Symposium organized by the Linguistic Association of Finland

Helsinki, November 14-15, 2003


1. Plenary sessions
2. Section papers
3. Posters


Christian Lehmann

University of Erfurt

Cognitive roles, participant roles and syntactic relations

Participant relations are analyzed at the following levels:


1. Individual grammatical system: main verb and NPs connected with it by syntactic relations, adpositions and other structural means.

2. Typology: predicate and arguments connected with it by participant roles.

3. Cognition: situation core and participants related to it by cognitive roles.

The third level is not properly linguistic, but serves as a tertium comparationis for linguistic strategies analyzed at the two lower levels.

        For example, in the German sentence Rita schnitt dem Kind die Haare 'Rita cut the child's hair', the NP dem Kind is a certain kind of adjunct (dativus sympatheticus) at level 1, a benefactive adjunct at level 2 and both a beneficiary and a possessor at level 3. At level 3, the relations among participants constitute a complex network only part of which is represented linguistically.

        From this point of view, a couple of functional subdomains of participation will be analyzed, including causation, benefaction, external possession and concomitance. Languages differ typologically by their choice as to which cognitive roles they express and which ones they leave to inference. In the above example, German leaves the possessive relation of the child to inference.

        In some participant constellations, the choice of a particular strategy is semantically motivated; in other constellations it is not so motivated, but may still be pursued by a language. For instance, Portuguese uses the above type of construction even in such sentences as descobri-lhe os defeitos 'I discovered his defects', where most languages would rather side with English. A language may be characterized at the typological level by the extension of a strategy over and above its proper functional domain.



Lehmann, Christian & Shin, Yong-Min & Verhoeven, Elisabeth 2000, Direkte und indirekte Partizipation. Zur Typologie der sprachlichen Repräsentation konzeptueller Relationen. München: LINCOM Europa (LINCOM Studies in Language Typologie, 4).

Lehmann, Christian & Shin, Yong-Min & Verhoeven, Elisabeth 2000, "Zur interlingualen Ebene in der typologischen Analyse." Göttinger Beiträge zur Sprachwissenschaft 3:57-71.

Van Valin, Robert D., Jr. & LaPolla, Randy 1997, Syntax. Structure, meaning, and function. Cambridge: Cambridge University Press.

Maria Vilkuna

Research Institute for the Languages of Finland


On expletive subjects and topics in Finnish

In Finnish, sentences without overt subjects are tolerated to a high degree. On the other hand, there is a tendency to fill the subject or topic position with overt material. A third tendency, hardly specific to Finnish, is that elements in the topic position are "light", often pronominal, in conversational language. Although the first of these three tendencies may speak against the existence of expletive subjects, the two other ones make it less surprising. Expletives have been suggested to exist in Finnish in investigations as diverse as Sirelius' overview of certain south-eastern dialects (1894) and the generative treatment by Holmberg and Nikanne (2002). However, they are often expletive topics rather than subjects. The main candidates for this function are illustrated in examples (1) and (2). They are the words se 'it' (1) and sitä (2), a particle originating from and formally identical to the partitive case of se. A third expletive element is tentatively proposed here, viz. tässä 'here', a pronominal adverb, in sentences like (3). Expletives are not obligatory in the examples, although deleting them might call for slight changes in word order. In each case, a referential NP could alternatively occupy the position of the expletive.


(1)   a. Se    on     ollut aika      kylmä   viime aikoina.

            it      has    been rather   cold      lately.

‘It has been rather cold lately.’

        b.Se   nyt   vaan on     tyhmää   maksaa    liikaa.

            it     now just  is      stupid     pay-INF  too.much

‘It's just stupid to pay too much.’


(2)   a. Sitä     voi   joutua  vaikka    millaiseen   onnettomuuteen.

            EXPL  can   get       whateverkind-of-ILLaccident-ILL

‘You can get in any kind of an accident.’

        b.Pitääkö            sitä      tuolla  tavalla ryntäillä?

            must-QUEST  EXPL  that      way     rush.about

‘Do you have to rush about like that?’


(3)   Minunko tässäpitää    kaikki         hoitaa?

        I-QUESThere must    everything

‘Is it me who has to take care of everything?’

The paper will discuss the status of these potential expletives from various points of view, without adhering to the assumptions made within the Chomskyan generative theory, such as those in Holmberg and Nikanne. A short look at these assumptions is in order, however, as it is this line of research that seems to have had most to say on the matter this far. The main focus of the paper is on the following questions: On what grounds can these elements be claimed to be expletives? How can it be shown that they occupy the same position? What do they tell us about the clausal architecture of Finnish, assuming a partially discourse-based field model of the Finnish sentence? What other functions do they have, apart from filling a particular position; do they preserve some of the semantic content of their other uses?



Sirelius, U.T. 1984: Lauseopillinen tutkimus Jääsken ja Kirvun kielimurteesta [A syntactic study of the dialect of Jääski and Kirvu]. Suomalaisen kirjallisuuden seura.

Holmberg, Anders and Urpo Nikanne 2002: Expletives, Subjects and Topics in Finnish. In Svenonius, Peter (ed.) Subjects, expletives, and the EPP. Oxford University Press.


Enoch O. Aboh

University of Amsterdam

On the right peripheral typing morphemes: The case of negation in Gbe (Kwa)


Under Aboh (1999), the Gbe (Kwa) languages of West Africa provide empirical evidence for Rizzi’s (1997) hypothesis that the complementizer system consists of a series of distinct slots that host fronted elements, such as, focus and topic. In this regard, the Gbe languages display discrete free morphemes that mark topicalized and focused constituents and could be thought as the manifestations of such slots (1). In other words, the topic and focus markers encode different functional heads that project to the left edge as components of the complementizer system.


(1)   Kòfík      [Lsì     Gúkm      tn]i      w    [IPék       x      ti

        Kofi    Top  rice      Gukome  Poss     Foc     3sg   go   buy

‘We went to buy the RICE FROM GUKOME’

However, these markers may occur to the right edge, when they take scope over the proposition. In (2), for instance, the focus and insistence markers cluster to the right because they take scope over the proposition. Accordingly, the Gbe markers may occur to the left or right periphery depending on their scope properties. Left peripheral elements take scope over a constituent that is attracted to the relevant specifier position within the complementizer system. Right edge elements, however, take scope over the proposition, which is attracted to a specifier position within the complementizer system, in a sort of predicate fronting (Aboh 2002).


(2)   [IPKòfí         x     lsì   Gúkm      tn]i      w         ti

            Kofi    go    buy rice Gukome  Poss     Foc Ins


Granting this analysis, the question arises how the Gbe right peripheral negation marker fits in the system. The Gbe languages express sentential negation in three ways.

1. The Gungbe-type languages display a preverbal marker —similar to French ne, or Italian non.


(3)   Kjó        x      kátìkátì   l          [Gungbe]

        Kojo   Neg buy  kite          Det

‘Kojo did not buy the kite’

2. The Fongbe-type languages resort to the Gungbe-type preverbal marker (4a), or a sentence-final marker ă—the equivalent of French pas (4b).


(4)   a. Kkú          x     àsn               [Fongbe]

            Koku   Neg  Fut  buy crab Det

'Koku will not buy the specific crab'

        b.Kkú          x     àsn              ă

            Koku   Fut   buy crab     Det Neg

'Koku will not buy the specific crab'

3. The Ewegbe-type languages require the simultaneous occurrence of the preverbal and sentence-final negative markers (5).


(5)   a. Kòfí    *()  ɖù            *(ò)         [Gengbe]

            Kofi    Neg     eat    thing   Neg

'Kofi did not eat'

These facts lead me to conclude that Gbe languages involve both a preverbal and sentential-final position for encoding sentential negation. Following Pollock 1989, Ouhalla 1990, Haegeman 1995, Zanuttini 1997, I propose that the negation markers /encode the negative functional Neg̊, within the inflectional system. On the other hand, the sentence-final negative elements manifest the complementizer system. Unlike the French adverbial negative element pas, the Gbe elements ă/ò express a negative head Neg̊[C] that is located within the complementizer system. Like other left peripheral markers, the negative marker surfaces to the right edge because it takes wide scope over the proposition. Granting the proposed analysis for the right edge markers of the complementizer system, I assume that Neg̊[C] attracts in its specifier, the proposition containing the narrow scope taking preverbal negative /, as shown in (7).


(7)   [CP… [NegP[C] [Neg[ă/ò]…[IP.. [NegP [Neg[má/mú]… ]]]]]]

                                        


This analysis of negative sentences as peripheral leads to a fine characterization of the architecture of complementizer system and sheds a new light on the interactions between negative constructions and certain A’-movements (i.e., fronting rules) that involve focus-, wh-, or topic-phrases.



Aboh, O. Enoch. 1999. From the Syntax of Gungbe to the Grammar of Gbe. Editions à la carte S.A., Sierre, Switzerland.

———. 2002. “Morphosyntaxe de la péripherie gauche nominale.” In Anne Zribi-Hertz & Anne Daladier eds., Recherches linguistiques de Vincennes 31: 9–26, La syntaxe de la definitude 2002. PUV, Paris.

Haegeman, Liliane. 1995. The Syntax of Negation. CUP, Cambridge.

Ouhalla, Jamal. 1990. “Sentential Negation, Relativized Minimality and the Aspectual status of Auxiliaries.” Linguistic Review 7: 183–231.

Pollock, Jean-Yves. 1989. “Verb Movement, Universal Grammar, and the Structure of IP.” Linguistic Inquiry 20: 356–424.

Rizzi, Luigi. 1997. “The Fine Structure of the Left Periphery.” In Liliane Haegeman, ed., Elements of Grammar. Kluwer, Dordrecht, Boston, London.

Zanuttini, Raffaella. 1997. Negation and Clausal Structure: a Comparative Study of Romance Languages. Oxford University Press, New York.

Maria Alm

Lund University


The Pre-Front Field of Spoken German: the Functions of the German Word "also"


The "pre-front field" in German is a controversial notion. The term has evolved from a common way of describing the German sentence structure as consisting of fields holding the clause elements. The first or left-most clause field of the nuclear German clause is considered to be the front field. The pre-front field it is in front of the front field, and it is therefore sometimes claimed that the pre-front field does not belong to the following clause at all.

        The pre-front field is claimed to be associated with certain (pragmatic) functions, which in this talk are discussed on the special case of the German word "also". The prototypical function of "also" is that of an (sentence) adverb. The adverbial function of also relates the propositional contents of its host utterance to the preceding utterance(s) as a consequence of what has been said. However, in spoken German the most the frequent use of "also" is in the pre-front field, where it seems to take on partly or entirely different functions than the functions as an adverb. The question is whether the pre-front field actually gives rise to the special functions of the items in this position, or whether functional changes are actually due to a change in the lexicon. If it is the same item, what difference does a change of position make and how can this change be explained?

        Thim-Mabrey (1985, 1988) holds that also is the same lexical item in the pre-front field and inside the nuclear clause. As for the pre-front field, Thim-Mabrey claims that when occupied by operative expressions like "also" this position per definition forces a metacommunicative interpretation. Auer (1996), on the contrary, argues that the items in the pre-front field have changed their category when moving out of the nuclear clause. His idea is that the pre-front field is a a position where items undergo semantic and syntactic change, a process that is referred to as grammaticalization. The consequence of Auer's model is that "also" developed into a discourse particle when moving into the pre-front field; it underwent semantic bleaching and took on contextually determined meaning instead of the lost lexical one. The result is that there are now (at least) two different lexemes "also": the adverb and the discourse particle(s).

        In this talk, the contribution of the pre-front field to the interpretation of "also" is examined in an authentic conversation. To pin down the differences between the pre-front field and the nuclear clause fields, the functional effect on "also" related to the pre-front field are compared to the functions of "also" inside the clause.


Auer, Peter (1996): "The pre-front field in spoken German and its relevance as a grammaticalization position." Pragmatics 6:3, 295-322.

Auer, Peter (1997): "Formen und Funktionen der Vor-Vorfeldbesetzung im gesprochenen Deutsch." In: Schlobinski, Peter (Hrg.): Syntax des gesprochenen Deutsch. Westdeutscher Verlag, Opladen. 55-97.

Thim-Mabrey, Christiane (1985): Satzkonnektoren wie 'allerdings', 'dennoch' und 'übrigens': Stellungsvarianten im deutschen Aussagesatz. Regensburger Beiträge zur deutschen Sprach- und Literaturwissenschaft. Verlag Peter Lang; Frankfurt am Main.

Thim-Mabrey, Christiane (1988): "Satzadverbialia und andere Ausdrücke im Vorvorfeld." Deutsche Sprache 16, 52-67.


Jóhanna Barđdal        &     Thórhallur Eythórsson

University of Bergen           University of Manchester

Oblique Subjects: A Germanic Inheritance!

In this paper we propose that non-nominative “logical subjects” of impersonal predicates in Modern High German exhibit some behavioural properties of syntactic subjects, similarly to Modern Icelandic. The grammatical relations in impersonal predicates in Icelandic are widely considered to be the converse of those in comparable constructions in German (cf. Andrews 1976, Zaenen, Maling and Thráinsson 1985 and many others). Thus, it is assumed that in Icelandic clauses of the type in (1a) the preverbal oblique NP (“me”) is the subject and the nominative (“the shark”) is the object, whereas in German the grammatical relations are the other way around.


(1)   a. Mér     líkar        þessi    hákarl.


        b.Mir      gefällt      dieser  Haifisch.


            ‘I like this shark.’


In other Germanic languages, such as Modern English and modern Mainland Scandinavian, impersonal predicates have been ousted. To the extent that morphological case is found at all in these languages (with pronouns), the subject is always in nominative case and the object in oblique case. It is a common view in Germanic syntax that the situation in German, exemplified in (1b), represents a more “original” state of affairs than the Icelandic one in (1a). This view has been spelled out most explicitly in an influential paper by Cole, Harbert, Hermon & Sridhar (1980) (cf. also Askedal 2001, Croft 2001, Faarlund 2001, Haspelmath 2001). Cole et al. (1980) propose that the non-nominative ?logical subject? of impersonal predicates has undergone a change from being a syntactic object to being a syntactic subject in Germanic, as well as in a number of other language families. We refer to this proposal as the Object-to-Subject hypothesis. Cole et al. (1980) postulate that at the earliest stage (A) the “logical subject” was a syntactic object which gradually acquired subject properties: first behavioural properties (stage B), e.g. control of reflexivisation and omission in second conjuncts and infinitives, and eventually coding properties (stage C), e.g. morphological case (nominative) and subject-verb agreement. Gothic and Modern High German are taken to represent stage A, while Modern Icelandic is at the intermediate stage B. The final stage C in the development is represented by Modern English and modern Mainland Scandinavian.

        We argue that much of the evidence adduced in favour of the Object-to-Subject hypothesis has not been properly interpreted. In particular, we demonstrate that the standard analysis of Gothic and Modern High German provides insufficient support for the view that the “logical subject” of the impersonal construction was a syntactic object, since it is based on only a small subset of relevant data. Crucially, we present novel data from German,

illustrating that the non-nominative “logical subject” of impersonal predicates can be omitted in control infinitives under identity with a preceding subject. This property is confined to syntactic subjects, and is by many linguists taken to be the most reliable test for subjecthood. Thus, our examination reveals that “logical subjects” of impersonal predicates exhibit some behavioural properties of subjects in Modern High German, similarly to Modern Icelandic.

        As against the Object-to-Subject hypothesis, we propose the Oblique Subject hypothesis, according to which “logical subjects” of impersonal predicates were syntactic subjects already in Old Germanic. We discuss data from a range of Germanic languages, including Old and Modern Icelandic, Modern Faroese, Old Swedish, Old and Early Middle English, and Modern High German. This evidence supports the Oblique Subject hypothesis and undermines the validity of the Object-to-Subject hypothesis. We conclude that there is no empirical basis for assuming a development from syntactic objects to syntactic subjects (stage A to B in the terminology of Cole et al. (1980)) in the history of Germanic.


Olivier Bonami          &     Daniele Godard

Université Rennes 2            CNRS-Universite Paris 7

Incidentally, there are (almost) no adjuncts in the left periphery in French


It is common for a modifier, in particular an adverb, to be linearly ordered at the left periphery of the sentence. (1-4) illustrate the four kinds of such cases found in French. We argue that of these four cases, only one, namely (4), is a genuine case of adjunction to a clause. We provide an analysis of the data in a version of HPSG with a specific linearization component (see Reape 1994, Kathol 2000).


(1)   Lentement/heureusement, Paul s’est endormi.

        'Slowly/fortunately Paul fell asleep.'

(2)   Mechamment, elle a repondu, la fille !

        'That girl answered nastily!'

(3)   Heureusement que Paul arrivera a l’heure.

        'Fortunately, Paul will arrive on time.'

(4)   Aussi arrivera-t-il a l’heure.

        'Thus he will arrive on time.'


(1) illustrates the case of an incidental adverb (IA). We call 'incidentals' expressions with a distinct intonation setting them apart from the rest of the sentence. Incidental intonation is not limited to modifiers (see e.g. left-topicalized arguments : 'Aux secretaires, le patron parlera' To the secretaries, the boss will speak). For adverbs, it is not limited to the left periphery : every IA can be linearized in any of the 6 positions indicated in (5). Adverbs with an 'integrated' intonation have a much more limited distribution.


(5)   IA Paul IA a IA explique IA le probleme IA a sa sœur IA.

All incidentals share a common property which suggests generation in a single position (contra e.g. Cinque 1999): they must outscope all nonincidental material (except quantifiers), and they do not obey the scope-order correlations found with nonincidental adverbs. We analyze incidentals as a special kind of adjuncts to the VP, which can be linearized anywhere in the verbal domain.

        Turning to (2), we show that this construction has two peculiar properties: the adverb can be an argument (6), and it can scope in an embedded clause (7). This is quite surprising if the adverb is an adjunct to the sentence, but it falls out directly if it is a filler in a filler-gap dependency.


(6)   Mechamment, elle les a traites, ses invites.

        She treated her guests nastily'

(7)   Mechamment, elle m’a dit qu’il avait repondu

        'She said that he answered nastily'

The construction in (3) is reserved to an small, idiosyncratic class of adverbs. We argue that in this construction the adverb is best thought of as the head of the sentence, taking the que-marked clause as its complement. This accounts directly both for the lexical limitation (which is simply a subcategorization idiosyncrasy) and for the presence of the complementizer, which is otherwise quite surprising, since root clauses lack complementizers in French.

