on the semantics and syntax of the latin ‘double dative’ construction

25
Das Nomen im Indogermanischen

Upload: uit

Post on 04-May-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Das Nomen im Indogermanischen

Das Nomen im Indogermanischen

Morphologie, Substantiv versus Adjektiv, Kollektivum Akten der Arbeitstagung der Indogermanischen Gesellschaft

vom 14. bis 16. September 2011 in Erlangen.

Herausgegeben von Norbert Oettinger und Thomas Steer

Das Nomen im Indogermanischen

Morphologie, Substantiv versus Adjektiv, Kollektivum Akten der Arbeitstagung der Indogermanischen Gesellschaft

vom 14. bis 16. September 2011 in Erlangen.

Herausgegeben von Norbert Oettinger und Thomas Steer

Wiesbaden 2014Reichert Verlag

Bibliografische Information der Deutschen NationalbibliothekDie Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind

im Internet über http://dnb.dnb.de abrufbar.

© 2014 Dr. Ludwig Reichert Verlag WiesbadenISBN: 978-3-95490-025-1

www.reichert-verlag.deDas Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt.

Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar.

Das gilt insbesondere für Vervielfältigungen, Übersetzungen,Mikroverfilmungen und die Speicherung

und Verarbeitung in elektronischen Systemen.Gedruckt auf säurefreiem Papier

(alterungsbeständig pH7 –, neutral)Printed in Germany

Gedruckt mit Unterstützung des„Interdisziplinären Zentrums Alte Welt“ der Universität Erlangen-Nürnberg.

InhaltsverzeichnisInhaltsverzeichnisInhaltsverzeichnisInhaltsverzeichnis

Vorwort 7

ACKERMANN, Katsiaryna: Bemerkungen zu den morphonologischen Resten einiger nicht belebter ieur. Kollektiva im Baltischen und Slavischen

9–23

CATT, Adam Alvah: A “Lost” i-Stem: Pāli pi��hi- ‘back’ 24–31

DAHL, Eystein: On the semantics and syntax of the Latin ‘double dative’ construction

32–50

DARDANO, Paola: Strategien der Nominalisierung im Hethitischen: Die Nomina agentis

51–64

FELLNER, Hannes A.: Das Femininum der thematischen Adjektiva im Tocharischen

65–77

FRITZ, Matthias: Vom Wandel zwischen den Dimensionen 78–87

GRESTENBERGER, Laura: Zur Funktion des Nominalsuffixes *-i- im Vedischen und Urindogermanischen

88–102

HARÐARSON, Jón Axel: Das Wort für ‚Eisen‘ im Keltischen und Germanischen und die indogermanischen -erno-Bildungen

103–112

KEYDANA, Götz: Ablaut in indogermanischen Primärnomina: Die hysterokinetischen Stämme

113–128

KIM, Ronald I.: Ablative and comitative in Tocharian 129–139

KLOEKHORST, Alwin: The Proto-Indo-European Acrostatic Inflection Reconsidered

140–163

KÜMMEL, Martin J.: Zum „proterokinetischen“ Ablaut 164–179

LÜHR, Rosemarie: Substantiv – Adjektiv – Pronomen als lexikalische und funktionale Köpfe

180–194

MALZAHN, Melanie: Das Kollektivum im Tocharischen 195–201

MEIER-BRÜGGER, Michael: Zur Bildung von urindogermanisch *melit- ‚Honig‘

202–204

MELCHERT, H. Craig: Anatolian Nominal Stems in *-(C)o- 205–214

NUSSBAUM, Alan J.: Greek τPκµαρ ‘sign’ and τPκµωρ ‘sign’: Why both? 215–260

OLSEN, Birgit Anette: On the Role of Stative Markers in Indo-European Noun Formation

261–272

PINAULT, Georges-Jean: Distribution and Origins of the PIE Suffixes *-ih2- 273–306

PLATH, Robert: Mykenisch e-u-te-re-u und der Lokativ Singular der i-Stämme im spätbronzezeitlichen Griechisch

307–317

6

PRONK, Tijmen: Proto-Indo-European mn-stems in Balto-Slavic 318–326

RAU, Jeremy: The History of the Indo-European Primary Comparative 327–341

RIEKEN, Elisabeth & WIDMER, Paul: Kongruiert alles? Zu den Kongruenz-mustern des Pronominaladjektivs der Bedeutung ‚all, jeder, ganz‘ im Griechischen und Hethitischen

342–359

SCHAFFNER, Stefan: Die slavischen Ethnonyme des Typs *polj'áne

‚Feldbewohner‘ und die griechischen Ethnonyme auf -Xνες 360–383

SOMMER, Florian: Avestisch viš 384–396

STEER, Thomas: Von der Hysterokinese zur Amphikinese: Akzentgebundener Ablaut bei der Substantivierung athematischer Adjektive

397–412

SUKAČ, Roman: Three Problems of Theoretical Morphology in Indo-European Languages

413–425

On the On the On the On the semantics and syntaxsemantics and syntaxsemantics and syntaxsemantics and syntax of the Latin of the Latin of the Latin of the Latin ‘double dative’ construction‘double dative’ construction‘double dative’ construction‘double dative’ construction****

Eystein Dahl

1. 1. 1. 1. Introductory remarksIntroductory remarksIntroductory remarksIntroductory remarks This paper takes a fresh look upon a copular construction in Latin which may be illustrated by the following examples:

(1a) Gaudio ero vobis joy-DAT be-1SG.FUT you-DAT

‘I will give you joy’ (Plautus, Poenulus 1217)

(1b) In Asiam ire nolui, quod et celebritas mihi

to Asia go not.want because and large.assembly-NOM I-DAT

odio est hate-DAT be-PRS.3SG

‘I decided not to go to Asia (both) because the large assembly is disagreeable to me (…)’ (Cicero, Epistulae Ad Atticum 3.19)

These examples illustrate that the construction under scrutiny consists of two dative-marked elements with a well-defined functional difference. One element typically consists of a personal pronoun or, somewhat less frequently, an animate noun in the dative, while the other element characteristically is an abstract noun denoting some state which affects the participant denoted by the personal pronoun or animate noun. Henceforth, I will refer to this construction as the ‘double dative’ construction. This paper aims to delimit its semantic and syntactic properties. The remainder of this paper is organized as follows. Section 1.1 discusses a few theoretical assumptions and terminological issues and section 1.2 attempts to ascertain the status of the double dative construction in Latin, suggesting that the double dative construction represents a fixed construction in Latin and that it may be plausibly ana-lyzed as a two-place compositional verb phrase consisting of an abstract noun in the dative and the copula. In section 2, the semantic and syntactic properties of the double dative con-struction are examined in some detail. Section 2.1 focuses on the lexical semantic proper-ties of the predicate types occurring in the double dative construction, showing that this construction type shows a relatively restricted and semantically coherent distribution, being restricted to experiential and benefactive/malefactive predicates. Section 2.2 scrutinizes the relative syntactic status of the two arguments in the construction under discussion. The gen-eral claims of the paper are summarized in section 3.

