.1.2. Results
The first experiment uncovered a number of interesting results concerning the co- occurrence possibilities of datives.
Firstly, clitic doubling of a core dative (Goal DP) seems to influence its co-occurring potential: while a non-doubled CD may successfully co-occur with a high dative expressed by means of a clitic (be it an ED or another NCD), a clitic doubled CD did not fare as well.
Compare the standardized results in Graph 1: ED & CD (no cl) fare much better than ED &
CD (+cl), similarly, NCD & CD (no cl) was found to be more acceptable than NCD & CD (+cl) instances).
Secondly, ED & NCD instances were assessed as acceptable, while situations where two other NCDs i.e., a benefactive and a possessive co-occur were found to be quite unacceptable. As it seems, while EDs may co-occur with NCDs, other NCDs may not co-occur with each other. Incidentally, this points to another difference between EDs on the one hand and the other NCDs on the other: while several instances of EDs may co-occur, instances of NCD with identical status may not.
Graph 1: co-occurrence of high and low datives.
Another, somehow unexpected result concerns the co-occurence of both EDs and other NCDs (possessive datives) with a clitic doubled CD. As may be seen from Graph 2 below, the difference in acceptability between the two pattern seems to be negligeable.
Under the announced hypothesis that EDs may actually occupy an even higher projection than the one occupied by the NCDs such a result is not expected. What we would expect, would be for ED & CD (+cl) patterns to be more acceptable than the NCD & CD (+cl) ones. Consider a possible explanation for this expectation: the dative clitic doubling a CD would occupy the ApplP. When a NCD (possessive) and a clitic doubled CD co-occur, both the Goal clitic and the possessive would occupy the same projection, hence the low acceptability of these lexicalisations built on these patterns. However, in the ED + CD (+cl) patterns, a higher acceptability rating is expected if EDs are taken to occupy a projection other than the (high) ApplP. Contrary to expectations, the results for this pattern are comparable to those obtained for the NCD & CD (+cl) one. For the moment we don´t have an explanation for these fact and leave the matter for further research.
Graph 2: Doubling vs. non-doubling.
One important finding relevant to the syntax of dative DPs that the first experiment contributes has to do with the merge postion of core datives. As it seems, dative DPs in ditransitives configurations are merged low, confirming the analyses in Cornilescu et al.
(2017) for Romanian, Harada and Larson (2009) for Japanese and Larson (1988, 2014) a.o.
4.2. Experiment 2 – high and higher datives 4.2.1. Design
The second experiment only focused on the interraction of EDs with other NCDs (again restricted to possessive datives). Two patterns were thus selected to undergo assessment:
1. ED & NCD (possessive dative expressed by means of a clitic)
2. ED & NCD (possessive dative expressed by means of a clitic doubled DP)
The examples under (26) show two lexicalisations built on these patterns: in (26) the ED co-occurs with a possessive dative expressed by means of a pronominal clitic, while in (27) the possessive surfaces as a clitic doubled full DP.
(26) Vulpea aşteptă până când ursul adormi şi când îl fox.the waited until bear.the fell asleep and when him.acc
auzi sforăind, ieşi tiptil din ascunzătoare heard snoring got out stealthily from hiding place şi mi-i şterpeli toţi peştişorii prinşi cu atâta trudă and me.ED-him.dat stole all fish caught with such effort de rămase jupânul cu buza umflată şi cu burta goală.
so that remained master.the with lip.def swollen and with belly empty
‘The fox waited until the bear fell asleep and when she heard him snoring she stealthily went out of hiding and stole the fish he had worked so hard to catch and left him with no food.’
(27) Vulpea aşteptă până când ursul adormi şi când îl fox.the waited until bear.the fell asleep and when him.acc
auzi sforăind, ieşi tiptil din ascunzătoare heard snoring got out stealthily from hiding place
şi mi-i şterpeli jupânului toţi peştişorii prinşi cu atâta trudă and me.ED-him.dat stole master.the all fish caught with such effort de rămase acesta cu buza umflată şi cu burta goală.
so that remained this with lip swollen and with belly.def empty
‘The fox waited until the bear fell asleep and when she heard him snoring she stealthily went out of hiding and stole the fish he had worked so hard to catch and left him with no food.’
For each of the two conditions we built 8 sentences, which we included in a single questionnaire. To the 16 lexicalisations thus obtained 24 fillers were added grouped into 8 expectedly inacceptable items, 8 completely acceptable items and 8 average items with respect to acceptability. The fillers had been previously checked for acceptability in a smaller, informal experiment. Each questionnaire thus ended up having a number of 40 items.
Just like in the first experiment, the questionnaire was formatted as google online forms in such a way that the respondents could only access one item at a time, without having the possibility of going back of forth. The questionnaire was further assessed by at least 80 native speakers of Romanian.
The results thus obtained were then verified and outliers were removed (questionnaires in which more than 6 fillers had been wrongly assessed – in this particular case 14 outliers were discarded). Remaining questionnaires entered statistical analysis.