        This leaves the construction in (4), where the presence of an initial adverb licenses subject clitic inversion, as the only case of adjunction to the left of the sentence in French. This construction is lexically restricted to a handful of adverbs, setting the number of possible adjuncts in the left periphery to a bare minimum.

Ernestina R. Carrilho

University of Lisbon

Expletives beyond the subject in a Null Subject Language


It has often been observed that languages allowing for null referential subjects (e.g. Italian) do not have overt expletives (Rizzi 1982, 1986, a.o.). This long-standing empirical correlation is in line with a well-known generalization about null subjects in natural language: conditions on the licensing of null subjects are less strict for expletive subjects than for argumental ones. This accurately accounts for the fact that expletives may be non-overt not only in so-called null subject languages, but in some non-null subject languages as well (e.g. Icelandic, German — see Platzack 1996 a.o.). A different type of languages regarding null subjects requires both expletive and referential subjects to be overt (cf. English). Accordingly, in different instantiations of the generative framework (Chomsky 1981 and subsequent work), overt expletives in languages such as English have traditionally been analyzed as a structural linguistic device strictly connected with some visibility requirement on the subject position.

        There is however increasing evidence that at least some non-standard (regional or colloquial) varieties of null subject languages exhibit optional overt expletives (e.g., among Romance languages, Spanish, Catalan, Galician, European Portuguese (EP, henceforth) — Bosque & Demonte 1999, Solà et al. 2002, Alvarez et al. 1986, Carrilho 2001). While unexpected, this combination of properties is still poorly understood. Optionality on the visibility condition on the subject could be tentatively invoked to explain the option between overt and null expletives in such languages. An alternative would consist in weakening the connection between overt expletives and subjects, an approach which will be pursued in this paper.

        On the basis of data extracted from a dialectal corpus of spoken EP, it will be suggested that the presence of the overt expletive in non-standard EP is tightly connected with discourse-related conditions. It will be shown that the expletive ele extends over an extremely wide spectrum of constructions, which goes far beyond impersonal constructions (1).

1. ele      estava  a          nevar

    EXPL was      PREP  snowINF

    ‘It was snowing’


        In particular, it will be shown that expletive ele behaves in a significant part of the corpus as an unambiguously peripheral element (2-4).

2. ele      a      fome   não   havia

    EXPL the   hunger not   had

    ‘As for hunger, it didn’t exist’

3. mas ele       aquilo   foi   feito

    but  EXPL  that       was done

    ‘but that was done’

4. ele      o      animal o  que      quer    é   fugir

    EXPL the   animal     what    wants  is  escapeINF

    ‘As for the animal, what it wants is to escape’

Close inspection will be devoted to the peripheral status of the expletive (specially w.r.t. other peripheral constituents in EP).

        The distribution of the peripheral expletive, together with the consideration of its discourse effect, will support the proposal that overt expletives in EP depend on discourse motivated conditions involving the left periphery (namely the loci for encoding the force of the sentence and the anchoring of information in discourse).


Álvarez, R., X. Regueira, & H. Monteagudo (1986) Gramática Galega. Vigo: Galaxia.

Bosque, I. & V. Demonte. 1999. Gramática Descriptiva de la Lengua Española. Espasa, Madrid.

Carrilho, E. 2001. Expletivos do Português europeu em foco: a evidência dos dados dialectais. in Actas do XVI Encontro Nacional da Associação Portuguesa de Linguística, September 2000. APL. Lisbon. 

Chomsky, N. 1981. Lectures on Government and Binding. Foris, Dordrecht.

Platzack, C. 1996. Null Subjects, Weak Agr and Syntactic Differences in Scandinavian. in H. Thráinsson, S. Epstein & S. Peter (eds.). Studies in Comparative Germanic Syntax.

Rizzi, L. 1982. Issues in Italian Syntax. Foris, Dordrecht.

Rizzi, L. 1986. Null object and the theory of pro. Linguistic Inquiry 17.

Solà, J., M.R. Lloret, J. Mascaró & M. Sadanya. 2002. Gramàtica del Català Contemporani. Empúries, Barcelona.

Pao-Yu Chen

Graduate Institute of Linguistics, National Taiwan University

Focus System in Kavalan: A Pragmatic Perspective

This is the first study that aims to explore the focus system in Kavalan in a pragmatic perspective. The significance of this study provides a closer understanding of the relationship between pragmatic condition and the focus system in Kavalan. The theories of discourse analysis and quantitative discourse analysis are adopted to help the analysis. The data, which comprises three Frog stories and three Pear stories for 20minutes and 18 seconds with a total of 418 IUs, were from three native speakers of Kavalan. An examination of word order, lexical transitivity, discourse transitivity, and topicality found that no pragmatic difference appears to underlie the choice of focus forms.

Tanya Christensen

Roskilde University

On the borderline of syntax: the importance of topology in word order languages CANCELLED


Even though syntax at first sight is a very linear matter of words following words, most researchers analyze and present syntactic relations in the form of trees, that is, layered or hierarchical structures with branches and nodes. But in languages where word order is densely coded, as it is in Danish (and other Scandinavian languages, and probably also in German), the linearity of the sentence is of massive importance, and should not be overlooked.

        This paper advocates an approach to the analysis of syntactic functions that allows topology (the study of “the places” (topoï) of a sentence, i.e. its positions) to play a major part in the analysis of (different aspects of) sentence meaning. It builds upon the appreciation of Danish grammarian Paul Diderichsen (1941, 1946) that Danish sentences can be divided into several positions which not only hold quite specific types of constituents, but also add extra meaning to the syntactic elements placed there. The point is, that the same material, the same word, can be placed in different positions and acquire a different meaning — though sometimes it seems that it is the meaning of the entire sentence, more than the meaning of the individual word, which has changed.

        One example of this relates to the pragmatic level, more specifically to information structure. It is possible to place the Indirect Object (syntactic level) in two positions: its ordinary, unmarked position between the Non-finite Verb and the Direct Object, or in First Position (topological level).








Sentence adverbs



Non-fin. Verb

Indirect Object

Direct Object
































    a. ‘She has given him the book’

    b. ‘It was to him that she gave the book’


Sentence (b) will often attain an emphatic reading in accordance with the more marked word order.

        Another example concerns the semantics of adverbs/particles. Many linguists have noted that adverbs or particles may be homonymic or functionally Janus-headed (Jespersen), but very few have realized that the different meanings correlate with different positional options (but see Jensen 2000, Hansen & Heltoft (forthcoming)):








Sentence adverbs



Non-fin. Verb














i Paris










i Paris










in Paris

    c. ‘He probably doesn’t live in Paris’

    d.‘He doesn’t live safely in Paris’


Generally, it is the case that the Danish adverbs/particles acquire modal meaning when placed in the regular SA-position between Subject and Negation in the Danish topological sentence structure, whereas they retain their historically prior, more concrete and “objective” meaning in 1st position (e.g. nu in 1st position: ‘now’, in SA-position: ‘not-true-what-was-said-before’).

        Whether the syntactic function of a unit remains intact or changes, the topological level is necessary to explain the effects of word order changes in languages such as Danish, I argue.




Diderichsen, Paul (1941): Sætningsbygningen i Skaanske Lov. [The Sentence Formation of the Scanian Law] København: Munksgaard

Diderichsen, Paul (1946): Elementær Dansk Grammatik. [Essentials of Danish Grammar [1964]] Køøbenhavn: Nordisk Forlag

Hansen, Erik og Lars Heltoft (forthcoming): Grammatik over det Danske Sprog [Grammar of the Danish Language]

Jensen, Eva Skafte (2000): Danske sætningsadverbialer og topologi i diakron belysning. [Danish Sentence Adverbials and Topology in a Diachronic Perspective] Ph.d.-dissertation.

Michael Daniel


Benefactive dative admissibility: making things uneasier. Evidence from Balkar. CANCELLED

The view of the verbal valency as a category “licensing” some arguments and banning others meets several remarkable obstacles. The two most well known cases are, first, verbs that may be used both transitively and intransitively, or labile verbs (e.g. [Haspelmath 1993; Kulikov 1999]), and, second, external possession constructions which apparently “license” an extra-argument [Payne&Barshi 1999]. Although both phenomena (especially labile verbs) cry out for further typological study, I will focus on a not less challenging but even less studied phenomenon of the enigmatic conditions that license appearance of a benefactive (or malefactive) argument, commonly coded by dative. The reason why for instance Russian otkroj mne dver ‘open the door for me’ sounds better than zakroj mne dver ‘close the door for me’ has few in common with the verb and its lexical meaning. Apparently, it is the situation described by the sentence rather than the verb itself that licenses the dative; practically, dative admissibility seems to base primarily on the force of individual native consultants’ imagination. This provides a convenient master-key explanation of benefactive datives — they are licensed conceptually and thus the issue of their admissibility is extra linguistic. This explanation is however false in several ways. To start with, [Shibatani 1996] makes a first step to show that languages vary in dative admissibility for the same real world situation. Then, whether we consider this variation to be a result of cultural or grammatical differences (or both), we better register dative admissibility constraints in the grammatical description. Second, even within individual languages there may be various benefactive constructions (verb + Dative vs. verb + benefactive serializer + Dative; Dative vs. benefactive adposition; etc.) which manifest various, more strict/loose constraints on benefactive admissibility. Last, benefactive semantics are problematic eo ipso. Do we have to distinguish between benefactivity and affectedness? Are we supposed to classify e.g. external possessors — indisputably affected participants — as benefactives? Whether we do and we are, or not, we need to be ready that dative may behave differently with various verbs, not because they are different verbs and reflect different situations, but because these are functionally different datives which obey different admissibility constraints. Consider Balkar.

        The issue of the admissibility of the Balkar benefactive dative puzzles from the very first moment of the study, producing the same unordered and non-systematic impression as reported by Shibatani in the lower part of his benefactive scale. The picture drastically changes and becomes much more systematic at the moment when we split Balkar datives into several types and consider them separately in interaction with some other categories — possession and creativity. Below I have summarized basic results; for convenience, I compare dative (in)admissibility with Russian and English:

        1. open the door for me type — no relation between the dative argument (beneficiary) and the patient — admissibility only depends on the conceptualization of the situation. Malefactive reading is impossible. If the dative is not admitted, a postposition is used. Differences between dative admissibility in Balkar, Russian and English may be viewed in terms of Shibatani’s scale of cross-cultural / cross-linguistic variation.

        2. paint my fence type — patient is possessed by the dative argument (external possessor ~ affected participant ~ beneficiary). The more inalienable is the possession relation, the less admissible is the dative. If the dative is not allowed, the participant is expressed as (internal) possessor. The difference from English is that the Balkar dative is allowed with alienable possessions (English Paint me a house); the difference from Russian is that with inalienable items the Balkar dative is out (Russian Postrigi mne(I.DAT) nogti ‘Cut my nails’)

        3 build me a house — the patient is created to be owned by the dative argument (prospective possessor ~ beneficiary). The dative seems to be always allowed with creative verbs, probably by attraction of the bitransitive pattern. Malefactive reading (negative impact on the prospective possessor) is possible. The difference from English and Russian is that even some intransitive verbs allow dative (*Mne vyros zub = *’A tooth grew to me’ = maNa tiS cyqdy).



Haspelmath, Martin. More on typology of inchoative / causative verb alternations. // Comrie and Polinsky eds. Causatives and Transitivity. Benjamins, 1993.

Kulikov, Leonid. May he prosper in offspring and wealth ... // Rakhilina and Testelets eds. Typology and Linguistic Theory. From description to explanation. Languages of Russian Culture. Moscow, 1999.

Payne and Barshi 1999. External Possession. Benjamins, 1999.

Shibatani, Masayoshi. Applicatives and benefactives. A cognitive account. // Shibatani and Thompson eds. Grammatical constructions: their form and meaning. OUP, 1996.

Bernard De Clerck

Ghent University


The syntactic and pragmatic analysis of let’s in present-day British and American English


Due to processes of grammaticalization involving semantic bleaching and phonological reduction , let’s has semantically and syntactically drifted away from the second person imperative let + us. This has led to a number of conflicting views with regard to the status of let and let’s. Seppänen (1977) treats let as a hybrid modal auxiliary, Costa (1972) still treats it as the imperative of the full lexical verb, while Fries (1952) identifies let’s as the only member of a separate class of function words. More recent descriptive grammars, Quirk et al. (1985) and Biber et al. (1999), describe let’s as an invariant or unanalysed pragmatic particle which functions as a first person plural imperative, proposing a joint action to be undertaken by the speaker and the hearer(s). Their observations are on a par with Joseph (1992), Hopper and Traugott (1993), Krifka (2001), who treat let’s as a monomorphemic constituent with an adhortative function. This particle-like status is corroborated by examples such as Let’s don’t, Let’s us, Let’s you and I, coming from AE (Quirk et al. 1985). Huddleston and Pullum (2002), however, are more hesitant towards treating let’s as a particle in varieties of English (including BE) which adopt a less lenient attitude towards these constructions. In these varieties, where these uses of let’s do not seem to occur, they say that “there is no compelling reason to suggest that there has been a reanalysis of the syntactic structure” (Huddleston and Pullum 2002: 935). In such varieties, Let is still regarded as a catenative verb - be it semantically bleached and partly fossilised in its syntax - used with an NP object (us or ‘s) and a bare infinitival clause as second complement. 

        In this paper, the observations made by Huddleston and Pullum will be tested against attested uses of let’s in AE and BE. On the basis of corpus analysis (BNC, ICE-GB, Santa Barbara Corpus of Spoken American English) and web searches, it will become clear that there are indeed striking differences between the two varieties, caused by a further development of let’s as a particle in AE. Examples such Let’s you and him make peace, Let’s they and them fight in AE seem to indicate that in some cases, the use of let’s as a particle has even been extended beyond first person plural subjects, not including the speaker in the proposed action. Even more striking are ambiguous examples such as Let’s us build your website, Let’s us know what business you are in. The meaning of these examples is not very definite and seems to hover over and between a requestive, suggestive or causative reading. Similar examples have not been found in BE. This paper further explores these “problematic” uses and focuses on the consequences for the syntactic and pragmatic analysis of let’s. Are these uses to be seen as confusion-induced mistakes or should they be treated as instances of an even further grammaticalized particle let’s, functioning as a mere adhortative?




Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan (1999). The Longman Grammar of Spoken and Written English. London: Longman.

Fries,. C.C. (1952). The Structure of English: an Introduction to the Construction of English Sentences. New York: Harcourt Brace.

Costa, R.M. (1972). Let’s solve ‘let’s’. Papers in Linguistics 5. 141-144.

Hopper, P & E. Clos Traugott (1993). Grammaticalization. Cambridge: cambridge University Press.

Huddleston, R. and G.K. Pullum (2002). The Cambridge Grammar of the English Language. Cambridge: University Press.

Joseph, B. (1992). Diachronic explanation: putting speakers back into the picture. In Davis, G and Iverson,G. (Eds.). Explanation in Historical Linguistics. Amsterdam: Benjamins. 123-144.

Krifka, M. (2001). Hand-out from university course on Lexikalische Semantik at Institut für Deutsche Sprache und Linguistik, Humboldt-Universität zu Berlin.

Quirk, R., S. Greenbaum, G. Leech and J. Svartvik (1985). A comprehensive grammar of the English language. London: Longman.

Seppänen, A. (1977). The position of let in the English auxiliary system. English Studies 58. 515-529.

Ricardo Etxepare

CNRS (UMR5478, Bayonne)

On the grammatical realisation of speech events: quotative que in Spanish

In colloquial speech, main clauses in peninsular Spanish can be headed by an overt complementizer (Spitzer, 42; Porroche Ballesteros, 95; Etxepare, 02):


(1)   Oye(que)   el     Madrid        ha   ganadola     Champions

        hey,that      the   Real madridhas  won     the   Champions

The apparent optionality of the complementizer masks an important semantic difference, however: whereas (1) w/o the complementizer constitutes an assertion whose propositional content is just that Real Madrid won the Champions, (1) with the complementizer contributes the additional meaning that someone else, who is not the speaker, said that Real Madrid won the Champions. In other words, que introduces reported speech in matrix contexts. This is not otherwise an inherent function of que, which is associated to different types of dependencies in Spanish (see Lopez García, 99). Now consider (2):


(2)   a. Siviene   tu    madre,     el    tabaco     estuyo

            if comes  yourmother,    the  tobacco   is yours

        b.Siviene   tu    madre,     que el     tabaco     es tuyo

            if comes  yourmother,    that the   tobacco   is yours

        Imagine the following situation: two teenagers are secretly smoking in a room. Fearing that his mother could show up and find out, one of them says (2a). By saying that, the speaker asks the other person to act as if the tobacco was his or hers. By saying (2b) the speaker asks more than just pretense: he/she asks the other person to say that the tobacco is his /hers. This is not, strictly speaking, reported speech. Both (1) with que, and (2b) can be accounted for together under the hypothesis that matrix complementizers are associated to an underlying speech eventuality. This speech eventuality can be referred to (Asher, 93; Stirling,93) and modified and quantified over (Parsons, 90). Consider in this regard the contrasts in (3a-b), (4a-b), and (5):


(3)   a. Siviene   mi   madre,     el    tabaco     estuyo,    y     *educadamente

            if comes  my  mother,    the  tobacco   is yours,  and politely

        b.Siviene   mi   madre,     que el     tabaco     es tuyo,   y      educadamente

            if comes  my  mother,    that the   tobacco   is yours, and  politely


(4)   a. Si viene mi madre, el tabaco es tuyo, *y   lo haces   educadamente

                                                                        and  cl do        politely

        b.Si viene mi madre, que el tabaco es tuyo, y lo haces educadamente


(5)   a. Siempre  que  el      Madrid   esel     mejor, qué  pelma

            always    that  the    Madrid   is the   best,    how boring

"Always (saying) that Real Madrid is the best, how boring"


        The present paper tackles the underlying syntactic structure of quotative constructions in Spanish. It argues that they come in two types: one of them incorporates a full VP, with an underlying verb of communication (à la Ross, 70), and supports overt thematic arguments:


(6)   Tu      padre    que cuándovienes

        Your   father   that when   you-come

"Your father saying "when are you coming?"