* This paper is a revised version of a talk I presented at the workshop ‘Changes in Case and Argument Structure

in the Ancient and Archaic Indo-European Languages’, University of Bergen 12th May 2011 and at the Ar-beitstagung der Indogermanischen Gesellschaft ‘Das Nomen im Indogermanischen. Morphologie, Substantiv

versus Adjektiv, Kollektivum’, Friedrich-Alexander-Universität Erlangen-Nürnberg 15th September 2011. I

thank the audiences of these conferences for critical suggestions and remarks, in particular Marina Benedetti, Chiara Fedriani, Mark Hale, Thomas Krisch, H. Craig Melchert, Georges-Jean Pinault, and Stefan Zimmer.

On the semantics and syntax of the Latin ‘double dative’ construction 33

I wish to emphasize from the outset that the results presented here are part of an ongo-ing investigation and hence to some extent are less conclusive than one might wish. How-ever, as the construction under discussion presents us with several interesting questions as regards the relationship between semantics and syntax in Latin, I believe a publication of my results, preliminary as they are, may be well justified, after all. 1.1. 1.1. 1.1. 1.1. Theoretical and methodological Theoretical and methodological Theoretical and methodological Theoretical and methodological presuppositionpresuppositionpresuppositionpresuppositionssss This paper is based on a functionally oriented approach to language structure, assuming that many aspects of morphology and syntax are motivated by semantics, and on a composition-al approach to linguistic semantics, which, among other things, presupposes that the mean-ing of any given sentence is determined by the meaning of the constituents in the sentence and the rules by which they combine. The same applies to the morphological and syntactic constructions in a given language, in the sense that their distribution is taken to be restricted by their inherent semantic properties. This assumption has a number of potentially impor-tant consequences for historical linguistics. In the present context, however, it suffices to note that it provides an important heuristic tool for determining the semantic properties of a given construction in dead languages, implying that distribution is indicative of construc-tional semantics. Consider, by way of illustration, the examples in (2).

(2a) ipsus Amphitruo optruncavit regem

himself-NOM Amphitruo-NOM kill-3SG.PRF king-ACC

Pterelam in proelio. Pterela-ACC in battle-ABL

‘Amphitruo himself killed king Pterela in the battle’ (Plautus, Amphitruo 415)

(2b) Iuppiter non minus quam vostrum quivis Jupiter-NOM not less than you-GEN anyone-NOM

formidat malum

dread-3SG.PRS evil-ACC

‘Jupiter dreads evil not less than any one of you’ (Plautus, Amphitruo 26 after Dahl & Fedriani 2012:345)

(2c) dum me civitatis morum piget

while I-ACC state-GEN manners-GEN regret-3SG.PRS ‘While I regret the manners of the state’ (Sallust, Bellum Iugurthinum 4.9.2)

These examples illustrate two distinct argument realization patterns in Latin. The passages cited in (2a) and (2b) instantiate the canonical nominative-accusative construction which is found with prototypically transitive verbs like obtrunco ‘I kill’, i.e. two-place verbs with a causative-transitive meaning, as well as verbs with a somewhat less obviously transitive se-mantics like formido ‘I am afraid’. Prototypically transitive verbs like obtrunco ‘I kill’ by definition select what is commonly referred to as an agentive first argument expressed by the nominative case (ipsus Amphitruo ‘Amphitruo himself’) and what is commonly referred to as a patientive second argument expressed by the accusative case (regem Pterelam ‘king Pterela’), a reflection of the fact that Latin generally shows nominative-accusative align-ment. Verbs like formido ‘I am afraid’, on the other hand, select what is commonly referred to as an experiencer first argument which is expressed by the nominative (Iuppiter ‘Jupi-ter’) and a stimulus second argument which is expressed by the accusative (malum ‘evil’).

34 Eystein Dahl

Finally, the example cited in (2c) instantiates a very different argument realization con-struction pattern with an accusative-marked and a genitive-marked argument which is found with a strictly limited number of verbs in Latin, namely pudet ‘is ashamed’, dispudet ‘is greatly ashamed’, suppudet ‘is somewhat ashamed’, taedet ‘is bored’, distaedet ‘is very bored’, pertaedet ‘is very bored’, miseret ‘feels pity’, commiseret ‘feels pity’, piget ‘feels annoyance, pity or regret’ and paenitet ‘repents, is sorry’, suppoenitet ‘is somewhat sorry’ (cf. Baños Baños 2003, Fedriani 2009, 2012 for discussion and references). This argument realization construction is limited to verbs denoting a (negative) emotional state, i.e. to a subclass of experiential predicates. It ascribes accusative case-marking to the experiencer argument (me ‘me’) and genitive case-marking to the stimulus argument (morum ‘man-ners’). The examples in (2) suffice to illustrate that certain argument realization construc-tions, like the nominative-accusative construction are quite inclusive, being compatible with agentive predicates as well as experiential predicates and others, whereas others, like the accusative-genitive construction have a far more limited distribution, being restricted to a small subclass of experiential predicates. Because of these lexical restrictions, the latter construction may be formally represented as EXPACC-STIMNOM. The nominative-accusative construction, on the other hand, does not appear to be subject to any lexical restrictions and may therefore be represented as XNOM-YACC, a notation which is intended to include con-structions like AGTNOM-PATACC and EXPNOM-STIMACC and several others.

The fact that the accusative-genitive construction does not occur with prototypically transitive verbs like obtrunco ‘I kill’, capio ‘I take’ or expello ‘I expel’ may be regarded as a consequence of the fact that this particular construction is semantically incompatible with agentive semantics. Assuming along the lines of Dowty (1991), Primus (1999), Grimm (2005) and others that traditional semantic roles like agent, patient, experiencer and stimu-lus may be analyzed in terms of abstractions over clusters of privative lexical entailments, it is reasonable to analyze different lexical distribution patterns of the type just discussed in terms of compatibility or incompatibility between particular lexical entailments and con-structional semantics. At least five lexical semantic properties may be assumed to be rele-vant for the classification of thematic roles. Specifically, verbs may differ with regard to whether they denote a (total) change of state or not [+BECOME], whether they involve causation [+CAUSATION], whether they entail that one or more of the core arguments are animate [+SENTIENCE] and, if so, whether they entail that the animate argument(s) con-trol the outcome of the situation [+VOLITION] or whether they entail that the animate ar-gument(s) is affected by the situation [+AFFECTED] (cf. Dahl 2009 for discussion). I fur-ther assume that causative-transitive verbs denoting a change of state may be regarded as prototypically transitive and it is reasonable to suggest that two-place verbs are organized hierarchically, with causative-transitive verbs in the most prominent position and other two-place verbs ordered downwards according to their respective lexical entailments (cf. e.g. Tsunoda 1985, Blume 1998, Grimm 2005, and Dahl 2009 for similar suggestions). Under this approach, a prototypical agent may be defined as an animate referent characterized by a volitional causation of a change of state in another referent. In contrast, experiential predi-cates typically entail that the experiencer argument is animate and affected by the situation but are underspecified with regard to and, hence, in principle compatible with the other three lexical entailments. Finally, I would like to suggest that the lack of the five mentioned entailments is characteristic of the benefactive/malefactive role. In other words, the main difference between an experiencer and a benefactive or malefactive argument is that the former is associated with the entailments [+SENTIENCE] and [+AFFECTED] whereas the latter is underspecified with regard to these features. Given these assumptions, one might

On the semantics and syntax of the Latin ‘double dative’ construction 35

hypothesize that the EXPACC-STIMGEN construction presupposes that the predicate entails that there is at least one animate argument which is affected by the situation, whereas the XNOM-YACC construction is not associated with any such presupposition as it appears to be compatible with virtually any type of predicate in Latin.1 It is tempting to connect the ex-clusive character of the EXPACC-STIMGEN construction with the fact that it represents a morphosyntactically rather marked constellation within the Latin argument realization sys-tem. Specifically, the construction or a part of the construction could be associated with some semantic feature which is incompatible with agentive semantics;2 however, a full evaluation of this possibility will have to be pursued elsewhere.