4.2.2. Results
As may be seen from Graph 3, instances where EDs co-occur with other NCDs (possessive datives) were found acceptable irrespective of whether the NCDs in question were expressed by means of a dative clitic or a clitic doubled full DP. With the exception of lexicalisation nr. 6 where the ED+NCD(only cl) condition did not fare so well, all lexicalisations were assessed as acceptable on a 7 step scale closely reaching and surpassing an acceptability threshold of 4, as shown in the table of mean values below:
Graph 3: ethical datives and other non-core datives.
ED+NCD(cl+full DP) ED+NCD (only CL)
4,545454545 3,96969697
4,393939394 4,060606061
4,575757576 4,060606061
4,454545455 4,696969697
4,96969697 4,909090909
4,151515152 3,545454545
4,121212121 4,666666667
4,424242424 4,060606061
The results uncovered by the two experiments point that ED pattern differently from other NCDs regarding their co-occurrence possibilities, not only allowing co-occurrence with CDs but also enabling other NCDs to co-occur with them. A suitable analysis of these expressions should account for these peculiarities. This is, indeed, the aim of the next section which endeavors to propose a tentative analysis of EDs in the light of the characteristics uncovered so far.
5. TOWARDS AN ANALYSIS OF EDs
A suitable analysis of EDs should account for the following: a) the possibility of EDs to co-occur both with NCDs and with CDs; b) their denoting a discourse participant and expressing an evaluative relation between this discourse participant and the event to which (s)he does not form part of; c) clitic clustering and the frozen word order 1stp >2ndp.
Person restrictions. As already pointed out above, there are important restrictions on the ED clitic sequence with respect to person in that only 1st and 2nd person EDs are allowed. Furtehrmore, within a cluster of two EDs, the 1st person instance always precedes the 2nd person one.
Another important observation has to do with the fact that EDs do not refer to event participants but to discourse participants, who are not part of the event contributed by the lexical verb. As such, EDs exhibit a very low “degree of integration” in the lexical event.
Along this dimension, one may view non-core datives as varying from fully integrated in the event structure of the lexical event (as is the Case with the Benefactive/Possessors), to non-integrated in the event structure of the lexical event (ED). This immediately raises the question of how the non-integrated event are semantically licensed. Since such Datives are often associated with specific illocutionary and perlocutionary flavours, it is likely that they are semantically related to the speech events. We thus posit that ED be reduced to the logophoric agent and the logophoric patient. As such, we take EDs to be marked for [+Participant], a feature which is typical for the logophoric agent and the logophoric patient and which gets checked in the C domain by some relevant functional head6.
6 Unlike EDs, other datives would be marked for [+iPerson]
The analysis would be in line with Sigurðson’s (2016) view that:
(28) Grammar computes or interprets (propositional event features) in relation to speech event features.
(29) [CP ... Force... Top ... ΛA ... ΛP ... Fin ... [IP ... Pn ... Nr ... T ... [vP v ... NP√Pn ...]]
The first and second person clitic are in fact participants in the speech event. Their function is that of evaluating the lexical event features, expressing interest in the truth of the propositional event, and in particular some emotional reaction: surprise, interest, etc.
Dative clitics are thus affected experiencers. EDs will thus anchor the Speaker/Addressee operators ΛA ... ΛP within the C domain.
Co-occurrence. The co-occurrence of ethical datives and other non-core high datives seems to suggests that Romanian allows two applicative phrases merging in different positions in the clause. We thus tentatively posit an ApplEDP outside of the lexical domain i.e., above the vP (see Boneh and Nash 2010 for French, Michelioudakis 2016 for Greek).
ΛAP
3
ΛA ΛPP 3
ΛP
...
ApplP 3 DP Eth.Dat Appl’
[+Participant] 3 Appl vP 3
DP v’
3
v ApplP
3 DP poss/benef Appl’
3
Appl VP
3 V DP
The Speaker/Addressee operators from within the C domain function as probes checking [+Participant] on the corresponding ED under long distance Agree, without concomitant movement. Following Baker (2008), we take Person to be a derivative notion, obtaining as a result of operator-variable agreement: all pronouns are in fact variables potentially bound by speaker/hearer operators in CP; the apparent person of the pronoun is
the result of the interaction of the nature of the operator (speaker/addressee) and the interpretable deictic features ([±participant], [±author], [hearer]) that the pronoun may carry (cf. Michelioudakis 2016).
As to the impossibility of doubling, one should recall Kayne’s idea the clitics (case) license their double (as already seen for Benefactives above). The proposal that we adopt (form Jouitteau and Rezac, 2008) is that Appl which introduces Ethical Datives does not have phi features and that such meta-clitis have some default inherent case.
6. CONCLUSIONS
This paper discusses a number of syntactic and semantic properties of EDs arguing that one should distinguish these expressions from other non-core, high datives with which they are usually ranged.