The other is a nominal structure, with the semantic import of an indefinite description, which contributes an existential quantification over variables referring to utterances (as in Lahiri, 02). In both cases, que has the function of subordinator (Bhaat&Yoong, 89; Szabolcsi, 94). The analysis defended here also has as a consequence a radical revision of so called "double comp" structures in Spanish (Lahiri, 91,02;Plann, 82;Rivero,95;Suñer, 91,94; Uriagereka,88).

Jocelyne Fernandez-Vest

Centre National de la Recherche Scienfique (C.N.R.S.) & Université Paris III-Sorbonne Nouvelle

Are discourse particles peripheral to syntax?


Discourse Particles (DIPs) were still some decades ago the " personae non gratae " of syntactic and stylistic theories. They have recently been rehabilitated and owed a considerable amount of attention in General Linguistics, as well as in the description of rare languages.

        Does this imply that their status is finally settled ?

        In fact, the features retained as relevant for defining DIPs still differs largely from one approach to the other, e.g.

        o linguistic - DIPs are invariable, often homonyms of other constituents (conjunctions, adverbs), syntactically optional ;

        o pragmatic - a universal prototypical definition is based on structural and functional criterai (DIPs lack a propositional meaning, they qualifiy the discourse process rather than the utterance structure, they anchor the speaker's messages in her attitudes/feelings implicitely).

        I will show that the felicity conditions for a study of DIPs include their investigation 1/ primarily in spoken language 2/ preferably in languages with no or little written tradition, 3/ framed by a preliminary distinction between an informative and a morpho-syntactic level 4/ with some insight in their variation in different genres and contexts.

        Examples will be taken from a few Uralic languages, mainly Northern Sami (from oral tradition to written typological change) and Finnic languages, compared with two Indo-European " world " languages - French and English.

        I will claim that, although DIPs belong, as a device for Information Structure, to the informative-enunciative level of discourse, they play also an important structural role : within oral utterances, DIPs regulate and modulate, together with prosody, the rythm and order of constituents In this sense, their function is not peripheral but central to language systems and their syntax.



Fernandez-Vest (M.M.Jocelyne), 1987, La Finlande trilingue, 1 - Le discours des Sames - Oralité, contrastes, énonciation, Paris, Didier Erudition, 990 p. //

1994, Les particules énonciatives dans la construction du discours, Paris, PUF, coll. Linguistique nouvelle, 283 p. //

2000, " Déixis, interaction, grammaticalisation : le cas des particules énonciatives en same du Nord et en finnois ", in Grammaticalisation aréale et sémantique cognitive: les langues fenniques et sames, - Areal Grammaticalization and Cognitive Semantics: the Finnic and Sami Languages, M.M.J. Fernandez-Vest (dir.), Actes du Colloque International du C.N.R.S. tenu les 9 et 10 avril 1999 en Sorbonne, Tallinn, Eesti keele sihtasutus, Oural-Ural, 1, 65-80. //

2001, " Cognitive processes and the construction of bilingual meaning: the case of Finnish spoken in California ", FU 9. Congressus Nonus Internationalis Fenno-Ugristarum, Tartu 7-13.8.2000, Pars IV, 237-246. //

2002, " Typology, pragmatics, standardization : will Northern Sami survive as Finno-Ugric ? ", in Finno-Ugrians and Indo-Europeans : Linguistic and Literary Contacts, Proceedings of the Symposium at the University of Groningen, November 22-24, 2001, R. Blokland & C. Hasselblatt (eds.), Studia Fenno-Ugrica Groningana 2, 64-71. //

2003, "Information structure and typological change : Northern Sami challenged by Indo-European models ", in M.M.J. Fernandez-Vest (dir.), Les langues ouraliennes aujourd'hui : approche linguistique et cognitive - The Uralic Languages to-day : a cognitive and linguistic approach, Préface de Claude Hagège, Paris, Ed. Honoré Champion, Bibl. de l'Ecole des Hautes Etudes à la Sorbonne, 563-576.)

Ekaterina Gruzdeva

University of Helsinki

On syntactic definition of finiteness: evidence from Nivkh

The traditional view on the category of finiteness is based on morphological distinctions of verb forms underlined by their syntactic (functional) properties. However, for some languages, in particular for Nivkh (Paleosiberian, isolate), whose data is discussed in the paper, the morphological treatment of this category seems to be not completely valid.

        In Nivkh, both "morphologically" finite and nonfinite verb forms may agree or not agree with the subject, may be inflected or not inflected for TAM categories. These facts lead to the conclusion that with regard to Nivkh data finiteness should be considered somewhat differently. The paper suggests the syntactic interpretation of finiteness defined in terms of the syntactic functions fulfilled by verb forms, that allows to draw quite a clear distinction between various types of verb forms in Nivkh.

        According to the syntactic approach to finiteness, the principal distinction is drawn between the syntactic characteristics of verb forms that may be supported by their morphological properties. Finiteness is therefore considered as the possibility of a verb form of being used as the only or the matrix predicate, regardless of markedness for different morphosyntactic categories. Unfiniteness is respectively understood as the impossibility of a verb form to appear in these syntactic functions.

        Moreover, the suggested approach implies the existence not only of a set of two absolute values, “finite” vs. “nonfinite”, but of a hierarchy of finiteness. The category is treated as a scale including three areas, i.e. two polar areas, “finiteness” and “nonfiniteness”, and an in-between area. The last area in turn represents a continuum whose members share the features of two polar areas and gravitate towards one or another polar. The degree of closeness / remoteness of a particular verb form from finite or nonfinite polar is determined by its prototypical, i.e. most common and textually frequent, and nonprototypical, i.e. less common and textually frequent, syntactic functions.

        Basing on syntactic interpretation of finiteness, verb forms in Nivkh can be divided into three classes:

        (1) "Pure" finite verb forms occur as predicates in single-predicate sentences or in matrix clauses of multi-predicate sentences (e.g. Nivkh negative indicative, preventive, interrogative and modal indicative forms).

        (2) "Pure" nonfinite verb forms occur as predicates in nonmatrix clauses of multi-predicate sentences (e.g. Nivkh coordinated forms) or as nonpredicates in all types of sentences (e.g. Nivkh participles and supines).

(3) In-between verb forms embrace the following two types:

(i) Verb forms that are prototypically used as finites and nonprototypically as nonfinites (e.g. Nivkh imperative and positive indicative forms).

(ii) Verb forms that are prototypically used as nonfinites and nonprototypically as finites (e.g. Nivkh converbs).

        In general, a scalar three-way syntactic representation of finiteness seems to be useful for analysis of verb forms not only language-specifically but also cross-linguistically. It is one of the possible solutions for getting over difficulties arising, first of all, on analysis of problematic forms and marginal cases that are widely attested in languages of different types.


Irmeli Helin

University of Helsinki

Evidentiality in Finnish and German


Evidentiality is a grammatical and lexical means for the speaker or writer to express that the source of information is somebody other than the sender of the message him/herself. The usual assumption is that there is no grammatical means in the Finnish language to express evidentiality except, maybe, the potential, which has other modal functions as well. It can also be used by the speaker to express that he or she "thinks" that a certain fact is true, not only to imply that the information has been forwarded by somebody other than the speaker. A practicable means in German to express evidentiality is to use the subjunctive in main clauses. It is a form of indirect speech, which often leaves the original speaker unknown. At the beginning of the last century, grammarians already condemned this mood to perish, but contrary to that prediction, it seems to be becoming more and more usual in modern texts, especially in newspaper texts and, even more in online texts published by newspapers. On the other hand, we must make a difference between mood as a grammatical means and modality as a semantic and even lexical means to express evidentiality. Certainly, in each language there is a method to express uncertainty or wish to exclude the speaker from the contents or implications of the utterance, even if grammatical evidentials are not very common (Boas 1911). In any case, mood often equates with modality. Frawley (1992:386) especially gives the subjunctive mood of different languages as an example of ways of expressing hypothetical assumption or uncertainty.

        The Finnish past perfect is often used when translating the German subjunctive into Finnish in indirect or mediated speech. Anyway, the role of this tense as a mood has not been very much researched in Finnish, even if Osmo Ikola (1950:44) points out that already E. N. Setälä mentioned the use of past perfect as the marker of hearsay in a reported speech.

        Normally the past perfect, being a tense with a low frequency (about 4% in different texts of different languages) is only mentioned as a tense situating both the event frame and the reference time prior to the tense locus (see Frawley 1992:348). Now, especially in Finnish, this remoteness is obviously used by the speaker or writer to create an interval between the information and him/herself and to avoid responsibility towards the recipients regarding the contents of the information. This tense is thus already for Setälä an evidential in Finnish, even if this term was not used by him, losing its temporal aspects and acquiring an increasing amount of modal function in speech and written texts.




Boas, Franz 1911. Handbook of American Indian Languages. Ed. by Franz Boas. Part 2. Government Printing Office. Washington.

Frawley, William.1992. Linguistic Semantics. Lawrence Erlbaum Associates, Publishers. Hillsdale, New Jersey, Hove and London

Ikola, Osmo 1950. Suomen kielioppien käsitys indikatiivin tempuksista ja niiden käytöstä. In: Virittäjä 1950, pp. 35 to 46.

Tiit Hennoste

University of Tartu

Verb or particle? Estonian kuule 'hear', oota 'wait', vaata 'look' at the beginning of the utterance.

There are some word forms used as verbs and as discourse particles in Estonian: kuule 'hear, 2nd person imperative', oota 'wait, 2nd person imperative', vaata 'look, 2nd person imperative', tead '[you] know', ütleme '[we] say'. In some contexts those words are interpretable as verbs, in some as particles and some form periphery between the two. Typically they are analysed in terms of conversation analysis Estonian linguistics to find their pragmatic functions. In my presentation I am interested in sentencies which consist of two parts: above mentioned verb form in imperative and the following clause. Some of those clauses are interpretable as complement clauses and the verb form as controlling verb of this clause (Tead, mida mina pean tegema 'do you know, what I must do'). There are also examples where those words are clearly particles (Kule ma homme ka ei tule 'PART I couldn't come tomorrow too'). But there are lot of cases where it is unclear whether there is particle or verb at the beginning of the sentence (Tead mina ka kohe täitsa ootan et sa tuleksid 'TEAD? I too wait that you would come'; Tere siin Jaanika. tead=et mul on sulle üks palve. 'Hi Jaanika here. TEAD?=that I have one wish to you.' etc).

        In my presentation I will analyse those sentencies, specially unclear cases, in grammatical terms and try to answer to the following questions. Is it possible to reduce the number of unclear cases analysing those utterancies syntactically? Is it possible to interpet some of those verb forms as syntactically reduced non-verbal predicates of complement clauses? Could the form of the clause help us to draw the borders between the verbs and particles (e.g are there differencies between clauses with and without complementizer at the beginning of the clause).

Mayumi Hosono

University of Durham

Defocalization strategy – suppression of a phonetic form in the canonically realized position CANCELLED

I argue that the issue of an ‘apparent imperfection’ of a human language (Chomsky 2000) will shift from whether a language has displacement to whether a phonetic form is suppressed in the position where it should canonically be realized. Pesetsky (2000) argues that the difference between overt and covert wh-movement is phonological: the distinction is whether a wh-phrase is pronounced in situ or in the position to which the wh-phrase has moved (Pesetsky 2000:8). This idea should be extended to the other kinds of movement, namely, DP movement, verb movement, and so on, since a moved element in DP/verb movement is, if it is supposed to move at all, pronounced in the higher position for some reason.

        The problem is that the principled way has been lacking to determine which position is pronounced and why that position is pronounced: it has been assumed that the decision is made optionally (Chomsky 1995, 2000). Assuming that the features are, as the occurrences of the same component, distributed to several places in the derivation (Chomsky 2000), the point will be why in the case of ‘movement’ the phonetic form of the in-situ occurrence is lost though nothing should prevent it from being realized there. A key to solve this problem will, I assume, lie in the saturation process that Rizzi (1986) proposes. He states that a θ-role is saturated (and its phonetic form is lost) when its referential content can be recovered from the context (Rizzi 1986:508). His statement can be extended in the way that when there exists ‘something’ (e.g., some morphological marking or the abstract entity like the context) that enables us to confirm the relevant element, the phonetic form of the element can be omitted in the position in which the element should canonically be realized. Then, I propose to formulate the saturation process as ‘defocalization strategy’ in general, which can ‘suppress’ the phonetic form of an element if and only if some ‘manifestation’ that lets us confirm the element can be found, for instance a morphological marking of [wh] by a wh-phrase or an agreement morpheme. I argue that the cross-linguistic difference in the presence/absence of movement (of all kinds) will be accounted for in terms of which position, either in situ or in the higher position, is pronounced.

Chia-Hui Huang

University of Washington

S-selected Case and the Structure of NP CANCELLED


This paper examines Genitive Case (GEN) in English, and argues that GEN exhibits a Case pattern whose properties are neither Structural nor Inherent under standard Case Theory. I argue that GEN constitutes a new Case type, "S-selected Case", derived from the notion of s-selection. The Case patterning in English suggests that the standard approach to Case is insufficient: Structural and Inherent Case alone cannot account for GENs in English. Therefore a three-way Case distinction is required. Within Minimalism (Chomsky 2000), Structural and Inherent Case are [-interpretable]; the significant property of the third Case, S-selected Case, is [+interpretable].

        Consider (1) where the postnominal NP 'Peter' is an argument, and receives a theta-role and Inherent Case from the head noun. In (2) however, the prenominal NP 'Peter' cannot be analyzed as Inherent Case, since it is not an argument, therefore it cannot receive a theta-role from the head noun. Nor is (2) Structural. Recall that Structural Case is independent from theta-role assignment therefore it can be checked with pleonastic or in Exceptional Case Markings. GEN cannot be checked with pleonastics or in Exceptional Case constructions.

        To account for GEN marking, I argue that postnominal GEN as shown in (1) is Inherent Case as it receives a theta-role from the head noun. Prenominal GEN, on the other hand, is analyzed as S-selected Case since it is capable of generating multiple readings that depend on the semantic context of the head noun, as shown in (2). In order to account for the properties of the two types of GEN in (1) and (2), following Carstens (2000) I assume that N is associated with uninterpretable Case feature checking since N assigns theta-roles to its complement, while n is associated with indirect argument selection in the sense of Kratzer (1996) and licenses the [+interpretable] S-selected Case. In other words, N is associated with Inherent GEN since N assigns theta-roles, while n is associated with S-selected Case. Applying the nP/NP analysis to GENs, in (1) the postnominal GEN Merges with N, which has a [-interpretable] Inherent Case feature. Through Merge, Agree is established to check the [-interpretable] feature of N. On the other hand, the prenominal GEN Merges with n, and receives S-selected Case -'s.

        Furthermore, S-selected Case can provide a solution for (3), which is unexpected given standard Case theoretic assumptions. In (3), postnominal GEN has double Case marking, namely of and -'s. Additionally, having S-selected Case we can finally account for temporal adverbials such as yesterday's lecture vs. *the lecture of yesterday where the head noun semantically selects an external argument that has a deictic reading.

        Thus, S-selected Case will be shown to account for three challenging cases that are problematic in standard Case Theory: prenominal GENs, GEN double Case markings, and temporal adverbials.


(1)   [The pronunciation of Peter] varies from person to person.


(2)   [Peter's book] is on the table. 'The book owned by Peter' or 'The book written by Peter'


(3)   [The book of Peter's] is on the table.

Esa Itkonen

University of Turku

An inflectional system with optional and interchangeable endings

The oldest extant document of Ancient Tamil is Tolkaappiyam, which contains both a grammar of this language and a poetics for its speakers. Rule 104 of Chapter 3 of the second Book (= ‘Collatikaaram') states that case endings may be dropped, and rule 106 states that case endings may be interchanged. The existence of these structural features is amply confirmed by texts (= so-called cankam or sangam literature) composed c. 100 - 300 AD, and they have been summarized as follows: "Die Auslassung der Kasussuffixe und auch der Lokativpostpositionen ist im Alttamil ein auffallend häufiges Phänomen", "ein Kasussuffix [kann] durchaus mit der Funktion eines anderen Kasus gebraucht werden" (Lehmann 1994: 52, 42). As a result of the former fact, even a complex sentence is most often just a string of nominal (and verbal) roots, apart from the last word which is a finite verb.

        From the typological point of view, these facts are quite surprising. Indeed, they are so surprising that discussants on FUNKNET tended to simply deny their existence, once they were brought up. This is, however, the wrong response (due, no doubt, to man's inborn ‘fear of the unknown').

        In the present paper I shall pursue the overlapping goals of finding crosslinguistic analogues of these facts and of explaining them.



Chevillard, Jean-Luc. 1996. Le commentaire de Ceenaavaraiyar sur le Collatikaaram du Tolkaappiyam. Institut francais de Pondichery.

Itkonen, Esa, 2000. Tolkaappiyam: the basic work of Ancient Tamil language and culture. SKY Journal of Linguistics.

Lehmann, Thomas. 1994. Grammatik des Alttamil. Stuttgart: Franz Steiner Verlag

Airi Jansons

University of Tartu

An overlapping area of the pragmatic particle and adverb: ‘siis’ and ‘nüüd’ in spoken Estonian

The author of this paper has focused on the overlapping area of pragmatic particles and adverbs – Estonian particles ‘siis’ and ‘nüüd’. As pragmatic particles both ‘siis’ and ‘nüüd’ have mainly two functions: a boundary margin in the beginning of a new subject or episode and a colouring particle emphasizing some emotional nuance (instigation, commanding, reluctance, wonder etc). As adverbs both of them refer to time: past (siis ‘then’) and present (nüüd ‘now’). The particle ‘siis’ may also refer to condition (kui…siis ‘if…then’) or circumstance.

        In theory it should be possible, and also seems easy to distinquish these functions, but in practice there is a number of examples of ‘siis’ and ‘nüüd’ that are difficult to define: the functions seem to blend: ‘siis’ or ‘nüüd’ may function as the adverb and particle at the same time. A frequent blend area can often be seen in case of monologues where ‘siis’ and ‘nüüd’ are used as boundary markers in the beginning of the text parts. If the events are retold in chronological order, ‘siis’ in the beginning of utterances seems to be a normal temporal adverb. However, in cases of chronological disorder (if the exact meaning of ‘siis’ is not specified), there is no doubt that ‘siis’ is a pragmatic particle: for the most part, it indicates the beginning of a new textual unit, yet retaining some of its temporal meaning. The particle ‘nüüd’ is used similarly in the beginning of comments about actions that take place at the speaking moment, for example in radio monologues or in giving instructions. In these cases ‘siis’ and ‘nüüd’ do not indicate clear temporal connections, but still have a certain temporal shade.