The observant reader will have noticed that I have not made any reference to syntactic relations like subject or object in the discussion so far. According to received lore, the sub-ject is normally expressed by the nominative case and the object is normally expressed by the accusative case in Latin (cf. e.g. Hofmann & Szantyr 1972:27, 30). This point is implic-itly included in the observation that Latin generally shows nominative-accusative align-ment; however, cases like (2c) where no nominative is present indicate that a nominative subject is not a necessary condition for a finite verbal construction to be grammatical. This construction also suggests that verb agreement is not a necessary condition for subjecthood, as neither the accusative-marked argument me ‘me’ nor the genitive-marked argument morum ‘manners’ agree with the finite verb which appears in the third singular form which by scholars like Benveniste (1960 = 1966:187–207) has been regarded as an unmarked or neutral form of the verb. This fact raises the question whether such constructions are best regarded as subjectless or impersonal. One important argument against regarding examples like (2c) as subjectless is semantic in nature. Verb phrases that are subjectless in a strict sense characteristically involve meteorological verbs like ningit ‘snows’, pluit ‘rains’ or tonat ‘thunders’ which in their concrete, basic sense do not seem to entail any participants in the situations they denote. In this respect, they have fundamentally different semantic properties than the verbs that are associated with the EXPACC-STIMGEN construction which entail two participants. Nevertheless it is far from clear which of the arguments represents the subject of the verb phrase, neither is it evident how this can be established in a princi-pled manner.3 The idea, first formulated by Keenan (1976) that subjecthood may be regard-

1 Unfortunately, the empirical consequences of this hypothesis cannot be further pursued in this context. At

present it suffices to note that the XACC-YNOM construction is sufficiently inclusive to include even verbs that occur in the EXPNOM-STIMGEN construction, as illustrated by the following example:

non te haec pudent?

not you-ACC these-NOM shame-3PL.PRS ‘Don’t these things embarrass you?’ (Terence, Adelphoe 754)

Note that this type of example is only attested in Archaic Latin and does not occur in Classical Latin.

2 It is tempting to suggest that an incompatibility of this kind could arise from the genitive case marking of the stimulus argument as this is the only formal feature that distinguishes this construction from another experien-

tial construction in Latin which is more frequently met with, namely the STIMNOM-EXPACC construction

which primarily seems to be associated with causative experiential predicates (cf. Fedriani 2012, Dahl & Fedriani 2012 for discussion).

3 Some readers might want to object to the use of the term ‘subject’ in cases where nominative case and verb

agreement is absent, as these two features are widely regarded as constitutive of the notion of subject in Latin (and many other languages). I therefore think that a terminological clarification is in order. In the following,

my use of the term ‘subject’ will in effect be co-referential with the term ‘Privileged Syntactic Argument’ in

Role and Reference Grammar (cf. Van Valin 2005) rather than with a morphologically based definition of subject.

36 Eystein Dahl

ed as a multidimensional concept, including coding properties like nominative case and verb agreement, behaviour and control properties like deletion, movement, case-changing properties etc. as well as semantic properties like agency, autonomous existence etc., as represented in Table 1, may provide a solution to this dilemma. On this assumption, the no-tion of subject might be regarded as a prototype concept, according to which different types of subjects are organized in a hierarchical fashion along a lattice structure with highly agen-tive nominative subjects of the type in (2a) at the top end and less agentive nominative sub-ject of the type illustrated in (2b) somewhere further down in the hierarchy (cf. Grimm 2005, Dahl 2009 for a discussion of the notion of a lattice structure).

Table 1: Promotion to Subject Hierarchy (Keenan 1976:324)

Coding Properties > Behaviour and Control Properties > Semantic Properties position > case marking deletion, movement, case changing agency, autonomous > verb agreement properties, control of cross-reference existence, selectional properties, etc. restrictions, etc.

Given these assumptions, constructions of the type illustrated in (2c) represent a kind of ar-gument realization pattern where none of the arguments shows the coding properties char-acteristic of prototypical subjects. In other words, morphological features appear to provide an insufficient means for identifying subjects in Latin. Moreover, both experiencer argu-ments and stimulus arguments can be selected as subject in Latin, something which rules out a decision on a strictly semantic basis. Keenan’s notion of behavioural properties, how-ever, implies that subjects may be expected to be identifiable by their syntactic behavior, as has, for instance, been convincingly argued to be the case in the Germanic languages (cf. e.g. Barðdal & Eythórsson 2003, Eythórsson & Barðdal 2005). This might provide a way of determining whether the accusative-marked experiencer or the genitive-marked stimulus represents the subject of constructions of the type illustrated in (2c). Taking prototypical nominative subjects in Latin as a point of departure, they are characterized by at least the following properties:

(3a) They tend to precede the other arguments of the clause,4 cf. e.g.:

ista tua pulchra libertas deos This-NOM your-NOM beautiful-NOM liberty-NOM gods-ACC

penatis et familiaris meos lares household.gods-ACC and household-ACC my-ACC gods-ACC

expulit expel-3SG.PRF

‘That beautiful Liberty of yours expelled my household gods and the eternal divinities of my hearth’ (Cicero, De Domo Sua 108.9)

4 Note that my inclusion of position as a behavioural property differs substantially from the account given in

Keenan 1976:324 who regards this as a coding property rather than a behavioural property. However, I

believe it is preferable to restrict the notion of coding properties to strictly morphological criteria and therefore take position as a syntactic and hence behavioural criterion.

On the semantics and syntax of the Latin ‘double dative’ construction 37

(3b) They are able to bind reflexives, cf. e.g.:

tua tamen dignitasi suumi locum your-NOM still dignity-NOM own-ACC place-ACC

obtinebit maintain-1SG.FUT

‘Your dignity will still maintain its place’ (Cicero, Epistulae ad Familiares 3.9.2.13)

(3c) They generally tend to be omitted in subject control constructions, cf. e.g.:

deinde sententiam meam tui facillime PROi then feeling-ACC my-ACC you-NOM easily

perspicere potuisti iam ab illo tempore understand-INF can-2SG.PRF already from that-ABL time-ABL

‘In the next place, you might without any difficulty have understood my feeling from that time onward.’ (Cicero, Epistulae ad Familares 2.16)

(3d) They tend to appear before other arguments in raising-to-object constructions, cf. e.g.:

volo tei PROi uxorem domum ducere want-1SG.PRS you-ACC wife-ACC house-ACC lead-INF

‘I want you to lead a wife home (i.e. to marry)’ (Plautus, Aulularia 150)

(3e) They are systematically omitted in conjunction reduction, cf. e.g.:

Consternatus egoi exsurrexi et Xi timui, terrified-NOM I-NOM rise-1SG.PRF and be.afraid-1SG.PRF

ne per tectum petauristarius aliquis that through roof-ACC acrobat-NOM some-NOM

descenderet. come.down-3SG.PRS.SBJ

‘I rose from my place in a panic and was afraid that some acrobat would come down through the roof.’ (Petronius, Satyricon 60.2)

Recent studies have shown that the accusative-marked experiencer argument behaves more like a nominative-marked subject with regard to these and other parameters than the geni-tive-marked stimulus argument with the relevant group of verbs (cf. e.g. Baños Baños 2003, Fedriani 2009, Dahl in preparation). This provides a methodologically well founded argument in favor of the assumption that Latin had oblique or non-nominative subjects. 1.2. 1.2. 1.2. 1.2. The status of the The status of the The status of the The status of the double dative construction in Latindouble dative construction in Latindouble dative construction in Latindouble dative construction in Latin Somewhat surprisingly, the double dative construction has not received much attention in the literature. For example, the philological handbooks provide a rather sketchy analysis of this construction, as illustrated by the following passage:

38 Eystein Dahl

(4) „Zwei Dative bei einem Verbum. Tritt zu einem persönlichen Dat. als Apposi-tion ein Dat. der Sache, so können dieselben eine mehr oder weniger enge Ver-bindung eingehen. (…) Die Hauptmasse der doppelten Dat. stellt jedoch die Ver-bindung eines persönlichen Dat. mit einem Dat. finalis“ (Hofmann & Szantyr 1972:99).

Although a description along these lines leaves many questions open, some of which will be addressed in the following, it does capture at least one important aspect of this construc-tion type. Assuming that the various readings associated with a morphological case catego-ry as a rule can be accounted for as lexically and contextually determined variants of a ba-sic general meaning, the double dative construction illustrates how two relatively well de-fined distinct functions of one and the same case category are used contrastively in one and the same syntactic construction. In my opinion, however, the above definition does not give a satisfactory account of the relationship between the two dative-marked elements, as it seemingly implies that the double dative construction represents a sporadic (or even ran-dom) combination of two more or less independent datives. This assumption faces at least two empirical objections. For one thing, the double dative construction is generally well at-tested from Plautus onwards and may therefore be regarded as a fixed or grammaticalized construction already in the earliest stage of literary Latin. It is also dubious to what extent the abstract noun may be regarded as an ‘apposition’ to the ‘personal’ dative. On the con-trary, from a semantic perspective, the personal dative intuitively seems to be dependent on the other dative-marked element, cf. for instance the following examples as well as the ex-amples in (1) above:

(5a) Aut tibi nuptiae haec sunt cordi. or you-DAT wedding-NOM this-NOM be-PRS heart-DAT

‘or is this marriage to your liking?’ (Terence, Andria 327)

(5b) Caesar … pollicitusque est Caesar-NOM promised-NOM.and be-PRS

sibi eam rem curae futuram

self-DAT this-ACC affair-ACC concern-DAT be-FUT.PRT

‘And Caesar promised that this affair would be of concern for him’ (Caesar, De

Bello Gallico 1.33.1.2)

(5c) quia nobis lucro fuisti potius quam decori since we-DAT profit-DAT be-PRF.2SG rather than grace-DAT

tibi you-DAT

‘Since you have rather been the cause of profit to us than of reputation to yourself’ (Plautus, Asinaria 192)

In cases like these and those cited in (1), the dative-marked abstract nouns gaudio ‘joy, gladness’, odio ‘hate’, cordi ‘heart, soul’ specify a state which in some way affects the ref-erents of the dative-marked pronouns mihi ‘me’, tibi ‘you’, sibi ‘himself’ and nobis ‘us’. It is therefore reasonable to regard the abstract nouns as belonging to what may be defined as the nucleus of the predicates, i.e. the innermost layer of the clause, and the dative-marked

On the semantics and syntax of the Latin ‘double dative’ construction 39

pronouns as arguments belonging to the core of the predicates, i.e. the second innermost layer of the clause (cf. Van Valin 2005:4f. for terminology and discussion).

Moreover, we find specific instantiations of the double dative constructions in various authors at different times; this is, for instance, the case with the constructions cited above: alicui gaudio esse ‘be a source of joy to someone’ is found e.g. in Plautus (1a), Sallust5 and Pliny the Younger,6 alicui odio esse ‘be disagreeable to someone’ is found in authors like Plautus,7 Cicero (1b) and Seneca,8 alicui cordi esse ‘be agreeable to someone’ in authors like Plautus,9 Terence (5a), Caesar,10 Cicero11 and Tacitus,12 alicui curae esse ‘be of con-cern to someone’ in authors like Plautus,13 Terence,14 Caesar (5b), Cicero15 and Sallust,16 alicui lucro esse ‘be of profit to someone’ in authors like Plautus (5c) and Seneca,17 and decori alicui esse ‘be of honor to someone’ in authors like Plautus (2c), Virgil,18 Sallust19 and Tacitus.20 Taken together, these facts provide prima facie evidence that the double da-tive construction represents a fixed construction more or less on a par with other syntactic constructions in the language.

It is unclear how an account of the type cited in (4) can accommodate these observa-tions. I therefore believe that it is preferable to analyze this construction as a two-place compositional verb phrase consisting of an abstract noun in the dative and the copula, as tentatively represented in (6).

(6) [XNOM, YDAT[ABSTRACT NOUNDAT<STATE> + ESSE]]]

In the next section, I examine the semantic and syntactic properties of this construction, i.e. what predicate types it is compatible with and the relative status of its two argument posi-tions. 2. 2. 2. 2. The semantic and syntactic properties of the double dative construThe semantic and syntactic properties of the double dative construThe semantic and syntactic properties of the double dative construThe semantic and syntactic properties of the double dative constructionctionctionction The main aim of this section is to establish the semantic and syntactic properties of the Lat-in double dative construction. Section 2.1 explores the semantics of the types of predicates that are found in the double dative construction in Latin, showing that although we find a considerable number of double dative predicates with an experiential character, an even greater number of predicates have a less obviously experiential character. Section 2.2 fo-cuses on the syntactic properties of the two arguments and discusses the relationship be-

5 Cf. e.g. Sallust, Bellum Iugurthinum 9.2.2.

6 Cf. e.g. Pliny the Younger, Epistulae 2.11.1.2.

7 Cf. e.g. Plautus, Trinummus 631. 8 Cf. e.g. Seneca, De Ira 1.14.1, 1.14.2.

9 Cf. e.g. Plautus, Cistellaria 109.

10 Cf. e.g. Caesar, De Bello Gallico 6.19.4.2. 11 Cf. e.g. Cicero, Epistulae ad Atticum 5.3.3.1.

12 Cf. e.g. Tacitus, Historiae 3.53.14, 4.58.35.

13 Cf. e.g. Plautus, Bacchides 1077. 14 Cf. e.g. Terence, Adelphoe 129, 680, 894.

15 Cf. e.g. Cicero, Epistulae ad Atticum 1.1.1.1.

16 Cf. e.g. Sallust, Bellum Iugurthinum 14.16.3. 17 Cf. e.g. Seneca, De Beneficiis 5.7.6.5.

18 Cf. e.g. Virgil, Eclogae 5.32.

19 Cf. e.g. Sallust, Bellum Iugurthinum 19.1.5, 73.4.2, 85.40.4. 20 Cf. e.g. Tacitus, Historiae 3.84.16.

40 Eystein Dahl

tween the double dative construction and other predicate types with a nominative-marked and a dative-marked argument. 2.1. 2.1. 2.1. 2.1. The semantics of the double dative constructionThe semantics of the double dative constructionThe semantics of the double dative constructionThe semantics of the double dative construction As noted in the previous section, the semantics of a given construction type is taken to be reflected in the range of predicates with which it is compatible. A representative list of nuclear dative-marked constituents with an experiential character occurring in the double dative construction in Archaic and Classical Latin is given in (7).