Unlike other HDs, EDs do not refer to event participants and as such they are not integrated in the event structure of the lexical event. Instead, they refer to discourse participants and express an evaluative attitude towards an event which the respective discourse participant does not form a part of. We formalized this semantic import by positing a [+Participant] feature on EDs, which is checked by the logophoric agent and the logophoric patient operators inside the CP.
The co-occurrence facts with HDs, verified experimentally, prompted us to posit a novel ApplP above the (high) Appl projection already proposed in the literature (Pylkkänen 2002, 2008).
BIBLIOGRAPHY
Baker, M., 2008, The syntax of agreement and concord, Cambridge, Cambridge University Press.
Baker, M., 1988, Incorporation: A theory of grammatical function changing, Chicago, University of Chicago Press.
Boneh, N., L. Nash, 2010, “A higher applicative: evidence from French”, Proceedings of IATL 25, http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.597.2062andrep=rep1&type=pdf Cornilescu, A., Dinu A., A. Tigău, 2017, “Romanian Dative Configurations’, Revue roumaine de
linguistique, LXII, 2, 179–206.
Cornilescu, A, 2016, “On the syntax of Datives in Unaccusative Configurations’, ms., University of Bucharest.
Cuervo, M. C, 2003, Datives at large, Doctoral dissertation, MIT.
Delbecque, N., B. Lamiroy, 1996, “Towards a typology of the Spanish Dative”, in: W. Van Belle, W. Van Langendonck (eds), The Dative, Vol. 1, Amsterdam/Philadelphia, John Benjamins, 71–117.
Draye, L., 1996, “The German Dative”, in: W. Van Belle, W. Van Langendonck (eds), The Dative, Vol. 1, Amsterdam/Philadelphia, John Benjamins, 155–215.
Georgala, E., 2011, Applicatives in their Structural and Thematic Function: A Minimalist Account of Multitransitivity, PhD Dissertation, Cornell University.
Georgala, E, Waltraud P., J. Whitman, 2008, “Expletive and Thematic Applicatives”, in: C.B. Chang, H.J. Haynie (eds), WCCFL 26, Somerville, Mass., Cascadilla Proceedings Project, 181–189.
Guţu Romalo, V. (ed.), 2005, Gramatica limbii române, Bucureşti, Editura Academiei Române.
Harada, N., R. Larson, 2009, “Datives in Japanese”, in: R. Shibagaki, R. Vermeulen (eds), Proceedings of the Workshop on Altaic Formal Linguistics II. MIT Working Papers in Linguistics, 54, 109–120.
Jouitteau, M., M. Rezac, 2008, “The French Ethical Dative. 13 syntactic tests”, Bucharest Working Papers in Linguistics, IX, 1, 97–108.
Larson, R., 2014, On shell structure, Routledge, London.
Larson, R., 2010, “On Pylkkänen’s semantics for low applicatives”, Linguistic Inquiry, 41, 701–704.
Larson, R., 1988, “On the double object construction”, Linguistic Inquiry, 19, 335–391.
Marantz, A., 1984, On the Nature of Grammatical Relations, Cambridge, MA, MIT Press.
Marantz, A., 1993, “Implications of asymmetries in double object constructions”, in: S.A. Mchombo (ed), Theoretical Aspects of Bantu Grammar, Stanford, CA, CSLI, 113–150.
Michelioudakis, D., 2016, Dative arguments and abstract case in Greek, PhD Dissertation, University of Cambridge.
Michelioudakis, D., E. Kapogianni, 2013, “Ethical datives: a puzzle for syntax, semantics, pragmatics and their interfaces”, in: R. Folli, C. Sevdali and R. Truswell, Syntax and its Limits, Oxford, Oxford University Press.
Pylkkänen, L., 2002, Introducing Arguments, Doctoral Dissertation, MIT.
Pylkkänen, L., 2008, Introducing arguments, MIT Press, Cambridge, MA.
Rákosi, G., 2008, “Some remarks on Hungarian ethical datives”, in: J. Andor, B. Hollo, Laczkó and P. Pelyvás (eds), When grammar minds language and literature, Festchrift for Prof. Béla Korponay, Debrecen, University of Debrecen, 413–422.
Roberge, Y., M. Troberg, 2009, “The high applicative syntax of the dativus commondi/incommodi in Romance”, Probus, 21, 249–289.
Sigurðsson, H. Á., J. M. Brandeis, 2012, “Silent Heads”, in: L. Bruge, A. Cardinaletti, G. Giusti, N. Munaro and C. Poletto (eds) Functional heads: The Cartography of Syntactic Structures, vol 7, Oxford, Oxford University Press, 368–379.
Speas, P., Tenny, C., 2003, “Configurational Properties of Point of View Roles”, in: A. DiSciullo, (ed.), Asymmetry in Grammar, Amsterdam, John Benjamins, 315–344.