        The other complicated area is the question of colouring particles which follow the verb. These are emotive and instigating utterances, in which ‘siis’ may also refer to a circumstance and ‘nüüd’ may refer to time.

        When speaking about ‘siis’ and ‘nüüd’, it is reasonable to take them for adverbs only in cases where they are clearly deictic, and treat others as particles with by- or additional function. Trying to simplify this complicated phenomenon, it may be said that the particles ‘siis’ and ‘nüüd’ have two general overlapping areas in spoken Estonian:


    1.boundary marker + adverb (temporal);

    2.colouring particle + adverb (temporal and conditional).

Baris Kabak

University of Konstanz

Adjunct clauses headed by postpositive markers in Turkish


Turkish employs a number of postpositive elements to convey various syntagmatic relations. Prototypical postpositional phrases consist of a postposition that normally assigns case to the nominal phrase that it follows (1). Several postpositions can also function as clause-combiners and become the heads of adjunct clauses (2, 3). Closer examination on such adjunct clauses reveals that their subjects can be either in the Genitive (2) or in the Nominative case (3).


(1)   [Binnaz-a        göre     ] bir   elbise

        [Binnaz-DAT  P         ] a      dress

"a dress suitable for Binnaz"

(2)   [[Ayşe-nin      anlat-dıg-ın]-a            göre    ], Osmanepey üzgün ol-malı

        [[Ayşe-GEN   tell-FNom-3sg]-DAT P        ]  Osmanvery sad     be-must

"According to what Ayşe says, Osman must be very sad"

(3)   [[Ayşe-Ø          var-dıg-ın]-a                  göre]  biz   gid-ebil-ir-iz

        [[Ayşe-NOM  arrive-FNom-3sg]-DATP     ]  we   go-able-Aor-1pl.

"Since Ayşe has arrived, we can go"

While the Genitive on the subject of the embedded clause comes from the possessive agreement marker on the verb in (2), the Nominative appears in (3) despite the presence of the same possessive agreement morphology. Interestingly, the same postposition can carry different meanings. For instance, while “göre” in (2) means “according to”, the same postposition in (3) conveys causality, expressing a conjunctive meaning.

        Furthermore, certain postpositional phrases (e.g., (3)) exhibit interesting similarities to adverbial clauses that carry case markers (4), and adjunct converb clauses (5).


(4)   [Ayşe-Ø           ev-e             gel-dig-in-de],                ben-i    ara

        [Ayşe-NOM   house-DAT come-FNom-3sg.-LOC I-ACC call

"When Ayşe comes home, call me"

(5)   [Ali-Ø          ev-e             gel-ince/-meden],                           ben-i       ara

        [Ali-NOM  house-DAT come-"when"/ “without having”  I-ACC    call

"When/ before Ali comes home, call me"

The adjunct clauses in (4, 5) also have subjects in the Nominative and carry meanings that are typically associated with conjunctive constructions, expressing temporal relations, sequentiality, contrastiveness, causality, etc.

        It seems that Turkish imposes restrictions on postpositive markers as to what kinds of syntactic hosts they can select. The differences in the kinds of meaning they express may depend on the kind of host they attach to. The empirical question, thus, remains as to whether the various functions of Turkish postpositive markers can be classified systematically.

        In this paper, I will first provide a detailed examination of various syntactic and semantic functions of such adjunct clauses. Second, I will propose an analysis that classifies postpositive markers into (at least) two groups: (i) those that select nominal complements (e.g., göre(1) in (2)), (ii) those that select only clausal complements (e.g., göre(2) in (3); Locative in (4)) (cf. Aygen, 2002). While in (i), sentential complements are genuinely nominal through a syntactic requirement, those in (ii) acquire noun-hood due to a morphological condition that requires the host of case-markers to be nominal. This account correctly predicts the nominal morphology on the subordinate predicates in (2, 3, and 4) since these complements serve as the hosts of case markers, which come from postpositions (e.g., Dative in (2, 3)), or are case-markers themselves that function as adverbials (e.g., Locative in (4)). The subject is in the Genitive in (2) since the embedded clause is genuinely nominal, as is required by the postposition göre(1). The nominative on the subject in (3) and (4), however, appears as a default case since the Genitive case assignment fails to apply in “semi”-nominal clauses (Kornfilt, 2003).

        Third, I will discuss possible explanations for the similarities between postpositions of type (ii) and other postpositive markers illustrated in (4) and (5). I will suggest that these postpositive elements have acquired converbial functions, multiplying their meaning to include conjunctive/ discoursive expressions.



Aygen, G. (2002). Subject case in Turkic subordinate clauses: Kazakh, Turkish and Tuvan. NELS 32. (563-579).

Kornfilt, J. (2003). Main embedding types in Turkish. Handout presented at the Workshop on Clausal Integration in Head-Final Languages. University of Konstanz.

Gerson Klumpp


Via Grammaticalization into the Periphery and Back to the Core: Main Verbs and their Aspectual Auxiliaries


Turkic languages as well as Samoyed languages from Southern Siberia are known for having grammaticalized verbs with prototype aspectual semantics like e.g. verbs meaning 'lie' (stative-imperfective, non-transformative) or 'depart' (resultative-perfective, final-transformative) into aspectual auxiliaries and further into aspectual or aspecto-temporal suffixes. In being grammaticalized, these verbs first lose all those of their semantic components which are of no aspectual value. The remaining aspectually salient components can be transferred syntagmatically to other verbs, which thus become aspectually modified. The process of modifying takes place within a converbal construction, which is a first of all a semantically empty means for connecting two or more verbs and gives rise to all kinds of coordinative and subordinative readings, depending on the context. The Kamas form in (1) may serve as an example: There is a main predicate in the syntactic core, modified by a verbal form from the syntactic periphery. The relation of the two converbally connected verbs can be understood as modificational, the direction of modification is from left to right, a head final pattern according to the areal syntactic law of modificator ante



(1)   Kamas

        s'arla'' nörb̊bjem

        joke.CV           say.PRET.1SG

'I said it jocularly' (DW: 210)

(Lit.: Joking I said it)


Aspectual auxiliaries, on the other hand, modify into the opposite direction, i.e. from right to left, cf. example (2a). The imperfective auxiliary i''b̊- 'lie' modifies a verb which can mean 'stand, stand up, stop'. The result of the imperfective modification is the selection of the nontransformative meaning 'stand' ('being standing'):


(2a) Kamas

        nula''       i''b̊m

        stand.CV lie.PRES1SG

'I am standing' (CM: 115)

(Lit.: Standing I lie)


Within this construction a disbalance between semantic and syntactic weight can be observed. Taking an aspectual auxiliary the semantically main verb has to retreat into the syntactic periphery to allow the auxiliary to take the position of the syntactic main predicate. Thus semantic weight is not represented syntactically:

Semantically heavy (main predicate) = Syntactically light (periphery, converb)

Semantically light (aspect auxiliary) = Syntactically heavy (core, finite)


Further grammaticalization processes lead to a formal distinction between the verb meaning 'lie' and the corresponding aspectual auxiliary which up to now couldn't be distinguished. Apocope, ellipsis and other sound changes lead to the contraction of converb ending and auxiliary stem.The stem of the main predicate as well as the inflectional endings of the auxiliary remain unaltered. The result is a new aspect suffix which follows the stem and precedes the inflectional endings for mood, tense and person. By amalgamating its converb ending and the stem of the following auxiliary into a suffix and by taking over the mood-tense-person-inflection of the auxiliary the main predicate has managed to be back in the core, cf. (2b):


(2b) Kamas

nula''b̊m [< nula'' i''b̊m]


'I am standing' (DW: 147)

The paper illustrates the phenomenon with data from Southern Siberian Samoyed and Turkic.


Ritva Laury                                       &     Eeva-Leena Seppänen

California State University, Fresno            University of Helsinki

& University of Helsinki

Et(tä)-clauses in spoken Finnish: their syntax, prosody and interactional function

This paper examines the use of et(tä)-clauses in Finnish conversations. Et(tä) is analyzed as a complementizer in standard descriptions of Finnish grammar (e.g. A. Hakulinen and Karlsson 1979:346-347; 353-354; Vilkuna 1996:66-69); however, in spoken Finnish, että-clauses are not always subordinate (Vilkuna 1996:69-70; see also Thompson 2002, Englebretson 2003 regarding complementation in English and Indonesian).

        In our data, että-clauses occur frequently with complement-taking predicates (CTPs) such as pelätä 'to fear' and sanoa 'to say'. These uses are associated with a prosodic pattern where että is grouped with the CTP, with the complement clause following in the next intonation unit (Chafe 1994). Consider the following excerpt, where Liisa is reporting on something told to her by a co-worker.


1  Liisa:  >et   sano        et< ,jos   niille           s:anoo jonku       kellonajan

                et     say-PST  et    if     3PL-ALL    say       some-ACC clock.time-ACC

That (she?) said that if (one) tells them some time

2             ni     se     täytyy  niinkum muuntaa,             .hh  etheillä               alkaa

               PTC 3SG must    PTC       transform-1INF          et3PLLOG-ADEbegin

Then it has to be like translated, that for them

3             se     aamu       siitä             ku        aurinko    nousee,

               DETmorning  3SG-ELA   when   sun          rise

the morning begins from when the sun rises,


In line 14, the että-clause which follows the verb sano 'said' could be considered a complement of that verb. However, the status of the et-clause beginning in line 2 is questionable. Is it a complement of the earlier token of sano, or an independent clause? In our view, rather than being a complementizer, että functions here indexically, regulating the footing of the utterance.

        In our corpus, että-clauses often appear without any CTPs in their environment. In such uses, as in lines 1 and 2 above, että occurs initially in an IU. These että-clauses can function as continuations of the previous unit, or as new units of their own. Consider the following exerpt, in which EL is asking AL to try on a sweater she is knitting for her boyfriend:


1  EL  Nouseks          vähä seisomaa            mää pruuvaan tätä.

            Arise-Q-2SG  little stand-3INF-ILL 1SGtry           this-PRT

Would you stand up for a bit (so) I (can) try this.

2          Kuin    pitkätää   on.

            how     long this is

How long this is.

3  RV  @Hänel       vai.

            3SG-ALL    or

On her?

4  AL  Mul              vai.

            1SG-ALL    or

On me?

5  EL  Nii.


6          Et   [mä] suurin              piirtein               tiärä.

            että 1SG large-INSTR   feature-INSTR   know-1SG

(So) that I can tell approximately.


The että-clause in line 6 could be analyzed as being integrated into the utterance in line 5, forming a nii(n) et(tä) 'so that' sequence, or it could be analyzed as a syntactically and interactionally independent unit, where niin functions as a response token and the että-clause as an account. The latter analysis is more plausible, given the prosodic and sequential evidence.

        In our paper, we will explore the syntactic and prosodic features of että-clauses, and how these factors are related to their interactional functions and their status as embedded and independent clauses.

Larisa Leisiö

University of Helsinki / Academy of Finland

Discourse particles in Nganasan


Nganasan (one of the one of the Northern Samoyed languages) has free and bound particles. Bound particles precede declination and conjugation markers.

        I will discuss the focus and contrastive particles. The bound morheme -RAA- marks exclusive focus: turku-raa-" lake-only-Pl 'only lakes'. The meaning of this particle seems to vary depending on the stem to which it is attached.

        Particles indicating are inclusive focus are a bound morpheme -ngalê and a word têbtê.

        Contrastive focus can be indicated by the free particle tê", as well as by morphemes -KÜMÜ, -KÜê, and -Kêlit'ê the latter being characteristic of negative polarity.

        I will analyse the scope and position of the focal particles, variation in their meaning in different contexts and in the interplay with other particles used in the utterance. I will also show their Russian substitutes which appear in Nganasan-Russian alternation.


Torsten Leuschner

Universiteit Gent

Left-Dislocation and the Degrammaticalization Debate: Some Diachronic Aspects of Nonspecific Free Relatives in German and Dutch

One of the most contentious issues in functional syntax at the moment is the problem of directionality in grammaticalization, and in particular the existence and significance of degrammaticalization. In a recent contribution, Haspelmath (2002) has proposed to divide alleged examples of degrammaticaliza-tion into "antigrammaticalization" and "retraction". Only cases of "antigrammaticalization" are said to represent genuine counterexamples to unidirectionality in grammaticalization.

        In my contribution to this symposium, I propose to revisit the historical development of nonspecific free relatives, particularly in German and Dutch, in the light of the degrammaticalization problem. In present-day German (and also Dutch and English), non-specific free relatives are introduced by the same kind of WH-based subordinator as a type of adverbial clause called "concessive conditional" (cf. Leuschner 2003):


(1)   Wer immer einen solchen Vergleich anstelle, habe in seinem Kabinett keinen Platz. (Deutsche Welle radio)

'Whoever draws such a comparison [wrote Schröder to Bush] has no place in his cabinet.'


(2)   Was immer wir auch June vorwerfen könnten - Zusammenarbeit mit J.F. Traber gehört be-stimmt nicht dazu! (Mannheimer Korpus)

'Whatever accusations we could raise against June - cooperation with J.F. Traber is certainly not one of them!'

In verb-second languages like German and Dutch, structures like (2) are unusual because the preposed adverbial clause is part of the syntactic periphery, whereas it ought normally to function as a consti-tuent filling the "forefield" of the main clause verb. Based on typological observations, Haspelmath / König (1998) suggest that concessive conditionals like (2) are (at least in some cases, including Ger-man and its sister languages) derived diachronically from highly integrated nonspecific free relatives like (1) by some kind of left-dislocation which shifted the free relative clause clause towards the peri-phery of the sentence. This would constitute a sharp decrease in condensation and hence a genuine counterexample to unidirectionality - i.e., antigrammaticalization.

        In my contribution I would like to present data from the history of free relatives in (mainly) Middle High German and Middle Dutch which show that Haspelmath / König's approach to the dilemma is too narrow and that the historical relationship between concessive conditionals and nonspecific free relatives is actually a case of retraction, not antigrammaticalization. Once we adopt this distinction, both the typological evidence and unidirectionality can be satisfactorily accomodated, and certain pro-blems concerning the historical development of other types of concessive conditionals in Germanic can be resolved as well.



Haspelmath, Martin (2002): "On Directionality in Language Change with Particular Reference to Grammaticalization." Lecture at New Reflections on Grammaticalization 2, University of Amsterdam, April 5. Available for download at pdf

Haspelmath, Martin / König, Ekkehard (1998): "Concessive Conditionals in the Languages of Europe." In: Johan van der Auwera (ed.), Adverbial Constructions in the Languages of Europe (= EUROTYP 3.) Berlin / New York: Mouton de Gruyter, 563-640

Leuschner, Torsten (2003): Between Discourse and Syntax: The Syntacticization of Concessive Conditionals in English, German and Dutch. Ph.D. dissertation, Freie Universität Berlin

Heather Lewis, Ida Toivonen, & Kate Kearns

University of Canterbury, New Zealand

The grammatical function of with-themes in English


This paper examines the so-called with-theme PP in English, exemplified in (1):


(1)   Josie stuffed her bag with old clothes.


With-themes are often called adjuncts in the literature (see, e.g., Jackendoff 1990), and indeed they display some characteristics typical of adjuncts: they are optional and marked with a preposition. However, there are also reasons to treat them as arguments: a with-theme is generally part of the core meaning of the verb, and they can be extracted out of (2):


(2)   What kind of things did she stuff her bag with pictures of?


Note also that (1) can be reformulated as (3):


(3)   Josie stuffed old clothes into her bag.


In (3), the NP 'old clothes' is not marked with the preposition 'with', and it is an uncontroversial argument, both syntactically and semantically. (1) and (3) exemplify the spray/load alternation (Fillmore 1968, Anderson 1971, Rappaport and Levin 1985, 1988). This paper considers with-theme phrases as they are used with several different classes of verbs; for example, the spray/load verbs illustrated above, but also fulfilling verbs like 'present', image impression verbs like 'inscribe', and others.

        We examine with-themes in Lexical Functional Grammar (LFG), where grammatical functions (GFs) are taken to be important grammatical building blocks. In addition to direct functions, LFG adopts OBL(ique) and ADJ(unct) functions, which can both be PPs. They differ from each other in that OBL has argument status and ADJ does not.

        In early LFG, the with-theme is analyzed as an OBL (Bresnan 1982). However, in modern LFG, which includes Lexical Mapping Theory (LMT), this analysis becomes problematic. LMT governs the syntactic level of a(rgument) structure, which links lexical semantics with GFs. As it stands, LMT falsely predicts that the with-theme is a secondary object: we assume that the two plausible possibilities are ADJ and OBL. Although the with-theme could be straightforwardly analyzed as an ADJ, by means of a lexical operation, we argue that the with-theme is most appropriately treated as an OBL. To achieve this, we must add a lexical specification on the lowest argument of the relevant verbs. However, the wide distribution of with-themes seems inconsistent with analysing them in terms of lexical idiosyncracy.

        As an alternative, we revisit the assignment of theta-roles at the level of lexical semantics. Specifically, we argue that the notion of 'patient-like' roles, which is crucial for LMT, is too vague and needs to be further refined. We relate our analysis of with-themes to more general issues regarding the distinction between OBL and ADJ (arguments and adjuncts).


Jan Lindström                &    Susanna Karlsson

University of Helsinki            University of Göteborg


Center or periphery? Verb-first declaratives as a syntactic and interactional resource in (spoken) Swedish


In this paper we will sketch a coherent picture of the diverse uses of declarative verb-first constructions, and thus pinpoint their common motivation and import in sentence- and turn construction in Swedish. Basically Swedish is a fixed word-order language with the typological characteristic of placing the finite verb always in the second constituent position in a declarative. Alike other V2 languages like German, there is sometimes an option for a verb-first order, whereby the first, normally obligatory clause constituent seems to be ellipted. V1 declaratives have been regarded as a peripheral phenomenon that is limited to certain textual or syntactic environments, like narratives (1), dialogic responses (2), and the non-realization of the formal, "dummy" subject (3):


(1)   Kom   hon  in där.      Kände             han igen henne.

        Came  she   in there.   Recognized    he     her.

(2)   -  Har hon  kommit?  - Vet      jag   inte.

            Has she   come?         Know  I       not.

(3)   [det]   Duger   utmärkt   med räkor       i stället för   kräftor,tyckerjag.