(7) acerbitati ‘be (a source) of hatred’, consolationi ‘(be of) consolation, comfort’, contemptui ‘be a source of contempt’, cordi ‘please’, curae ‘(be a source of) anx-iety’, damnationi ‘(be a source of) condemnation’, delectationi ‘be a source of pleasure’, dolori ‘’be a source of pain’, fastidio ‘be a source of aversion’, invidiae ‘source of jealousy, grudge’, inrisui ‘source of derision’ iucunditati ‘be agreeable’, laetitiae ‘be a source of joy’, levationi ‘be a relief’, maerori ‘be a source of grief’, molestiae ‘be troublesome’, odio ‘be an object of hate’, offen-

sioni ‘source of vexation’, ridiculo ‘be a source of laughter’ rubori ‘be a source of shame’, solacio ‘be a source of comfort’, sollicitudini ‘be a source of anxiety’, spectaculo ‘be an amusement’, stomacho ‘be unpleasant’, taedio ‘source of wea-riness’, timori ‘source of fear’, voluptati ‘be a source of pleasure’.

These data indicate that some of the double dative predicates in Latin have a relatively clear-cut experiential character, in that they presuppose that the dative-marked participant is sentient and emotionally affected by the situation denoted by the verb. The participant which causes the emotion is often referred to as the ‘stimulus’. Consider the following ex-amples, as well as examples (1ab) and (5ab) above:

(8a) omnibus iste ceteris Siculis odio est all-DAT this-NOM other-DAT Sicilians-DAT hate-DAT be-PRS.3SG

‘He is a source of hate for the other Sicilians’ (Cicero, In Verrem 2.4.15)

(8b) nam Ti. Graccho (…) senatus severitas for Tiberius Gracchus-DAT senate-GEN severity-NOM

dolori et timori fuit pain-DAT and fear-DAT be-PRF.3SG

‘The severity of the senate was a constant source of grief and fear to Tiberius Gracchus’ (Cicero, De Haruspicum Responso 20)

(8c) si populo ludorum magnificentia voluptati if people-DAT games-GEN magnificence-NOM pleasure-DAT

est be-PRS.3SG

‘If the magnificence of games is a pleasure to the people (…)’ (Cicero, Pro

Murena 19)

In these and similar cases, the dative-marked participants omnibus ceteris Siculis ‘the other Sicilians’, Ti. Graccho ‘Tiberius Gracchus’ and populo ‘the people’ represent fairly typical experiencer arguments, whose sentiments are specified by the dative-marked nouns odio

On the semantics and syntax of the Latin ‘double dative’ construction 41

‘source of hate’, dolori et timori ‘source of pain and fear’ and voluptati ‘source of pleasure’. Predicates of this type may generally select animate experiencer arguments with dative case marking (DATEXP) and inanimate or, somewhat more rarely, animate stimulus arguments with nominative case marking (NOMSTIM).

In the cases discussed so far, the double dative predicates included abstract nouns de-noting a situation presupposing an animate participant which is emotionally involved in the situation, i.e. abstract nouns associated with the entailments [+SENTIENCE] and [+AF-FECTED]. However, although experiential predicates clearly represent a salient subclass of double dative predicates, predicates with a less obviously experiential character are like-wise very frequent. A representative list of non-experiential predicates is given in (9)

(9) adiumento ‘be of help’, argumento ‘(serve as an) argument’, auxilio ‘(be of) as-sistance, support’, bono ‘be good’, calamitati ‘(be of) damage, mischief’, crimini ‘(be a source of) accusation’, decori ‘be a source of dignity’, dedecori ‘unseem-ly’, detrimento ‘be of injury’, dignitati ‘be a source of dignity’, documento ‘(serve as an) example’, emolumento ‘(be of) benefit’, exemplo ‘(serve as an) ex-ample’, exitio ‘destruction’, fastidio ‘be a source of aversion’, firmamento ‘sup-port’, fraudi ‘(be of) injury, fructui ‘(be of) profit’, frugui ‘be useful, proper’, gloriae ‘(be a source of) honor’, honori ‘(be a source of) honor’, impedimento ‘be an impediment’, indicio ‘serve as proof’, infamiae ‘be a source of disgrace’, labori ‘be of trouble’, laudi ‘be a source of fame’, lucro ‘be of profit’, maculae ‘be a stain’, malo ‘be bad’, morae ‘be a source of delay’, oneri ‘be a burden’, opprobrio ‘be a source of disgrace’, ornamento ‘mark of honor’, saluti ‘be a means of safety’, subsidio ‘protection’, sumtui ‘expense’, testimonio ‘proof’, usui ‘be useful’, utilitati ‘be useful’, vituperationi ‘blame’

Consider the following examples as well as (5c) above:

(10a) qui magno nobis usui ad bellum which-NOM great-DAT we-DAT use-DAT to war-ACC

gerendum erant be.waging-ACC be-PRS.3PL

‘Which were of great use for us in the waging of war’ (Caesar, De Bello Gallico 2.9)

(10b) si propinquos habeant imbecilliore vel animo vel if relatives-ACC have-PRS.3PL weaker-ABL either mind-ABL or

fortuna (…) eisque honori sint et fate-ABL they-DAT.and honor-DAT be-PRS.SBJ.3PL and

dignitati dignity-DAT

‘If they have relatives with a weaker mind or fortune (…) and they should be a source of honor and dignity for them’ (Cicero, De Amicitia 70)

(10c) sed vereor ne (…) dedecori sim but fear-PRS.1SG that unseemliness-DAT be-PRS.SBJ.1SG

42 Eystein Dahl

studiis ac litteris nostris studies-DAT and writings-DAT our-DAT

‘I fear being a source of unseemliness to my philosophy and my writings’ (Cicero, Epistulae ad Atticum 8.11)

(10d) cui rei testimonio sunt litterae which-DAT matter-DAT witness-DAT be-PRS.3PL letters-DAT

quas Bussenius ad me misit which-ACC Bussenius-NOM to I-ACC send-PRF.3SG

‘The letters that Bussenius sent to me bear witness of this matter’ (Cicero, Epis-

tulae ad Atticum 8.12)

These examples illustrate that non-experiential double dative predicates like usui esse ‘be of use’, honori ac dignitati esse ‘be a source of honor and dignity’, dedecori esse ‘be un-seemly’ and testimonio esse ‘bear witness’ may select animate (nobis ‘us’, eis ‘for them’) or inanimate (studiis ac litteris ‘philosophy and writing’, cui rei ‘for this matter’) dative-marked arguments. Note that there appear to be extremely few cases of the latter type, a distributional fact which indicates that the constructional semantics of the double dative construction is in principle compatible with animate as well as inanimate referents in the ar-gument position triggering dative case marking but tends to give preference to animate ref-erents in this position.