        [it]      Does     excellent with shrimps   insteadof    crabs,  think  I.


We will argue, however, that the scope of V1 declarative constructions is much less constrained in the grammar of Swedish than is traditionally recognized. Example (3) reveals this point partly: it is notable that the subsequent comment clause, "tycker jag" 'I think', also has a V1 order, not only the preceding clause. This is the case with similar modal semi-formulaic expressions and report clauses that follow a quotation:


(4)   Det är     varmt, sade hon.

        It     is     warm, said  she.


Instances of this kind are not traditionally analysed as V1 clauses in their own right but the preceding clause is understood as a first constituent (the object) to the following verb phrase and subsequent clausal core. Evidence from spoken language suggest though other ways of analysis that upgrade the intrinsic constructional value of V1 declaratives. Further, it can be noted that several discourse particles seem to be lexicalizations of short commenting declarative clauses, for example "vet du" 'you know' (lit. 'know-you'), "ser du" 'you see' (lit. 'see-you'), and "vet jag" 'I say' (lit. 'know-I'). The V1 constructions that can be attested in dialogic responses of the type in (2) could be seen as variations of this generally available V1 constructional format in the grammar of Swedish.

        There is compelling evidence for that V1 declaratives are an important resource in especially spoken Swedish in the management of turn construction and response elicitation (cf. also discourse particles above). Constructions like "tycker jag" are used as a means for making relevant the possibility for heteroglossic diversity, opening up for other interactants to offer their point of view. Accordingly, they tend to appear in context of turn-taking difficulties, such as lack of uptake. The V1 format offers a method by which the speaker can add a continuation to his/her utterance, thus re-signalling and re-positioning, indeed "post possible completion", the turn-closure. It is also worth noting that V1 structures are the central resource by which pivot constructions are tied together into valid turn constructional units in Swedish, e.g. "jag brukar cykla dit på höstarna brukar jag" 'I use to ride a bike there on Sundays use I'.

        Our study is empirically based and examples of real occurences in speaking and writing will be provided. We will also explore theoretical, syntactically and interactionally inspired models for the description of the linguistic phenomenon at hand.

Edoardo Lombardi Vallauri

Università di Roma Tre

Grammaticalization of syntactic incompleteness: Italian unembedded conditionals

Spoken Italian shows the presence of a construction that belongs to the "periphery" rather than the "center" of the grammar, and is completely absent in writing. Many subordinate clauses introduced by "se" (= engl. 'if', the subordinating conjunction for conditionals and indirect interrogatives) are actually not embedded in any (overt) main clause.

        This sometimes causes syntactic ambiguity, because the "se" clause can be interpreted both as a conditional and as an indirect interrogative clause. Cf. the following examples from the LIP corpus of spoken Italian:


LIP, Rb8:

B:        se mo' stai lavorando # ti posso richiama'_

C:no no non sto lavorando


C:                                                                    m'hanno annullato un sacco di persone ieri tra l'altro una ragazza che non sta bene_ un altra <?>

B:        ah io ci ho mio fratello con una bronchite pazzesca


LIP, Nb55:

A:eh visto che lui eh venerdi'_ scorso ci presento' dei_ dei trasferimenti di proprieta' al PRA

B:si' aspetta che piglio nota ahah

A:                                                    ecco

B:si'? trasferimenti proprieta' si'

A:ahah eh dovrebbe avere lui la_ la distinta

B:                                                                    ah

A:                                                                        per ritirare gli attestati PRA # eh SE E' POSSIBILE_ EH REPERIRLA altrimenti SE MI PUO' FAR SAPERE come devo fare per ritirarli se posso_ se posso_ andare da qualcuno qualche amico suo che sta la' al PRA

B:allora se e' possibile averla o da chi

A:                                                           mi posso rivolgere

B:si' # <?>


In such cases the ambiguity is also due to the fact that in present day spoken Italian the indicative has almost completely replaced the subjunctive mood, and indicative is also the standard mood for reality and possibility conditionals, leading to perfect formal identity between the two constructions.

        However, the majority of unembedded "se" subordinate clauses are not ambiguous, and can only be interpreted as conditional clauses "lacking" a main clause. In all these cases the utterance shows a certain degree of semantic incompleteness, because the addressee is obliged to imagine what could be the content of the main clause which is actually not produced.

        It will be showed that the lacking main clauses take a small number of semantic values, such as (1) the generic question: "what will happen?", (2) the statement that there is either no need or no hope to act in the situation: "everything's fine", or "there's no hope", (3) the strong denial of what has just been stated: "you are wrong!", (4) the request or offer to do something which is suggested by the conditional clause or by the context.

        The frequency of these semantic values is so regular that they have become higly expectable and, more importantly, strictly associated with the "incomplete" construction under examination. For this and other reasons (among which, intonation), it will be argued that, for spoken Italian, these sentences must no longer be described as unembedded (and in some way incomplete) conditional clauses. Rather, they must be viewed as a new construction, typical of the spoken language (initially belonging to the periphery of the grammar but now actually deserving an "official" place in it), constituted by an independent clause introduced by "se" and having its own, almost codified meanings. The absence of the main clause (whose causes will also be inquired) has taken a precise grammatical function, allowing the unembedded conditional to become a new kind of sentence. This is why we suggest to speak of "grammaticalization of incompleteness". Such facts having clear pragmatic grounds, it is highly probable for them to take place, in similar ways, in other languages. The hypothesis will be verified in some indoeuropean and non-indoeuropean languages.

Christiane Migette

Université Paris 13

Syntactic function(s) of clauses or phrases introduced by assuming


Quite a number of -ing forms are used in English to introduce clauses and/or noun phrases with varying degrees of grammaticalisation as conjunctions and/or prepositions. The clauses or phrases introduced by such forms are generally considered as adverbial adjuncts. This paper will concentrate on assuming: it will start with a corpus-based study of the different patterns of syntactic behaviour of this form used as a connector: examples show that it can introduce a that-clause or a noun phrase, which is a very common use, but also an infinitive clause. These clauses or phrases appear in different positions in the superordinate sentence: initial, final or mid-positions are all possible, and one question to be studied is whether those different positions and the type of complementation of assuming are linguistically significant.

        Diacritics such as commas, brackets or dashes are regularly used in written texts, which suggests at least for the last two mentioned a peripheral status. Yet, the basic semantic value of assuming is conditional and it can be permuted with if when it is followed by a that-clause, which indicates that the assuming-clause cannot be suppressed without an important change of meaning. This is in contradiction with a peripheral status and lends support to the idea that the connector can be considered as a predicator in itself with two arguments being what is traditionally called the main and the subordinate clauses: in the specific case of assuming, a sentence of the form assuming p, q could thus be considered as equivalent to < the truth of p conditions q>. In the light of this analysis, the complex sentence includes three predications. We shall see if this hypothesis can be considered valid whatever the degree of integration of the clause in the sentence and the scope of the condition expressed.

        Another factor which should be taken into account when analysing assuming is the influence of its lexical origin as a verb. This influence can still be seen in the attested use of dependent infinitive clauses and in the fact that, though it can be permuted with if, the reverse is not always possible, and it is not simply a semantic problem. Clauses or phrases introduced by assuming are in fact similar to dangling participle clauses: assuming can thus be considered as still being a verb requiring an animate subject, even though this subject is not explicit; the re-building of the missing grammatical subject can help to analyse the level of the constituent in the structure of the sentence. In conclusion, a comparison will be made with French in order to see if similar phenomena occur.

Peep Nemvalts

Tallinn Pedagogical University / Stockholm University


Reconsidering syntactic functions — Estonian example

If we agree that syntax is that layer in language structure which is used to connect meanings (semantic layer) and forms (morphological layer), we may have to reconsider the syntactic functions.

        1. Taking into account that every layer has a multilevel structure, it is necessary to distinguish sentence/clause level functions from those at phrase level.

        1.1. Functions like SUBJECT or OBJECT are meaningful for relating particular sentence level SbP-s to the sentence nucleus (VP). In Estonian grammars, ADVERBIAL has become an overwhelmingly dispersing function, occurring everywhere in the structure in tens of claimed subclasses, which hardly have anything in common (cf. EKG 1993). However, functions of sentence level constituents can be determined more exactly by matching the semantic and morphological layers consistently.

        INDIRECT OBJECT was already suggested in the 1960s, on somewhat different grounds. Furthermore, INDIRECT SUBJECT can be clearly separated as a syntactic function of its own (Nemvalts 2001), e.g., meil (we+ADESS) in Meil tuleb mõelda ‘We have to think’.

        The question is: Do we need ADVERBIAL as a syntactic function at all?

        Because the concepts adverb and adverbial have become too diffuse, I have introduced the function of DESIGNATOR (Nemvalts 2001, 2003) for the sentence level phrases referring to circumstantial notions.

        1.2. It is only at the phrase level that the functions PREDICATIVE and ATTRIBUTE are purposeful. The attributive constituents in a SbP are either qualifiers or quantifiers. There is no sense in postulating ‘adverbial’ as a function of phrase level satellite constituents like intensifiers, e.g. väga (‘very’), in a QlP väga väike (‘very small ~ little’), nor in other

nominal phrases.

        2. Are some functions more peripheral than others? I shall discuss ATTRIBUTE vs. APPOSITION, APPOSITION vs. exclamatory ADDRESS.

        3. Another question is: Are syntactic functions inevitable for every constituent at every level?



EKG 1993 = Mati Erelt et al., Eesti keele grammatika II. Süntaks. Lisa: Kiri. Eesti Teaduste Akadeemia Eesti Keele Instituut. Tallinn.

Nemvalts, Peep 2001. Süntaksi ja semantika koostoime eesti keeles. In: Congressus Nonus Internationalis Fenno-Ugristarum 7.-13.8.2000 Tartu, Pars V, Dissertationes sectionum: Linguistica II. Redegit: Tõnu Seilenthal, p. 438-441.

Nemvalts, Peep 2003. Interaction of different levels for communication in Estonian. In: Les Langues Ouraliennes aujourd́hui. The Uralic languages today. Bibliothèque de l'Ecole des Hautes Etudes à la la Sorbonne. (forthcoming)

Frederick J Newmeyer

University of Washington and École Normale Supérieure

'Peripheral' grammatical relations in English: theoretical implications of non-dp subjects and objects

Most work in generative syntax defines the grammatical relations 'subject' and 'direct object' only for DP (or NP). This paper argues that there are also CP, VP, and PP subjects and direct objects, a conclusion which has implications for frameworks as different as the Minimalist Program and Cognitive Linguistics.

        Examples of non-DP subjects and objects are given in (1) and (2):


(1)   a. [CP That John passed] surprised her.

        b.[VP To attend] would give the wrong message.

        c. [PP Between six and seven] suits her fine.

(2)   a. The average fourth grader believes [CP that the earth is round].

        b.Everyone would prefer [VP to come early].

        c. I’ll choose [PP after the holidays] (to hold my party).


I show that the bracketed phrases in (1) are subjects, based on tests using tags, raising, agreement, adverb choice, and emphatic reflexives. Tests based on passivization, 'tough'-movement, and exceptional case marking show that the bracketed phrases in (2) are objects. But some non-DPs do fail to occur in subject or object position. For example, sentential subjects seem not to occur in embedded clauses (3a); with inverted auxiliaries (3b); or after a topicalized element (3c); nor do they show up in passive 'by'-phrases (3d):


(3)   a. *Although [that the house is empty] may depress you, it pleases me.

        b.*Why did [that Mary liked old records] bother him?

        c. *Such things [that he reads too much] doesn^Òt prove __.

        d.*The situation wasn^Òt helped by [that you spoke out of turn].


Examples like (3a-d) have led some linguists to argue that non-DP subjects and objects do not exist. (3a-b), however, are not ungrammatical, but rather unacceptable for processing reasons -- lengthening the constituents following the sentential subject improves acceptability considerably. Whatever is wrong with (3c), it has nothing to do with sentential subjects. If we replace the bracketed phrase with a DP, the unacceptability remains. And the ungrammaticality of (3d) follows from categorial constraint in English prohibiting a 'that'-complement from ever following a preposition. I go on to similarly deconstruct the arguments against sentential objects, and against VP and PP subjects and objects.

        I then discuss how the existence of non-DP grammatical relations supports GB-style free movement and challenges MP-style triggered movement. In a nutshell, if movement is free, the failure of CP, VP, and VP to occur in subject or object position in certain circumstances can be explained by the sorts of independently-needed generalizations needed for (3a-d). MP-based stipulative mechanisms (such as uninterpretable triggering features) that license movement into surface subject or object position lead to loss of generality, since separate triggers would be needed for DP, CP, and PP. Economy considerations, then, argue in favor of free movement.

        As far as Cognitive Linguistics is concerned, the existence of non-DP subjects and objects challenges the strong form of form-meaning iconicity advocated by Langacker and others, where there is a biunique relation between syntactic and cognitive categories. The mismatch between form and meaning shows that, at least to a certain degree, there must be an independent level of syntactic patterning.

Marina Nielsen

Åbo Akademi University

The syntactic function of the word coup in French nominal compositions of the type coup de N and coup de SN.

This paper discusses the syntactic role and value of the word coup in French expressions of the type coup de N or coup de SN. In the latter case, the N is always preceded by an article of some sort (normally, the possessive or the definite article), while this is entirely impossible in the constructions coup de N. Nevertheless, both types can, in some instances, be defined as nominal compositions: following a terminology and a theoretical framework borrowed initially from Benveniste (1974), we can here distinguish between conglomerates (ex. le coup de notre ami Bill), synaptic (ex. coup de main), and semi-synaptic (ex. coup du lapin) expressions. The present paper discusses, and hopes to further clarify, the problematic surrounding these types of nominal structures, with the immediate focus lying on the syntactic role(s) played by coup (un coup de). Through which criteria can they be defined? Do the roles played by coup (un coup de) depend on the type of nominal structure (conglomerate, semi-synaptic, or synaptic)? Do they vary in accordance with the type of structure? The paper shows that the models developed by, e.g., Anscombre (1999) and Gross (1984), do not suffice to make a clear separation of these types of expressions. While Gross (1984) mainly accounts for the morpho-syntaxic, predicative role of the word coup in synaptic constructions like (donner) un coup de main, where the coup de N is preceded by a predicatively empty “supporting verb”, Anscombre proposes to make a clearer distinction between cases of both fixed compositions and “free” syntactic structures having the construction N+preposition+N (translated in our study, into coup de N). Our analyses of the two tests or models of research proposed by Anscombre in his article (1999), have shown that they are not sufficient, and that one needs to integrate, into this question, the very close relation that exists between a certain syntactic structure and the semantic use of the composition or the word. This paper therefore wishes to bring forward the importance of the semantic role played a) by the N, and b) by the word coup. The semantic value of coup is not always the same in constructions of these types, but it can vary from a semantic near-emptiness to a complete lexical value. In order to make a clear definition of coup de in all of its variations (un/le coup de N/SN), it is therefore essential to take into account the full (syntactic and semantic) role and value of the N, as well as the full semantic and syntactic value of the construction as a whole. It is only then, that the syntactic role and value of the word coup can be fully apprised. The proposed solution and model of analysis consists, in this paper, as in our previous research (Nielsen, 2001; Nielsen, to be published), of a further development of a model of distinction presented initially by Guillaume (1919) and Carlsson (1966).

Anette Rosenbach

Heinrich-Heine University, Düsseldorf

On the squishy borderline between determination and modification in English


The present paper will elaborate on what Plank (2003:6) calls the "elusive" nature of the determination-modification distinction. On the basis of an analysis of nominal adnominals in English (i.e. possessives [John's book] and noun+noun constructions [a Cambridge student]) it will be shown that there is some considerable indeterminancy between the syntactic functions of DETERMINATION and MODIFICATION in English. It will be argued that determination and modification are not syntactic primitives but rather instantiations of a clustering of more primitive semantic features (i.e. the referentiality and the restrictiveness of the nominal dependent), which prototypically map to certain formal properties (e.g. adjacency to head, co-occurrence with articles). This mapping between meaning and form is not idiosyncratic for English, but appears to have cross-linguistic validity, too (see e.g. Koptjevskaja-Tamm 2003).



Koptjevskaja-Tamm, Maria (2003). "A woman of sin, a man of duty, and a hell of a mess: non-determiner genitives in Swedish". In: Plank (ed.), 515-558.

Plank, Frans (ed.) (2003). Noun Phrase Structure in the Languages of Europe. Berlin/New York: Mouton de Gruyter.

Jouni Rostila

University of Tampere


From parallel predications to prepositional objects and progressive constructions: The grammaticalization of an economical strategy of expression


Traugott & König (1991:197) propose that temporally overlapping static predications like in (1) are apt to give rise to causal implicatures and their conventionalization.


(1)   I couldn¹t work when the television was on.


My purpose is to show that the pattern in question might form a widespread strategy of expression applied for the sake of its economy. A product of its application can be seen in prepositional objects like in (2), which, in my view, have developed out of adjuncts:


(2)   Er hat Angst vor Bären.

My suggestion is that speakers introduce prepositional objects like vor Bären ‘(because) of bears’ as locational adjuncts (literally ‘in front of bears’) without making explicit the causal connection between the two predications (Pred 1: Er hat Angst; Pred 2: Pred 1 vor Bären) they nevertheless want to express. Instead, they leave it for the hearer to judge in which way the adjunct predication is relevant to the main predication of the sentence. Being underspecified in this way, the strategy is economical; in addition, the strategy makes use of already existing ways of expression. The conventionalization of the causal implicature intended by the speaker and used by the hearer creates a complement (prepositional object) relation and a new meaning of the morpheme originally used to form the adjunct. Thus vor in (2) acquires the reading [+causal] in the course of this grammaticalization process (cf. Rostila, in press).

        Basically the same strategy and its grammaticalization might lie behind the development of progressives out of postural verb constructions (Ebert 1996). In cases like (3), which instantiates the postural verb construction that one of the present-day Dutch periphrastic progressives has arisen from (cf. Van Pottelberge 2001), no causal implicature is available:


(3)   Hij zit en eet. (literally ‘He sits and eats.’)

Instead, such expressions invite the hearer to infer which of the two predications conveys the main purport of the speaker, i.e. would mean saying ‘enough’ in the sense of Grice’s maxim of quantity. The conventionalization of such implicatures brings about the establishment of a class of verbs that regularly DO NOT constitute the main predication of the sentence but form a temporal frame for it. In other words, the process results in the development of new, auxiliary-like readings for postural verbs.