In examples like those cited in (10), then, the dative-marked abstract nouns denote a situation which neither entails that any of the participants in the situation is directly affected by the situation nor that they are animate. Above, it was suggested that the lack of these en-tailments is characteristic of the benefactive/malefactive role. Thus, the double dative con-struction primarily appears to include experiential and benefactive/malefactive predicates and, as a consequence, we may render our preliminary definition of the double dative con-struction somewhat more specific, as in (11). Note that the abbreviation BEN is intended to represent both the benefactive and the malefactive role.

(11) [STIMNOM, EXP/BENDAT[ABSTRACT NOUNDAT<STATE> + ESSE]]

2222.2. .2. .2. .2. The syntactic status of the arguments in the double dative constructionThe syntactic status of the arguments in the double dative constructionThe syntactic status of the arguments in the double dative constructionThe syntactic status of the arguments in the double dative construction Having established that the double dative construction mainly includes experiential and benefactive/malefactive predicates, we may now turn to the second of the two questions posed above, namely the relative status of the nominative-marked and the dative-marked ar-gument. Specifically, which of these two arguments represents the first argument or syntac-tic subject and which represents the second argument or syntactic object? As Latin repre-sents a nominative-accusative language, the obvious answer to this question is that the nominative-marked element is the subject and the dative-marked argument represents an object argument with non-standard case marking. However, although many scholars would probably accept this account without hesitation, it is worth pointing out that it takes for granted that there is a symmetrical relationship between (case) morphology (or coding pro-

perties) and syntax (or behavioural properties) and that the syntactic subject is invariably expressed by nominative case. Although this hypothesis would appear to account for a con-siderable majority of the various argument realization patterns in Latin, it was argued above that this assumption fails to account for certain constructions in Latin, e.g. the EXPACC-STIMGEN construction which appears to constitute an exception to this general tendency. As

On the semantics and syntax of the Latin ‘double dative’ construction 43

regards their respective behavioural properties, Baños Baños (2003) and Fedriani (2009) have convincingly shown that the accusative-marked participant behaves syntactically like a nominative-marked subject, a point which justifies the assumption that Latin allowed ob-lique or non-nominative subjects.

For our present purposes, another relatively salient argument realization construction in Latin which characteristically involves a dative-marked and a nominative-marked argument comes into mind; cf. e.g. the examples in (12). This argument realization pattern mainly oc-curs with verbs of pleasing/displeasing but is also found with certain other experiential predicates as well as predicates with a less evidently experiential character (cf. Fedriani 2012 for discussion).

(12a) mihi et res et condicio placet I-DAT and property-NOM and terms-NOM please-PRS.3SG

‘I am happy both with the property and the terms’ (Cicero, Epistulae ad Atticum 12.25)

(12b) Varroni quidem displicet consilium pueri, Varro-DAT in.fact displease-3SG.PRS counsel-NOM youth-GEN

mihi non. I-DAT not

‘In fact, the youth’s plan displeases Varro, not me’ (Cicero, Epistulae ad Atticum 16.9.1.8)

(12c) damna evenerunt maxuma misero mihi mishaps-NOM happen-3PL.PRF greatest-NOM poor-DAT I-DAT

‘The greatest mishaps have befallen wretched me’ (Plautus, Stichus 209)

(12d) tua disciplina nec mihi prodest your-NOM learning-NOM not I-DAT be.useful-3SG.PRS

nec tibi not you-DAT

‘Your learning is neither useful to me nor to you’ (Plautus, Bacchides 135)

These examples illustrate that simple predicates like placeo ‘please’, displiceo ‘displease’, evenio ‘happen, befall’ and prosum ‘be useful’ select a stimulus argument in the nominative and an experiencer (mihi ‘me’, Varroni ‘Varro’) or benefactive/malefactive (misero mihi ‘poor me’, mihi) in the dative. Data like these indicate that the double dative construction may be regarded as a special instantiation of a more general STIMNOM-EXP/BENDAT con-struction in Latin which also occurred with simple predicates. Both the nominative-marked stimulus and the dative-marked experiencer or benefactive show some but not all of the properties associated with syntactic subjects. It is significant that the dative-marked argu-ment shows some behavioural properties characteristic of nominative subjects. We note that it strongly tends to precede the other arguments of the clause21 (13a), it tends to be omitted

21 According to a recent investigation (Fedriani 2012), this constellation is found in 72% of the cases in her cor-

pus. More generally, Devine & Stephens 2006:37 note that the subject strongly tends to precede other verbal arguments in Latin.

44 Eystein Dahl

in conjunction reduction (13b) (cf. also Bauer 2009:280), it can control reflexives (13c) and appears to be raised to object in certain cases (13d).

(13a) mihi et res et condicio placet I-DAT and property-NOM and terms-NOM please-PRS.3SG

‘I am happy both with the property and the terms’ (Cicero, Epistulae ad Atticum 12.25)

(13b) scio Xi quid dictura es: hancj esse pauperem. know-PRS.1SG what say-FUT.2SG she-ACC be-INF poor-ACC

Xi haecj pauper placet she-NOM poor-NOM please-PRS.3SG

‘I know what you will say: that she is poor. But I like her poor’ (Plautus, Aulularia 174)

(13c) Placet ille meus mihi mendicus please-PRS.3SG this-NOM my-NOM I-DAT beggar-NOM

suos rex reginae placet own-NOM king-NOM queen-DAT please-PRS.3SG

‘This beggar of mine is agreeable to me; her own king is agreeable to the queen’ (Plautus, Stichus 133)

(13d) Ne ego, inquam, si itast, velim so.that I-NOM say-1SG.PRS it so be-3SG.PRS want-1SG.PRS.SBJ

tibi eum placere quam maxume. you-DAT he-ACC please-INF as greatest

‘If that is the case’, answered I, ‘I would like you to think of him as favourably as possible.’ (Cicero, Brutus 249.5)

These examples indicate that the dative-marked element in the second non-canonical con-struction show some behavioural properties characteristic of subjects. Although a more thorough investigation of the behavioural properties of the dative-marked and nominative-marked arguments with verbs like placere ‘please’ in (13) is necessary in order to ascertain their syntactic status, we may, for the time being, adopt the assumption that the dative-marked participant shares some behavioural properties with nominative subjects. If this is correct, the question remains whether the double dative construction is analogous to con-structions of the type just discussed in the relevant respects, not the least because it also in-stantiates the STIMNOM-EXP/BENDAT construction. The available evidence is not over-whelmingly rich but it is nevertheless significant that the dative-marked argument in the double dative construction shows an analogous behaviour to the dative-marked argument in (13). Specifically, it tends to precede the other arguments of the clause (14a), it can be left out in conjunction reduction (14b), it can bind reflexives (14c) and it sometimes is raised to object (14d).

(14a) si L. Flacco tantus amor in bonos if Lucius Flaccus-DAT so.great love-NOM for good-ACC

On the semantics and syntax of the Latin ‘double dative’ construction 45

omnis, tantum in rem publicam studium calamitati all-ACC so.great for the.republic-ACC devotion-NOM injury-DAT

fuerit be-PRF.SBJ.3SG

‘If the exceeding affection of Lucius Flaccus for all good men, and his great devotion to the republic turns out an injury to him’ (Cicero, Pro Flacco 42)

(14b) magis Xi curae est magisque Xi adformido, more anxiety-DAT be-3SG.PRS more.and fear-1SG.PRS

ne is pereat neu that he-NOM be.ruined-3SG.PRS.SBJ nor

corrumpatur. be.destroyed-3SG.PRS.SBJ

‘The greater is my concern, and the more do I dread lest he may be ruined or corrupted’ (Plautus, Bacchides 1078)

(14c) ovibusi suai lana decori est sheep-DAT own-NOM wool-NOM dignity-DAT be-PRS.3SG

‘For sheep their own wool is a source of dignity’ (Ovid, Metamorphoses 13.849)

(14d) scio tibi eum non esse odio know-1SG.PRS you-DAT he-ACC not be-INF hate-DAT

‘I know you don't dislike him’ (Cicero, Ad Familiares 8.6)

The above examples appear sufficiently clear to warrant the conclusion that the dative-marked argument in the double dative construction shares some behavioural properties with nominative subjects. However, an important question is to what extent these behavioural patterns are exclusively associated with subjects in Latin or whether they also occur with other types of arguments. Significantly, most, if not all of the above properties are found with non-subject arguments as well. First, Latin has a fairly flexible word order, thereby sometimes allowing direct objects to precede the subject or other arguments of the clause, as illustrated by the example in (15a). Moreover, Latin allows (and in some cases apparent-ly requires) direct objects to be omitted, consider, for instance, the passage in (15b) (cf. also Luraghi 1997). Furthermore, one does indeed find some passages where a reflexive is bound not by the subject of the clause but by the direct object, as seems to be the case in (15c) (cf. also Pieroni 2007, 2010).

(15a) reliquos omnes consecuti equites nostri rest-ACC all-ACC having.pursued-NOM horsemen-NOM our-NOM

interfecerunt slay-3PL.PRF

‘Our cavalry pursued and slew all the rest of them’ (Caesar, De Bello Gallico 1.53)

46 Eystein Dahl

(15b) senatus haeci intellegit, consul Xi videt

‘the senate understands those things, the consul sees them’ (Cicero, Catilina 1.2 after Luraghi 1997)

(15c) nam is illius filiami conicit in navem miles for he-NOM her daughter-ACC throw-3SG.PRF soldier-NOM

clam matrem suami secretly mother-ACC own-ACC

‘For this soldier put her daughter on board the ship without her mother knowing’ (Plautus, Miles Gloriosus 112)

At first glance the raising to object feature, illustrated by the examples in (13d) and (14d), might give rise to some optimism, as there are numerous examples of raising to object con-structions where the subject of the subordinate clause precedes the object, as illustrated by examples like the following (cf. also (3d) above):

(16a) volo te hoc scire want-1SG.PRS you-ACC this-ACC know-INF

‘I want you to know this’ (Cicero, Epistulae ad Atticum 1.18.6)

(16b) audio enim eum ea senatus consulta hear-1SG.PRS in.fact he-ACC those-ACC senate-GEN decrees-ACC

improbare (…) reject-INF

‘I hear that he in fact rejects those decrees of the senate (…)’ (Cicero, Epistulae

ad Atticum 11.7.1)

(16c) ‘scio me’ inquit ‘paene incredibilem know-1SG.PRS I-ACC say-3SG.PRF almost incredible-ACC

rem polliceri’ thing-ACC promise-INF

‘“I know”, he said, “that I promise an almost incredible thing”’ (Caesar, De

Bello Civile 3.86.2)

(16d) te animo magno et forti istam you-ACC spirit-ABL great-ABL and strong-ABL this-ACC

rem agere existimo thing-ACC conduct-INF estimate-1SG.PRS

‘I think you conduct this business with great and strong spirit’ (Cicero, Epistulae

ad Atticum 8.12C.1)

However, although data like these appear to represent a significant tendency in Latin, there are several counterexamples to this general rule. Consider the examples in (17) where a non-subject argument precedes the subject argument in the subordinate clause:

On the semantics and syntax of the Latin ‘double dative’ construction 47

(17a) hoc te intellegere volo This-ACC you-ACC understand-INF want-1SG.PRS

‘I want you to understand this’ (Cicero, Epistulae ad Atticum 1.10.2)

(17b) de Hortensio te certo scio about Hortensius-ABL you-ACC surely know-1SG.PRS

dolere; equidem excrucior be.sorry-INF similarly be.devastated-1SG.PRS

‘I know for sure that you are sorry because of Hortensius. I am also devastated’ (Cicero, Epistulae ad Atticum 6.6.2)

Finally, I would like to draw attention to the fact that the nominative-marked stimulus in the double dative construction does not bind reflexives and does not tend to precede the da-tive-marked argument. Significantly, however, the nominative-marked argument is system-atically omitted in subject control constructions (18a), it is sometimes omitted in conjunc-tion reduction contexts (18b) and it is found in raising to object constructions (18c).

(18a) magno etiam adiumento nobis Hermogenesi great-DAT still help-DAT us-DAT Hermogenes-NOM

potest PROi esse in repraesentando. can-3SG.PRS be-INF in paying.immediately-ABL

‘Hermogenes could be of great help to us in the case of an immediate payment’ (Cicero, Epistulae ad Atticum 12.31.2)

(18b) quam quod modo senexi intro ad nos venit than just now old.man-NOM in to we-ACC come-3SG.PRF

errans. mihi solae Xi ridiculo fuit going.astray-NOM I-DAT alone-DAT laughter-DAT be-3SG.PRF

quae quid timeret scibam. who-NOM what-ACC be.afraid-PRS.SBJ know-1SG.IPF

‘(…) than that the old man just now came in to us, going astray; He was only a source of amusement to me who knew what he was afraid of’ (Terence, Eunuchus 1002–1004)

(18c) Haec non voluptati tibi esse these-ACC not pleasure-DAT you-DAT be-INF

satis certo scio. sufficient surely know-1SG.PRS

‘I am quite sure that these things are not to your liking’ (Terence, Heauton Timo-

roumenos 71)

The data cited in (14) and (18) suggest that the dative-marked and the nominative-marked arguments have overlapping but not identical behavioural properties. Specifically, both can be omitted in conjunction reduction contexts and both are found in raising to object con-structions but only the nominative-marked argument is omitted in subject control construc-tions and only the dative-marked argument shows a strong tendency to appear early in the

48 Eystein Dahl

sentence and to bind reflexives. These considerations suggest that neither the nominative-marked nor the dative-marked argument behave in a manner identical with prototypical subjects.