        Apart from dealing with cases like (2) and (3) in more detail, my purpose is to address the question of whether the conventionalization of the progressive implicature could take place in present-day Swedish, which also displays structures like (3), and in contemporary Finnish, which likewise exhibits a possible source structure, cf.

(4) Han sitter och läser. (= (3))

(5) Hän seisoo katselemassa juoksijoita. (lit. ‘She/he stands watching the runners.’)


Finally, (1-2), (3-4) and (5) correspond to three different syntactic structures. If all of them can nevertheless grammaticalize in largely the same way, this would speak for the semantic nature of grammaticalization and its independence of structure (in contrast to, e.g., Roberts 1993).



Ebert, Karin (1996): Progressive Aspect in German and Dutch. Interdisciplinary Journal for Germanic Linguistics and Semiotic Analysis 1, 41-62.

Roberts, Ian (1993): A Formal Account of Grammaticalization in the History of Romance Futures. Folia Linguistica Historica XIII/1-2, 219-258.

Rostila, Jouni (in press): Zur Grammatikalisierung bei Präpositionalobjekten. In: Leuschner, Torsten & Mortelmans, Tanja (eds.), Grammatikalisierung im Deutschen. Berlin: de Gruyter.

Traugott, Elizabeth Closs & König, Ekkehard (1991): The Semantics-Pragmatics of Grammaticalization Revisited. In: Traugott, Elizabeth Closs & Heine, Bernd (eds.), Approaches to Grammaticalization (Vol. I). Amsterdam/Philadelphia: John Benjamins, 189-218.

Van Pottelberge, Jeroen (2001): Verbonominale Konstruktionen, Funktionsverbgefüge: Vom Sinn und Unsinn eines Untersuchungsgegenstandes. Heidelberg: Winter.

M.Rusakova, A.Rusakov and S.Say

St. Petersburg State University

Do syntactic relations really exist? Three case studies in support of the (Radical) Construction Grammar approach

The paper addresses the problem of necessity and adequacy of the concept of syntactic relations in a anthropologically-oriented description of syntactic phenomena. The three sets of data discussed here belong to overtly unrelated domains; however, all of them imply similar theoretical generalizations. The first group of data comes from the analysis of a corpus of speech errors in the Russian Adj+N construction. It is traditionally assumed that the adjective receives its inflectional characteristics from the "head" noun. However, an analysis of such NPs, in which an erroneous discord between adjective and noun is registered, shows that speakers often provide the adjective with its morphological setting independently from the noun and, moreover, the noun often gets its inflectional properties from the (preposed) adjective. In the normal case, grammatical and/or semantic analysis that is necessary for the morphological setting of the components of the NP is undertaken only once, the second component being "mechanically" assimilated to the first one, should it happen to be adjective or noun. It is further concluded (at odds with the traditional viewpoint) that syntactically the components of the Russian Adj+N construction are mutually independent and the morphological categories of the adjective are neither semantically void, nor syntactically bound. In other words, the structure of the NP is characterised by agreement as a resulting superficial property of the construction as a whole, but not as an underlying process. The second group of data comes from the analysis of contact phenomena in the North Russian Romani Dialect (NRRD), namely, of its syntactic interference with Russian. In particular, it is shown that the firm correspondences have been established between the NRRD case forms and Russian case forms or prepositional phrases. For instance, the NRRD instrumental case corresponds to the Russian [s 'with' + Instrumental] construction, even for the idiosyncratic uses of the latter (due to the mechanism of loan translation). It is thus clear that a striking discrepancy between the constructions of the two languages in terms of traditionally assumed syntactic relations does not preclude the possibility of their cross-language identification. In other words, a bilingual speaker identifies the constructions and not the categories of the two languages. The third part of the paper addresses a typologically rare phenomenon in Russian, namely, the non-anaphoric verbless utterances such as Ja v magazin lit. 'I - to the shop', that is, 'I am about to go to the shop'. These utterances are strikingly non- compositional; besides, they are rather unwelcome for any strictly verbocentric theory of syntax, in which the form of the arguments is always determined by the verb. Based on a detailed syntactic and pragmatic analysis of such utterances it is argued that they are represented in speakers' competence as relatively flexible syntactic patterns, endowed with their own semantics, and are not merely contextual realisations of "full" (verbal) syntactic models. It is shown that the components of these collocations are usually NPs used in prototypical context-free case meanings. The three groups of data are compatible with the theoretical premises of the Radical Construction Grammar, "a nonreductionist theory of syntactic representation, [in which] constructions, not categories and relations, are the basic, primitive units of syntactic representation. The categories and relations are derived from them:" (Croft 2001: 48).

Rami Saari

Hebrew University of Jerusalem


Maltese ilu — The only postposition in a language of prepositions?


Maltese, a modern vernacular form of Arabic with its own linguistic standards and peculiarities, is one of the only Semitic languages written in the Latin alphabet. Unlike Classical Arabic, Maltese does not use any syntactic cases. Instead, it has a well developed prepositional system. Maltese prepositions play a crucial role in the structure of prepositional phrases, functioning like heads of the nominal complement. These phrases are used mainly as the expression of a semantic complementation, which represents the adverbial relation. Therefore, when related to the verb or to its agent, the prepositional phrases substitute for objects or for adverbials. The only Maltese adposition which may be considered a postposition is ilu 'ago'.

        The purpose of this paper is to examine the roots and the usages of this adposition and to compare them with the usages of the parallel adposition in Arabic, English and Italian, the three most important languages in regard to the historical development of Maltese. The paper also deals with the difficulty of using the term "postposition" when referring to languages which are traditionally considered not to have such a category. While giving examples of the problematic definition of postposition versus adverb and postposition versus case ending from several other languages (Basque, Finnish, German, Greek, Hungarian and Turkish), the paper examines critically some of the linguistic principles used to define syntactic functions.

Ivo Sánchez-Ayala

University of California Santa Barbara

A Usage-Based Approach to Spanish Oblique Arguments CANCELLED

Current usage-based linguistic models such as those of Bybee (1985, 1998, 2002, 2003) and Langacker (1987, 2000, forthcoming) have put forward a view of linguistic structure as largely emergent from local patterns of frequency of use. In this paper, we will apply the insights of usage-based models to the morphosyntactic distinction between types of oblique arguments in Spanish conversation. We will contrast received categories of oblique with the different types of oblique subconstructions that emerge from the patterns of lexical recurrence of its formatives at different degrees of specificity.

        Observing recent quantitative methodology in corpus linguistics (Hunston and Francis 1998 and Gries and Stefanowitsch forthcoming), we extracted from a corpus of one million words of European Spanish conversation the 5,510 collocations of verbs and NP's that were attracted to the preposition para. We coded the verbs and NPs in the construction at three levels of specificity: word-form, lemma and semantic type. Employing the Fisher exact test (Pedersen 1996, Gries and Stefanowitsch forthcoming), we ranked all the different instantiations according to significance, arriving at a model constellation of the constructions with para.

        In the discussion of results, we will show that the three most significant instantiations of the schema are at a high degree of abstraction (high type frequency): the Benefactive, Final and Allative construction illustrated in (1) to (3). At more intermediate levels of specificity, we found entrenched daughter constructions that required some additional morphological factor for their characterization. Instantiations of this type are the Temporal construction which attracts verbs inflected with future morphology or aktionsart and a daughter construction of the Perlative construction that requires the feature [low deixis] in the complement of the preposition. We will show how instantiations at this level of abstraction present challenges to the traditional categories complement and adjunct. Moreover, at the highest degrees of specificity (high token frequency), we found that the most significant instantiations belonged to the "unattached" PP subschema. These constitute independent uses of PP's also present in constructions of the constellation including a verb. This is the case of the subjective construction para mí 'for me/as for me' or the negative para nada 'not at all' illustrated in the double negative construction in (5) and independently in (5). Based on these types of subconstructions, we will argue for a revision of the received categories conjunct and disjunct. Furthermore we will relate these independent PP's to the notion of syntactic "fragment."

        Thus, the present study shows that the degree of obliqueness in prepositional phrases is a syntactic emergent property that largely correlates with their frequency of use in discourse.



(1)   Benefactive: V[type:tr.motion] P[para] NP[type:human]

        A:   à  Y le he traído estas flores para Belén

́And I brought these flowers for Belen.́

(2)   Final: V[type:free] P[para] NP[type:inanimate]

        B:       Eso quiere decir

that means

               à  que el producto ya no vale para el mercado

that the product is not good for the current market'.

(3)   Allative: V[type:motion] P[para] NP[type:location]

        A:       Mi Dama se va mañana para Lanzarote

               à  ́My "Lady" is leaving tomorrow for Lanzarote, 

... con todo el equipo

with all her stuff".

(4)   V[type:motion] P[para] NP[word-f:nada]

        A:       Allí no ladraba?

'It didn't bark there?'

Al loro Pepe

'Listen Pepe!'

               à  Que allí no ha ladrado para nada!

'It hasn't barked there at all.'

(5)   P[word-f:para] NP[word-f:nada]

        A:       Yo creo que el problema es igual aquí que en Estados Unidos.

I believe that the problem is here the same as in the United States.

        C:       Que no,


Que no Juan,


               à  Para nada      

Not at all.



Bybee, Joan L. 1985. Morphology: A Study of the Relation Between Meaning and Form. Amsterdam: John Benjamins.

Bybee, Joan. 1998. "The Emergent Lexicon". Proceedings of the Chicago Linguistic Society. 34.

------- 2003. "Sequentiality as the Basis of Constituent Structure." In: Givón, T. and Bertram F. Malle (eds.) The Evolution of Language out of Prelanguage. Amsterdam: John Benjamins.

Hopper, Paul. 1998 "Emergent Grammar" in Michael Tomasello, ed., The New Psychology of Language: Cognitive and Functional Approaches to Linguistic Structure. 155-176. Englewood Cliffs, NJ: Erlbaum.

Hunston, Susan and Gill Francis. 1999. Pattern Grammar. A Corpus-Driven Approach to the Lexical Grammar of English. Amsterdam: John Benjamins.

Langacker, Ronald. (1987). Foundations of Cognitive Grammar. Volume I. Theoretical Prerequisites. Standford: Standford University Press.

------- 2000. "A Dynamic Usage-Based Model". In: Barlow, Michael and Suzanne Kemmer (eds.) Usage-Based Models of Language. Standford: CSLI.

------- forthcoming. Constructions in Cognitive Grammar. Journal of English Linguistics.

Pedersen, Ted. 1996. "Fishing for exactness". Proceedings of the SCSUG 96 in Austin, TX, 188-200.

Stefan Th. Gries and Stefanowitsch, Anatol. forthcoming. Collostructions: Investigating the Interaction of Words and Constructions. International Journal of Corpus Linguistics.

Christopher Keiichi Schmidt

University of Kiel


The syntactic function of converbs in japanese

Japanese has been characterised as a converb-type language, i.e. a language that makes extensive use of converbs. Two types of converb constructions have been recognised in the typological literature: The first type, called conjunctional, is used for adverbial subordination, while the second type, called narrative, is employed in clause-chaining constructions. Both types are attested in Japanese: (1) exemplifies adverbial subordination and (2) clause-chaining.


(1)   Fune   ga         kekkoo   shi-te           okure-ta.

        ferry   NOM   suspend  do-CONV    be_late-PST

'Because the ferry was suspended, [I] was late.'

(2)   Koobee  it-te,             tomodachi  ni         at-te,               issho      ni      tabe-mashi-ta.

        K        to go-CONV    friend         DAT    meet-CONV   together  ADVeat-POL-PST

'[I] went to Kobe, met a friend and [we] ate together.'


There is, however, as this paper will show, a third type, which has not been recognised before. The two most commonly used converb suffixes in Japanese, -te and -i, are employed both in adverbial and clause-chaining constructions. In the third type of construction, which will be called "semantic conjunct" for want of a better term, the converb clause is not semantically dependent on the finite clause as in the case of adverbial subordination and neither do the converb clause(s) and the finite clause form a narrative sequence of events as in the case of clause-chaining. Rather, they are rather semantically independent of each other as would be the case in a coordinative construction. But since the converb clause is nevertheless syntactically dependent on the predicate of the finite clause the symmetry is only of a semantic nature. These constructions typically express a contrast as in (3), but there are also cases of simple coordination of two propositions as in (4).


(3)   Taroo wa   Amerika   ni    it-te,            Hanako    wa      Furansu  ni  it-ta.

        T.       TOPA.             to    go-CONV  H.            TOP   F.           to  go-PST

'Taro went to America and Hanako went to France.'

(4)   Jon  wa   e          o      kak-i,              haiku  o          tsukur-u.

        J.     TOPpicture ACCpaint-CONV   poem  ACC    compose-NPST

'John paints pictures and composes poems.'


This paper draws not only from the relevant linguistic literature, but also from empirical data as novels, newspapers and internet pages.

Maria-Maren Sepper

Tallinn Pedagogical University / Institute of the Estonian Language

On the development of referring adpositions on the basis of verba dicendi in standard written Estonian

Reference is part of a general semantic category called evidentiality, indicating the source of information and thereby its evidence, or reliability. According to Thomas Willett the certification of information is a semantic category, which is divided into subcategories. The main opposition is direct evidentiality versus indirect evidentiality. Usually languages distinguish between direct and indirect evidentiality, in Estonian this difference is associated with the category of mode of reporting. There are three ways of marking evidentiality: 1) morphological, 2) secondary morphological means and 3) lexical means. The distinction between these ways comes from A. Aikhenvald. She also suggests distinguishing between grammaticalized evidentiality and evidential strategy. The former expresses evidentiality with special morphemes attached to verbs. Estonian has two morphological categories that express subcategories of indirect evidentiality. These are the reported indicative mood, or the quotative (ta olevat haige ‘he is said/reported to be ill’) and the reported imperative mood, or the jussive (ta tulgu linna ‘he should come to town’).

        An evidential strategy implies the use of the linguistic means that have another, primary function beside the expression of evidentiality. The Estonian language has several constructions that can be used for expressing reported eviden-tiality, e.g. indicative pluperfect (ta oli olnud haige), the past simple form of the modal verb pidama with ma-infinitive (ta pidi haige olema), da-infinitive (ta olla haige). The meaning of all the examples is ‘he is said to be ill’.

        In Estonian the lexical-morphological means of expressing indirectness are related to verba dicendi. Constructions containing verba dicendi are handled in terms of reference, where indication to the source of information is obligatory. Usually reference is constituted as a complex clause consisting a verba dicendi as a superordinate verb, and a subordinate clause. There are several cases in world languages where the collocation of a verba dicendi and a sentence specifying the content of the communication has been grammaticalized. Estonian has a good illustration of that process with verba dicendi – the morpheme -vat marking the quotative mood has developed by way of grammaticalization of the subordinate clause.

        Over the past decades reporting has increasingly come to be conveyed by prepositional phrases. My report deals with the grammaticalization of referring adpositions on the basis of verba dicendi, e.g. ta väitis, et ‘he alleged that’ -> tema väitel ‘according to his allegation’. It would be easy to say that in Estonian the superordinator and subordinator have grammaticalized into one adposition. But this is a much more delicate case. There are several constructions that could be the source of analogy in the genesis of referring adpositions. These are

- adpositions kohaselt, järgi ‘according to’;

- a lexicalized des-syntagm from the transitive verb arvama ‘to think’;

- postpositions meelest and arust ‘in (someone’s) opinion’;

- the phrase structure of the nominalization;

- adverb sõnutsi ‘in (someone’s) words’.

Remarkably, almost all of the referring adpositions are in the singular adessive case, two of them are in the plural (andmete-l, sõnu-l) and in another context they can all be treated as a purely lexical unit.

Kaius Sinnemäki

University of Helsinki

Critical view on Hypotactic clauses and embedded clauses


Subordinate clauses are generally divided in two groups, those that are embedded and those that aren’t. There are many views on how to define what is embedded and what isn’t. One such definition is from Halliday (1985/1994). Embedded clause is defined as functioning as the head of a nominal phrase, or as a postmodifier of a nominal or adverbial phrase. A clause is not embedded if it’s in a direct relationship to another clause. Such clauses are called hypotactic clause. Since they are in a direct relationship to another clause, they are outside the boundaries of periphery.

        Based on criteria from Quirk et al (1985) it is argued that a large number of hypotactic clauses are in fact embedded. Others have found similar results based on different argumentation. The criteria here include negation, polar questions, content questions and focusing words. Their scopes extend only to the constituents of the clause. Only a small number of hypotactic clauses fall outside these criteria, namely those that function like non-restrictive relative clauses or disjunct adverbials in general.

        The behavior of hypotactic clauses in relation to coordination and embedding is also tested. Long sequences of hypotactic clauses will be compared to long sequences of coordinated clauses and to recursively embedded clauses. Based on a corpus analysis of such clauses in written Finnish, it is shown that hypotactic clauses resemble embedding rather than coordination. This is contrary to what e.g. Halliday (1985/1994) says about the nature of hypotactic clauses.

        The results show that, at least in a Finnish, a large number of hypotactic clauses should in fact be placed in the periphery of the clause and not outside of it. Some could even be placed in the core of the clause. Clauses that, nevertheless, seem to fall outside the periphery are disjunct adverbial clauses. Their syntactic status remains somewhat unclear.

Volker Struckmeier

Universität zu Köln

Sentential properties of attributive structures in German

The talk focuses on adnominal modifiers, especially attributive participles (ATPs) in German. It will be shown that German makes use of a "deep" functional structure within the DP, which allows to describe the hitherto miraculous ‘mixed’ properties of participles with regard to their word class and syntactic function. The structure of German adnominal modification receives a unified description which associates modification structures with clausal constructions.

        The typological situation with ATPs in German and English can roughly be demonstrated by the contrast in (1): German allows for a complex participle argument structure, while the corresponding structure in English is ungrammatical:


1. a)Der dem treuen Fido von den Kindern an den Schwanz gebundene Knallfrosch

    b)*The by the children onto the tail of faithful Fido attached cracker

'The cracker, that had been attached to the tail of faithful Fido by the children'

Attempts to attribute these ATP properties to their deverbal morphological make-up fail to describe their actual properties for two reasons:

        First, German ATPs show verbal properties, but only in clusters, i.e. once certain adjectival properties are used in a given structure, all verbal properties disappear. E.g., while ATPs are in principal subject to adjectival prefixation by 'un-', the rich verbal argument structure of (1) vanishes when 'un-' occurs:


2. a)Das unbepflanzte Beet    'the unplanted bed of flowers'

    b)*Der dem treuen Fido von den Kindern an den Schwanz ungebundene Knallfrosch

('the cracker, that had not been attached to the tail of faithful Fido by the children')

        Secondly, properties of ATPs do not always match properties of German adjectives in the first place: ATPs allow for time reference, rich argument structure, assignment of structural cases and so on. None of these properties hold of German adjectives. On the other hand, present participles do not allow for adjectival inflection, 'un'-prefixation etc. at all:


3.  laufend, *laufender, *am laufendsten, *unlaufend, ...


Other representations have therefore focused on verbal properties of German ATPs and argued that ATPs are in fact verbs, placed in complex syntactic structures before the head noun inside the nominal projection. However, these approaches can be demonstrated to fail to take into account that ATPs may in fact show completely adjectival properties at times.

        In my talk, I will propose a unified syntactic structure that enables ATPs to display either adjectival or verbal properties: A relative clause-like structure holds for both adjectival and verbal types of adnominal modification. One crucial assumption about this structure is that a whole bundle of its properties can be explained by assuming a phonologically zero element which resembles the overt auxiliary 'sein'. The (non-) occurence of this element receives independent confirmation, e.g. from facts about the class of elements that can be made into adnominal modifiers in the first place.

        The talk describes the specific functioning and categorial status of participles as an instance of a general structure of adnominal modification in German. This structure receives a unified description which is capable of explaining properties of participal and adjectival modification alike.

Ilona Tragel

University of Tartu

On the syntactic function(s) of the Estonian Serial Construction

The paper focuses on Estonian serial constructions (SCs): finite clauses consisting of two or three verbs describing a single event, as in


Ma          lähen           käin             toon                sulle        shokolaadi.

I:NOM   go:1P-SG    walk:1P-SGbring:1P-SG   you:ALL chocolate:PRT.

I'll go and get you some chocolate.

It is argued that Estonian SCs meet all the criteria established for a 'prototypical' SC in typological research.

        Although such constructions consist of two or three finite verbs, their syntactic function(s) is (are) the same as in case of a single finite verb. The first components in SCs - typically motion verbs (minema, tulema or käima) - show signs of grammaticalisation.

        The grammatical functions examined in the paper include the inchoative aspect, intentionality and the imperative. The paper will also discuss the question of clause borders.

Anne Vainikka

Johns Hopkins University

Adverb movement in English and Finnish


Cinque (1999) has developed a new approach to adverbs which does not rely on adjunction; rather, adverbs are base-generated in specifier positions. This approach, however, leads to a proliferation of projections. The following universal order of functional heads is proposed (in addition, AgrP and NegP are needed), several of which are only posited for adverbs:

1) T1 [absolute tense] > Modal [epist] > Mood > Modal [root] > Aspect [habitual] > T2 [relative tense] > Aspect [perfective] > Aspect [progressstrive] > Aspect-O[compl.] > Aspect-S [completed] > Voice


Applying Cinque's approach for Finnish, Manninen (1999) proposes to relax the correspondence between hierarchical structure and linear order in order to account for the apparent free word order of sentence-final adverbials.

        I propose here a variant of Cinque's and Manninen's approach: most adverbs are base-generated inside the VP, and are optionally raised to (empty) specifier positions, contra Cinque who does not assume movement of adverbs between specifier positions. The following functional heads are required for English and Finnish (cf. Mitchell 1991, Holmberg under this approach, with X being designated for certain adverbs:

2) English: Infl > X > Modal > Negation > Aspect > Progressive > Voice


3) Finnish: Agr > X > Negation > Mood > Aspect > Voice


I propose that in both languages many adverb types, in particular degree, time, and frequency adverbs, are base-generated in the VP and raise to specifier positions in the lower half of the syntactic tree (typically below NegP). This includes the Adv2 category in Holmberg (1993), with Finnish adverbs such as AINA 'always' and PIAN 'soon', as well as several of Jackendoff's (1972) categories of English adverbs. In addition, there is a class of adverbs that Jackendoff discusses -- mostly involving short adverbs without the -LY suffix such as WELL, MORE, LESS, and FAST -- that are base-generated in the VP and must remain there.

        The approach proposed here combined a minimal number of projections with the idea that adverbs occur in specifier positions. The main challenge to this approach is the MERELY-class adverbs discussed in Jackendoff (1972) which occur only in the "auxiliary" position between the subject and the finite verb. The special projection XP (in 2-3 above) is required to account for this class, as well as certain instances of sentential adverbs; furthermore, the finite verb would need to remain in X rather than raising all the way to Infl/Agr. The alternative to this analysis of the MERELY-class and other such adverbs would involve traditional adverb adjunction.

Richard Valovics

Danish University of Education, Århus

Extraposed relative clauses


In this paper I investigate the so-called extraposed relative clauses, focusing on finite relative clauses in languages, in which finite relative clauses prototypically follow their heads immediately. I define an extraposed relative clause in such languages as a relative clause that either precedes its head or is separated from it by constituents that do not belong to its head. Thus a relative clause that does not follow its head immediately but is separated from it only by material that too belongs to the head, is not of concern here.

        Extraposed relative clauses thus defined are attested in a number of languages. The most well-known examples are cases, in which therelative clause is separated from its head by a verbal particle or an infinite verb that is part of the predicate that governs the head’s NP. In such a case, the assignment of the relative clause to its head and thereby the interpretation of the structure are unproblematic as the verbal particle or the infinite verb cannot possibly be the head of the relative clause. There are, however, more extreme cases. In Danish and Hungarian, there can be NPs intervening between the relative clause and its head. These intervening NPs do not belong to the head of the relative clause; they are other clause constituents. Such constructions in these languages are seemingly hopelessly ambiguous as they are not accompanied by any special marking, intonational or otherwise, that could indicate which possible head the relative clause belongs to. Hungarian even allows relative clauses that precede their heads.

        Despite the apparently unresolvable syntactic ambiguity that extraposed relative clauses present, speakers of these languages seem to have no difficulties whatsoever with interpreting such constructions. In many cases, extraposed relative clauses even seem to be preferred to ‘ordinary’ relative clauses, which follow their heads immediately. Concentrating on these more extreme cases, I will show what makes it possible to use extraposed relative clauses and under what circumstances they are used.

Laura Visapää

University of Helsinki

To be or not to be? — On the degree of marginality of infinitive constructions without a finite verb


Infinitives are hardly ever examined as such. As the central focus of syntactic research has for long been on finite sentences, findings on infinitival structures have — for obvious reasons — accumulated as observations on the dynamics between the finite verb and its verbal complements. In my presentation, I will adopt a critical attitude towards such a perspective and argue that it loses sight of many essential aspects of infinitives. Subsequently, I will sketch as broad a picture as possible about such usage environments of Finnish infinitive structures that do not catch one’s attention when examining the data through the lenses of finiteness.

        The contexts I am interested in are centred around infinite structures which are not used as elements of finite clauses but which have a constructional figure and a function of their own (see Visapää 2002; fortch.). Intuitively, one might expect such ‘independent’ infinitives to be found in spoken interaction and in restricted contexts only (e.g. in an adjacency pair: A: What should we do today? B: Go to the beach). The conversational data studied thus far strongly suggests that spoken interaction does indeed provide a fruitful context for infinitival utterances; the infinitive-centred utterances are not, however, restricted to the context described above. There is a vast supply of infinitive structures without a finite verb in written Finnish, too.

        I will approach my data chiefly in the frameworks of Cognitive Grammar and Conversation Analysis. The analysis follows several lines. The beginning of the presentation concentrates on the marginalized status of ‘bare’ infinitive structures as well as the general tendency of peripherical phenomena being attributed qualities that are the antitheses of what are seen to be the ideal qualities of the ‘core’. I will broaden the picture with the analyses of spoken and written data: by going through some examples in detail, I aim at shedding light on the pragmatic nature of infinitives; I will also try to show that the structures at hand can be labelled marginal only in a framework that treats finiteness as a norm for all language use. I will conclude my presentation by pondering upon the concepts of marginality and mainstream, and, finally, suggest that the status of infinitives be reconsidered.


Visapää, Laura. 2002: Finiittiverbittömien infinitiivikonstruktioiden affektisista käytöistä ja rakenteesta. Pro gradu. Helsingin yliopiston suomen kielen laitos.

Visapää, Laura. forthcoming: Alastomat, inhimilliset infinitiivit. In Herlin, Ilona and Laura Visapää (eds.), Infiniittisten rakenteiden lähihistoriaa. SKS, Helsinki.


Elena Bakhmoutova

Arkhangelsk State Technical University

Interjections in terms of cognitive approach

This paper attempts to investigate the role of the interjection in the regularities observed in the cognition-language relationship.

        Since the days of Varro the interjection has been regarded in traditional grammars as a part of speech though not without reserve. According to quite a few researches interjections don’t enter the construction of sentences therefore their claim to recognition among the parts of speech is a small one.

        Indeed, any syntactic relation of the interjection is a sample of transcategorization phenomena between interjection and other parts of speech.

        An interjection belonging to a sentence can hardly be outside its structure. Otherwise we should admit that it is outside the cognitive construction which a sentence helps to set up.

        An interjection included in a sentence reshapes the expectation of relevance created in the course of the comprehension process. It cannot be structurally unrelated to the process of understanding. The impact and meaning of a sentence is dramatically changed with the interjection removed from it.

        An isolated interjection expresses the semantics of a whole sentence and can be used by speakers to induce the hearer to set up a new mental space. Nevertheless it is not recognized as a fully fledged sentence.

        So far no satisfactory definition of sentence has been developed. Those we have don’t enable us by applying them in the case of the interjection to find out whether an isolated interjection is a sentence or not. The need of new approaches is obvious.

        Mental space theory brought together with analysis of the structure of language should provide its contribution to the study of the syntactic function of the interjection. The latter in its turn would contribute to the notion of sentence and the definition of it.

Franca Ferrari-Bridgers

New York University

Luganda Initial Vowel and the CP Layer of the Grammar CANCELLED


Beghelli and Stowell (1997) assume the existence of a Referential Projection (RefP) in CP where English definite and indefinite referential DPs (QPs) covertly check their referentiality feature. Referential DPs are DPs whose referent is assumed or presupposed to exist by the speaker/context, which contrast with non-referential DPs, i.e. generics or existentials, for which LF movement to RefP is excluded. Using the validity of the above referential hypothesis as my starting point and drawing from the semantic-syntactic analyses of Bokamba (1972) and of Mould (1972),in this paper I will provide a novel syntactic analysis for the initial vocalic segment (henceforth IV) found prefixed to noun class markers of Luganda nouns (see 1).


I will argue that IV is the overt morphological expression of referentility. This predicts that IV is obligatory in referential DPs, but is absent from non-referential DPs. From the interplay between DP features [+/-referentiality] and [+/-definite] I propose two principles regarding the presence of IV in Luganda.

        I.In affirmative clauses, IV is present with referential [+/-definite] DPs. In negative clauses, IV is present only with definite DP. (see 2,3,4).

        II. When the DP is non-referential, IV is usually omitted in both affirmative and negative contexts (see 5, 6).

        >From the two principles above I will demonstrate that referential indefinite DPs are semantically impossible, because they cannot be found under the scope of negation (see 7).


By further exploring the overt nature of the feature referentiality in Luganda and building on Rizzi's (1997) CP representation, I will propose an improvement to Beghelli and Stowell's representation of RefP in the CP. In particular I will demonstrate the following.

        1. Contrary to Beghelli and Stowell's assumption RefP is not 'a sort' of TopicP, but is an independent node situated under TopicP. The data show that topicalized objects in Luganda have IV and are left dislocated (see 8) and IV is also found prefixed on non-topicalized objects occupying the canonical SVO position (see 9).

        2. RefP is located above NegP and above FocusP.

        With regards to negation I will argue against Hymes & Katamba's (1993) generalization that IV is never found in Luganda DP under the scope of negation. As the data in (10) and (11) show the presence of IV is independent from the presence of a negation. The necessity to check the referentiality feature in (11) triggers DP movement to RefP above NegP. Such a movement is not expected for non-referential DPs in (10).

        In Luganda focalization is achieved by the presence of a particle [E] preceeded by the noun class prefix of the element under focus (Ashton 1954:441). The particle [E] is here assumed to be the head of FocusP. As the data in (12) and (13) show, the focalized referential DP still retains the IV in presence of [E]. The leftward position of the DP is taken to indicate DP movement to RefP located above FocusP.

        >From the above observations I propose the following CP representation for Luganda


[ForceP [ Topic P [ RefP [XP [Focus P [XP [IP [NegP]]]]]]]]]



1. a-ba-ntu (people)

    IV-noun class prefix (ncp)-root

    e-ki-ntu (thing)


    o-mu-ntu (man)



Affirmative contexts [+REF,+DEF] and [+REF,-DEF]

2. [+DEF, +REF]

    o-mu-bba            o-y-o                           yayingiramu   nnyumba

    IV-ncp/sg thief    IV-pc-dem of ref.       entered in     the house.

    the thief this one entered in (the) house

    the thief entered in the house. [Ferrari-Bridgers 2000]

3. [-DEF, +REF]

    o-mu-bba    y-ayingira    mu  nnyumba

    a thief         entered         in    (the) house [Ferrari-Bridgers 2000]

Negative contexts only [+REF,+DEF]

4. [+DEF,+REF]

    Yokaana ta-yagala  o-mu-kazi                oyo

    John       not likes   IV- ncp/sg woman   this

    John does not like the woman    [Ferrari-Bridgers 2000]


5. Yokana  ta-yagalira   mukazi

    John       not likes       woman

    John does not like any woman   [Ferrari-Bridgers 2000]


6a.   Nnoonya                 o-musawo

        I am looking for      IV-a/the doctor [+REF]

6b.   nnoonya              musawo

      I am looking for    a/any doctor [-REF] [Mould 1972]

7. NEG [-DEF,+REF] DP is an impossible reading

The presence of a negation in a sentence such as (7) will automatically assign a non-referential reading to the DP, a thief.

7. a thief did not enter into the house = no thief entered in the house

If a thief did not enter in the house, it means that "no thief" entered the house. Logically speaking "no thief" cannot have a referent, because it does not exist.



8. o-mu-sawo          na-mu-laba

    IV- the doctor,    I- him-saw

    I saw the doctor

9. Nnoonya             o-musawo

    I am looking for  IV-a/the doctor [+REF] [Mould 1972]


10    to-gula nte [-ref]

        NEG+buy cow

        don't buy a cow

11.   E-nte to-gigula [+ref]

IV+cow Neg+buy

don't the cow [Ashton 1954]



12.   Njagala   o-mu-dumu (-focus)

        I want     IV-ncp-jug

I want the/a jug


13.   O-mu-dumu    gw-e                       njagala (+focus)

        IV-ncp-jug      ncp-focus particle  I want

        I want THE/A JUG (not ...) [Ashton 1954]

Selected bibliography


Ashton, E., et all, 1954, Luganda Grammar, Longmans, London

Beghelli, F., Stowell, T, 1997, Distributivity and negation: the syntax of each and every, in A. Szabolcsi, Ways of Scope Taking, 71-107, Kluwer, Netherlands

Bokamba, G., 1971, Specificity and definiteness in Dzamba, in Studies in African Linguistics, Vol.2, Number 3, p.217-237.

Hyman, L., Katamba, F., 1993, The augment in Luganda: syntax or pragmatics?, in Mchombo (ed), Theoretical aspects of Bantu Grammar 1, CSLI publication, Stanford, California.

Mould, M., 1972, The syntax and semantics of the initial vowel in Luganda, in Voeltz, E. (ed), Third Annual Conference of African Linguistics, p.223-231, Indiana University Press, Bloomington.

Margit Hallmägi

Predicative or subject complement in English


A copula or a link verb has the function of connecting the subject of the sentence with its complement or predicative. The subject complement in a sentence is a peripheral part.

        The predicative may be treated with respect to its form, function and meaning. The formal level may be somewhat ambiguous as far as the link verb is concerned.

        P. Christophersen and A. Sandved draw our attention to the fact “that most linking verbs may occur as non-linking verbs also:

Linking:                          Non-linking:

The man is my brother.   The man is here.” (1980 : 147)


The mistake is quite common and reveals how authors tend to mix up one case of the copula expressed by a noun brother and the other by an adverb here. So only the functional aspect is related to the syntactic level and sentence proper. And as it is generally accepted that the copula has got only a linking function between the subject and its complement, it cannot have the dual function in the same contextual framework, for it would thus contradict the laws of logic.

        There are cases with: (1) a nominal subject taking a nominal complement: I think the same goes for Fess’s mother because she knows Fess is Bess, too (p.10, Bigamist), So the novel is actually many stories with the same mysterious man as the center of each (p. 13, Bigamist); (2) an impersonal subject taking a nominal complement: The author believes it’s a reflection of the way we live now (p. 10, Bigamist), It’s the working girl’s answer to a demanding marriage (p. 11, Bigamist); (3) a nominal subject taking an adjectival complement: And the divorce was painful (p. 6, Herzog), The grass was dense and soft with the peculiar beauty it gains when ... (p. 8, Herzog); (4) an impersonal subject taking an adjectival complement: It’s painful to have to say I never loved you (p.9, Herzog), It is unfortunately true, however, that he blames me for the breakup with Madeleine and ... (p. 11, Herzog); (5) a nominal subject taking adverbial complements of: (5a) place or location: If he’s not out by ten forty-five, Íll come in ... (p. 6, Bigamist), “Well, it́s all in the book,” he says and ... (p. 6, Bigamist); (5b) moment: ..., he was about to do what she longed most to do, strike a blow)... (Herzog, p.8), What he was about to suffer, he deserved (Herzog, pp. 8-9); (5c) result: He was in love with Madeleine; ... (Herzog, p.6), She was in an ecstasy of consciousness (Herzog, p.9). The above-referred examples serve to express the meaning of the subject complement as follows: identification (examples 1-2), description (examples 3-4), restriction (5). Identification leads us to metaphor, some descriptive subject complements to simile. Restrictive use has generally been left out of research in that connection altogether, being thus a very intriguing area.


McDermott, Alice. A Bigamist’s Daughter. Perennial Library, 1988.

Bellow, Saul. Herzog. London, 1992.