Summing up the discussion in this section, we note that a relatively strong case can be made for the claim that the double dative construction represents a special case of a more general STIMNOM-EXP/BENDAT construction, both on semantic and on syntactic grounds. Both of these constructions include experiential as well as benefactive predicates and, inter-estingly, the dative-marked argument behaves in a very similar manner in both cases. How-ever, as regards the question about the relative status of the dative-marked constituent in these constructions, the behavioural criteria identified earlier do not provide unambiguous results. Although the dative-marked argument to some extent appears to behave like a syn-tactic subject, so does the nominative-marked argument and, moreover, neither of the be-havioural patterns associated with the dative-marked argument are exclusively associated with syntactic subjects. Thus, while the dative-marked argument represents a relatively good candidate for subject selection from a semantic point of view, the morphological pro-perties of this argument differ substantially from those of prototypical subjects whereas its syntactic behaviour is analogous but not identical to that of prototypical subjects in Latin. 3. 3. 3. 3. Concluding remarksConcluding remarksConcluding remarksConcluding remarks This paper has scrutinized some of the semantic and syntactic properties of the ‘double da-tive’ construction. It was argued that this construction represents a fixed or grammaticalized construction already in the earliest stage of literary Latin, as it is well attested from Plautus onwards. This construction has a well-defined semantics, being restricted to predicates with an experiential or benefactive/malefactive meaning. However, although the construction sometimes shows an inanimate referent in the argument slot triggering dative case marking, we noted that such cases are rare and that this argument slot shows a strong preference for animate referents. In some respects, the ‘double dative’ construction shows a distribution similar to that of simple predicates with a nominative-marked and a dative-marked argu-ment, something which suggests that the double dative construction may be regarded as a subtype of a more general STIMNOM-EXP/BENDAT construction which includes both simple and complex predicates.

As regards the relative syntactic status of the two arguments in the ‘double dative’ con-struction, it was noted that both the nominative-marked and the dative-marked argument show a syntactic behaviour characteristic of prototypical, nominative-marked subjects but that none of them show all of the relevant properties and that they even share some behav-ioural properties. This observation can be variously explained. First, one might simply con-clude that behavioural properties are irrelevant for identifying syntactic subjects in Latin and that one should regard morphological criteria, notably nominative case and verb agree-ment, as decisive in this respect. Another possibility would be that the double dative con-struction as well as the STIMNOM-EXP/BENDAT construction more generally vacillates with regard to the relative status of the two arguments. Specifically, one might speculate that the apparently inconsistent behaviour of the two arguments in the ‘double dative’ construction is a consequence of the fact that it is subject to two competing restrictions. One is morpho-logical in nature and ultimately language-specific, reflecting the general tendency of sub-jects to be ascribed nominative case marking and to trigger verb agreement in Latin, some-thing which would explain why the nominative-marked stimulus argument shows a behav-iour which to some extent corresponds to the behavior shown by prototypical subjects. The other is semantically based and may be hypothesized to originate from universal principles

On the semantics and syntax of the Latin ‘double dative’ construction 49

of argument selection according to which two-place predicates tend to select arguments re-sembling prototypical agents as syntactic subjects (cf. Dowty 1991 for discussion). As ani-macy is generally considered a central element in the complex prototypical notion of agen-cy, the fact that the argument slot in the ‘double dative’ construction which triggers dative case marking shows a strong preference for animate referents might somehow explain the fact that dative-marked experiencers and benefactives/malefactives to some extent show a syntactic behaviour characteristic of subjects rather than objects. Needless to say, the re-sults presented here are far from sufficient to arrive at a conclusive decision between these two explanations and future research will show to what extent the approach outlined in this paper represents a fruitful way of dealing with subjecthood in languages like Latin.

BibliographyBibliographyBibliographyBibliography

Baños Baños, José Miguel. 2003. “Paenitet y los verbos impersonales de sentimiento en latín: sintaxis y pragmática del acusativo personal”. In PRAEDICATIVA. Estudios sobre complementación

verbal en griego y latín (Anejos de la revista Verba 53), ed. by José Miguel Baños Baños, Concepción Cabrillana Leal, M. Esperanza Torrego Salcedo, and Jesús de la Villa Polo. Santiago de Compostela.

Barðdal, Jóhanna, and Thórhallur Eythórsson. 2003. “The change that never happened: the story of oblique subjects”. Journal of Linguistics 39:439–472.

Bauer, Brigitte L. M. 2009. “Word Order”. In New Perspectives on Latin Syntax. Vol. 1: Syntax of the

Sentence, ed. by Philip Baldi and Pierluigi Cuzzolin. Berlin / New York. Benveniste, Émile. 1960. “‘être’ et ‘avoir’ dans leurs fonctions linguistiques”. BSL 55:113–134. ———. 1966. Problèmes de linguistique générale I. Paris. Blume, Kerstin. 1998. “A contrastive analysis of interaction verbs with dative complements”.

Linguistics 36/2:253–280. Dahl, Eystein. 2009. “Some semantic and pragmatic aspects of object alternation in Early Vedic”. In

The Role of Semantics and Pragmatics in the Development of Case, ed. by Jóhanna Barðdal and Shobhana Chelliah. Amsterdam, 23–55.

———. In preparation. Studies in the syntax of subjecthood in Archaic and Classical Latin. Dahl, Eystein, and Chiara Fedriani. 2012. “The Argument Structure of Experience: Experiential Con-

structions in Early Vedic, Homeric Greek and Early Latin”. TPS 110/3:342–362. Devine, Andrew M., and Laurence D. Stephens. 2006. Latin Word Order. Structured Meaning and

Information. Oxford. Dowty, David. 1991. “Thematic Proto-Roles and Argument Selection”. Language 67:547–619. Eythórsson, Thórhallur, and Jóhanna Barðdal. 2005. “Oblique Subjects: A Common Germanic Inheri-

tance”. Language 81:824–881. Fedriani, Chiara. 2009. “The ‘Behaviour-Before-Coding’ Principle: Further Evidence from Latin”.

Archivio Glottologico Italiano 94/2:157–184. ———. 2012. Experiential Constructions in Latin. A Synchronic and Diachronic Study. PhD disser-

tation, University of Pavia. Grimm, Scott. 2005. The Lattice of Case and Agentivity. MSc Diss., University of Amsterdam. Hofmann, Johann Baptist, and Anton Szantyr. 1972. Lateinische Grammatik. Zweiter Band: Syntax

und Stilistik. München. Keenan, Edward L. 1976. “Towards a universal definition of ‘subject’”. In Subject and topic, ed. by

Charles N. Li. New York, 303–334. Kühner, Raphael, Carl Stegmann, and Andreas Thierfelder. 1976. Ausführliche Grammatik der latei-

nischen Sprache. Zweiter Teil: Satzlehre. Erster Band. Darmstadt. Luraghi, Silvia. 1997. “Omission of the direct object in Classical Latin”. IF 102:239–257. Pieroni, Silvia. 2007. “Soggetto e riflessivo”. In Morfosintassi latina. Punti di vista, ed. by Nunzio La

Fauci and Silvia Pieroni. Pisa, 27–39.

50 Eystein Dahl

———. 2010. “Deixis and anaphora”. In New Perspectives on Historical Latin Syntax. Vol. 3: Con-

stituent Syntax: Quantification, Numerals, Possession, Anaphora, ed. by Philip Baldi and Pierluigi Cuzzolin. Berlin / New York, 389–501.

Perseus Greek and Roman Collections. Tufts University. http://www.perseus.tufts.edu/hopper/ collection?collection=Perseus:collection:Greco-Roman.

Primus, Beatrice. 1999. Cases and Thematic Roles: Ergative, Accusative and Active. Tübingen. Tsunoda, Tasaku. 1985. “Remarks on transitivity”. Journal of Linguistics 21:385–396. Van Valin, Robert D. 2005. Exploring the Syntax-Semantics Interface. Cambridge.

Eystein Dahl Department of Linguistic, Literary and Aesthetic Studies University of Bergen PO Box 7800, 5020 Bergen Norway [email protected]