Mayumi Hosono

University of Durham

Agreement as a defocalization marker – from the perspective of information structure CANCELLED

I focus on the issue of what role an agreement morpheme plays from the point of view of information structure. An agreement morpheme causes displacement (Emonds 1978, Pollock 1989, Holmberg and Platzack 1990, Roberts 1993), therefore ‘apparent imperfection’ of human languages (Chomsky 1995, 2000). In the present system (Chomsky 2000, 2001, 2002), however, it is assumed that by the operation Agree all the uninterpretable features are deleted in the original positions. Therefore, the observation that morphological marking of agreement is obligatory when a subject is preverbal, while the expression of its marking is various between languages when a subject is postverbal (Rizzi 1982, Alexiadou and Anagnostopoulou 1998, Belletti 2001) will not clearly be accounted for.

        A key solution to the problem seems to lie in Rizzi’s (1982) observations that in Italian a tensed agreement morpheme allows the “definite pronoun” interpretation (Rizzi 1982:130). His statement means that an element that a verb should agree with is presupposed, therefore defocalized in the context. Then, I suppose, originally based on Holmberg (1999), that an agreement morpheme is a defocalization marker that inherently has [-F(ocus)] and is merged to a verb. Here it is assumed that [-F] is uninterpretable (notated as [u-F]), since it has no meaning and causes displacement. I propose the following process of deleting [u-F]: (i) the feature related to the information-structural component (i.e., (interpretable) [+F]) is inherently attributed to T, and the value (i.e., [+]) of [F] is determined at the point of a derivation at which a (main) verb with or without [u-F] is canonically realized in a language (e.g., after all the French verbs or the English Aux have moved to T, or when T is merged to vP/VP in the case of the English main verb). (ii) Generalizing the notion of Agree operation (‘feature identity,’ Chomsky 2000) in the way that an element with an uninterpretable feature, whether it is a functional category or a lexical category, seeks the interpretable counterpart to delete the uninterpretable feature, an agreement morphology with [u-F] itself seeks an element with the interpretable counterpart [-F] nearest to it after [+F] is determined for the sentential components. (iii) [u-F] is deleted by the interpretable counterpart [-F] when it is c-commanded by an element that carries [-F], originally based on Holmberg (1999), ‘licensing of [-F] by [+F] in terms of government.’

        As an example of case studies, the contrast of the morphological marking between a preverbal and a postverbal subject will be accounted for as follows: the postverbal subject in (Italian) ‘Parla Gianni’ (‘Speaks Gianni’) is focalized (Lambrecht 1994, Belletti 2001), thus [+F] is assigned to Gianni and [-F] to parla. [u-F] of –a is deleted by [-F] of parl(-a), which is the nearest element that c-commands –a. The preverbal subject in ‘Gianni parla,’ on the other hand, is a topic (ibids.), thus [-F] is assigned to Gianni and [+F] to parla. Gianni moves to [Spec,TP], from where it c-commands and deletes [u-F] of –a.

Mayumi Hosono

University of Durham

Defocalization strategy in Merge – Parallelism of Clitic Left Dislocation with null object construction CANCELLED

I make an attempt to provide a unified account for both empty object construction as in Chinese and Cl(itic) L(eft) D(islocation) as in Italian, by analyzing both a null object and a clitic as the elements that are inherently defocalized in Merge. Concerning a null object, Rizzi (1986) argues that a language has an option of saturation in lexicon or saturation in syntax, and that the availability of empty objects depends on the presence of saturation process in lexicon. Huang (1984, 1991), contrary to Rizzi, argues that an empty object as in Chinese is a variable that is bound by an empty topic. Following Huang, the structure of the example ‘Zhangsan shuo Lisi bu renshi.’ (‘Zhangsan say Lisi didn’t know [him].’) is ‘[Top ei], [Zhangsan shuo [Lisi bu renshi ei]]’: an object is topicalized and moves to the initial position, in which it is deleted (Huang 1984:543).

        A clitic, on the other hand, is an element that has to be attached to a functional category (Kayne 1991). The status of a clitic in CLLD has not been clear, though. According to Cinque (1990), a clitic appears in the construction ‘Gianni, lo ho visto.’ (‘Gianni, him-Cl have-1sg seen.’), in which Gianni is topicalized. Why does a clitic show up in CLLD to begin with, though it does not appear in focalization (Cinque 1990, Rizzi 1997)?

        Remarkable is the parallelism of CLLD with the null object construction in Chinese. Both the referent that is associated with a null object and the one that is connected with a clitic have topicalized statuses. To unify their properties and seize the parallelism, I assume first that, following Rizzi’s (1986) observation that the referential content of an object pro can be recovered from the context, saturation in lexicon is ‘defocalization strategy’ in general that makes a language available of an object pro inherently assigned [-F(ocus)] in Holmberg’s (1999) sense. Second, I assume that cliticization is also defocalization strategy, and that a clitic is a defocalization marker that has a θ-role associated with the relevant element, is inherently assigned [-F], and is attached to a functional category: a clitic is base-generated in a functional category, and the canonical object position is not created. The CLLD case can be explained as follows: the role of a clitic that it plays in CLLD will be to show that a topicalized phrase is not focalized in a sentence. Therefore, it appears only in CLLD, not in focalization.

        Then, I argue that the relationship between the realization of defocalization markers and the presence of structural positions that arguments occupy will be summarized as follows: in the case in which a defocalization marker is overtly realized, a structural position of the argument is unnecessary to be projected. When a language has no overt realization of a marker, on the other hand, a structural position for an empty pronoun has to be created.

Nobuo Inui & Yoshiyuki Kotani

Tokyo University of Agriculture and Technology

What kind of words can modify? A case study of a Japanese adverb TOTEMO

Modification relations are useful for natural language processing and understanding. However adverbial modification relations are not available from usual corpora. One of the reasons is that adverbial information is not well formalized and unknown. In this paper, we explain the modification relation of a Japanese degree adverb "TOTEMO" (very, extremely). "TOTEMO" has interesting features such that: a. It can modify both adjectives and verbs. b. There are some restrictions to the modifications; negative forms of adjectives, negative forms of some verbs and positive forms of some verbs cannot be modified.

        Usual linguistic dictionaries explain that "TOTEMO" is a polysemy word: one meaning is for the emphasis of adjective, for example, "TOTEMO SHIROI" (extremely white): another meaning is for the concord to negative words, for example, "TOTEMO YOMENAI" (cannot read - at all). But our investigation shows that "TOTEMO" cannot always modify adjectives and co-occur with negative words. We model the meaning of "TOTEMO", based on the actual usages. This is useful for descriptions of dictionaries for natural language processing.

        a) Modifying adjectives: In this case, "TOTEMO" cannot modify any negative forms of adjectives. Since this adverb, different from "slowly", is unable to add a new attribute, it is necessary for adjectives to have a salient attribute. Adjectives themselves are considered to have an (salient) attribute. A salient attribute of the negative forms is not specified because there are many possibilities, for example, for a sentence ‘this flower is not beautiful’, we guess like ‘this flower is strange’, ‘this flower is pretty’ and so on. As same as an English adverb "extremely", "TOTEMO" cannot modify ungradable adjectives like "forever".

        b) Modifying verbs: In this case, positive or negative forms of verbs are possible to be modified. "TOTEMO" can only modify some kinds of verbs. For example, "TOTEMO YOMANAI" (I do not read — extremely) is not natural but "TOTEMO YOMENAI" (I cannot read — extremely) is natural. In addition, "TOTEMO YOMU" (I read — very much) is natural. From our investigation, "TOTEMO" can only modify an attribute "Ability" which is equipped with the negative forms of verbs. We consider that there is a salient attribute of each verb. "TOTEMO" can emphasize the salient attribute. For the previous example, "TOTEMO YOMU", we can consider the meaning such as "I frequently read", depending on the context. In this case, the salient attribute is "habit". Our investigation indicates that "TOTEMO" is suitable for quantitative expressions like time, frequency and so on. In this meaning, "TOTEMO YOMERU" (I can read — very much) becomes unnatural because "YOMERU" means the ability.

        From our considerations above, we construct the mathematical model of nouns, adjectives, verbs, adverbs and modification relations. Our model describes meanings of words as a kind of vector expressions. "TOTEMO" indicates a salient attribute of each word and can modify words which have a salient attribute. This model explains how and what word can be modified by "TOTEMO". We showed that "TOTEMO" is not a statement adverb but a degree adverb in all cases, in the meaning that "TOTEMO" restricts the verbs modified. "TOTEMO" is an interesting adverb because the modified adjectives and verbs are very restricted. We conclude that a word dictionary should contain descriptions of abilities and salient attributes for each verb and adjective to handle adjectives like "TOTEMO".

Petri Jääskeläinen

University of Jyväskylä


Shadow argument adjuncts with instrumental verbs in Finnish


In this poster, I will focus on the relationship between denominal verbs and shadow argument variation, which I analyse in the light of Finnish instrumental verb derivation. As a semantic class, instrumental verb is a denominal verb base or cognate noun that refers to an instrument, as in the noun-verb pair saha 'a saw' : sahata 'to saw'. On a conceptual level, these types of Finnish noun-verb pairs are similar to the English noun-verb conversion (a knife - to knife).

        Although instrumental verbs lexically conflate in their meaning the instrument used, the instrument of the activity (which is associated to the cognate noun) can be expressed as verb's adjunct in adessive case, e.g. sahata veitsellä (saw+INF knife+ADE) 'saw with knife' in cases where the semantic content of the shadow argument (saw) is fused with the semantic content of the adjunct nominal phrase (knife) (Pustejovsky 1995). This fusion between verb's cognate noun and adjunct noun can vary from hyponymy and synonymy to co-hyponymy and hyperonymy or between different semantic domains, e.g. keihästää puukeihäällä 'to spear with a wooden spear'~ piikillä 'with a thorn' ~ sarvella 'with a horn' ~ haarukalla 'with a fork' ~ mailalla 'with a stick' ~ kainalosauvalla 'with a crutch' ~ eturaajoilla 'with a foreleg' ~ katseella 'with a look; eye', etc.

        In the literature (e.g. MaCawley 1976, Green 1973, Ljung 1977) it has been noted that the broad variation scope of the instrumental adjunct makes it difficult to establish an exact definition for denominality. If we can hammer with a stone, can the hammer, then, be considered as the base or motivation for the meaning of the verb? Kiparsky (1997) has suggested that the quality of shadow argument variation in English correlates with the denominal interpretation of the verb. That is, if the adjunct noun can only be a hyponym or a synonym for the verb's cognate noun, then the verb is a true denominal instrumental verb (e.g. to tape). If, however, the adjunct can have a broader meaning than the cognate noun, then verb is a pseudo-instrumental verb (to hammer).

        However, the analysis of ca 220 prototypical instrumental verbs (and their counterpart nouns) listed in The Basic Lexicon of Finnish ("Suomen kielen perussanakirja") and the respective contextual occurrences thereof in the computer-based corpus Textbank of Finnish ("Suomen kielen tekstipankki") does not support Kiparsky's classification: neither does the historical derivational relationship (real denominal derivation vs. back-derivation, loan relationships, or root-associated word clusters) nor the synchronic-syntactic derivational relationship between a cognate noun and a verb correlate with the qualitative borders of the adjunct.

        On the basis of the Finnish data, I shall argue for the following: 1) The scope of the variation doesn't imply denominality as such, and the variation for the whole verb class can not be reduced to one explaining factor only. 2) Instead, the process of argument fusion, seen as conceptual integration (blending) (Coulson - Matlock 2001), positions verb's different uses on a continuum from literal to non-literal meaning (literal meaning - literal mapping - metaphor). 3) On this account, the motivation of the individual variation is based on the verbs flexible potential, which is motivated by the canonical tool use scheme (instrumental construction), by the functional category of the instrument nouns (analogical applicability), and by the nature of the activity referred. This results in idiosyncrasies and non-predictability, and can lead to semantic changes and lexicalization (from flexible meaning to polysemy; cf. Fellbaum 2000).




Coulson, Seanna - Matlock, Teenie 2001. Metaphor and the space-structuring model. - Metaphor & Symbol 16(3).

Fellbaum, Christiane 2000. Autotroponymy. - Ravin, Yael - Leacock, Claudia (eds.). Polysemy. Theoretical and Computational Approaches. Oxford University Press.

Kiparsky, Paul 1997. Remarks on Denominal verbs. - Alsina, Alex - Bresnan, Joan - Sells, Peter (eds.). Complex predicates. CSLI Publications.

Pustejovsky, James 1995. The Generative Lexicon. MIT Press.

Galina Kalmykova

Ulyanovsk State Pedagogical University


Core and Periphery of Noun Phrases with Cause-Effect Meaning

In the paper I research causation as type of cause-effect relations. In the focus of my attention are noun phrases with cause-effect meaning. The investigation is fulfilled on the comparative basis of German and Russian. I argue that noun phrases in the function of adverbial modifier have their own core and periphery. The distinction criterion is the degree of their dependence on contextual environment.

        The means for core representation are noun phrases with prepositions aufgrund, vor, von, wegen, dank, aus, durch, mit and their Russian equivalents. For example: wegen der Krankheit (uz za bolezni — because of illness). In the sentence this noun phrase represents the core causative meaning: Wegen der Krankheit war sie abwesend. As peripheral means serve noun phrases with prepositions auf, bei, unter, über, noch, zu in German and their equivalents (with/ or without prepositions, but by means of case system) in Russian. Such phrases can realize their causative function only in some definite contextual environment or in some definite sentence models. In the sentences: Sie war froh über diese Aufmerksamkeit  (ona radovalas’ etomu vnimaniju - She was glad about this attention.) Preposition über in German or Dative in Russian do not correlate with causative function. It arises from the context. Or: Es ist schwer, ihn in diesem Anzug  zu erkennen. (ego trudno uznat’ v etom kostjume. — It is difficult to recognize him in this suit.). In this case the function of adverbial modifier is not obligatory, but it’s resulting from the context.




Helbig, Gerhardt & Buscha, Joachim (1994) Übungsgrammatik Deutsch. Langenscheidt, Verlag Enzyklopädie.

Zolotova G.A. (1998) Communicative Grammar of Russian. Moscow: MGU.

Kalmykova G.A. (2002) Means of “unmarked” causality in German// In: Theory and Practice of Germanic and Romance Languages. - Ulyanovsk: Pedagogical University.

Elena Sergeevna Kokanova

English sentence adverbs in the framework of the theory of mental spaces

The aim of this poster is to show the relevance of the so-called peripheral elements for the interpretation of essential properties of language.

In linguistics there is interest to the units of language which are hardly entered in the traditional classification of parts of speech. It is possible to attribute sentence adverbs to such units. English has a subclass of adverbs, called sentence adverbs, which are rather different from ordinary adverbs. While "ordinary adverbs" describe some aspect of the action, the sentence adverbs express the speaker's view of the whole sentence and modify the whole sentence. The attempts to describe sentence adverbs by foreign and Russian grammarians are characterized by a significant variety of opinions at the essence of the given class of words. Till now there is no precise definition of this class of words. Traditionally, Russian authors referred the given class of words in the English language to the modal words and examined them as a separate part of speech [Kovner 1951, Vinocurova1954, Zhigadlo, Ivanov, Iophik1956, Ganshina, Vasilevskaya1964, Smushkevich1968, Ilish1971, Ivanov, Burlacova, Pochepzov 1981].

Russian linguists raised the problem of the existence of modal words in Russian for the first time( See Vinogradov 1947). We haven't come across the term "modal words" in the foreign grammars. The question about their detachment into a separate part of speech was not discussed either. The authors of foreign grammars single out the given group of words in a special group of adverbs, defining them as "sentence modifiers" [Sweet 1891, Poutsma 1928], "sentence adverbs" [ Kruisinga 1925, Curme 1957, Zandvoort 1966, Schreiber 1973, Bellert 1977, Trask 1993], "sentence adverbials" [ Katz, Postal 1964], "independent adverbs" [Smart 1946], "stance and linking adverbs" [ Biber, Conrad, Leech 2002]. As far as we know, the term "sentence adverbs" has existed since the beginning of the twentieth century. For the first time it appeared in E.Kruisinga's grammar.

In spite of the long-term practice of the research of the given class of words, the term 'sentence adverbs' is not fixed in more or less widely accepted and recognized frameworks in linguistics. Besides, we find traditional ways of examining sentence adverbs are not completely sufficient and we try to review them from the point of view of cognitive linguistics, or more narrowly from the point of view of the theory of mental spaces which is applicable to the analysis of the given class of words. According to E.Sweetser and G. Fauconnier's opinion, the theory of mental spaces provides a model of the connection between semantics and cognition which allows us to address these important issues and offers theoretical concepts intended to account for the regularities observed in the cognition-language relationship. Syntax and grammar are involved in the expression and formation of human understanding of the world. Designs of mental spaces meet various interpretations of a sentence in a context. Statements are understood and interpreted to the relatively developed designing of mental spaces which represent dynamic cognitive formations. Creating an adequate mental space is understanding. A speaker forms new mental spaces and changes them according to the directions given in the discourse. Mental spaces are not incorporated as ready structures in our consciousness, and during the generation of discourse a new one appears each time. Language has special devices of creation and construction of mental spaces. Grammatical markers are so-called "traffic signs" which direct a listener and a speaker by a general mental trajectory. We can apply the theory of mental spaces to the analysis of sentence adverbs which can act as builders of mental spaces or as connectors. If the first ones (typically, actually, unfortunately, etc) modify mental spaces, the former (alternatively, anyway, thus, etc.) connect them. Sentence adverbs are so- called navigators in discourse, their function is to direct an interlocutor in it. They are convenient, compact. But, at the same time, they are significant. The theory of mental spaces is a rather interesting direction which allows in a new way to look at the language status of various parts of speech, including sentence adverbs analyzed here. If we take sentence adverbs, traditionally, they have an optional, peripheral meaning from the point of parts of the sentence. They do not enter the predicative minimum of the sentence. However, in the framework of cognitive linguistics we can reevaluate their status in English. Over the years they have come to possess high frequency value which is the reflection of their increased cognitive load. We can note the tendency of appearing new sentence adverbs which number is growing (historically, linguistically, economically, etc.) For example: Linguistically, babies and animals are 'borderline beings, who may not entirely qualify as 'persons'. From the point of view of linguistics, babies and animals are 'borderline beings, who may not entirely qualify as 'persons'. Thus, a sentence adverb gives a convenient way of packing information, an economical way of bringing thoughts to an interlocutor.

Liina Lindström

University of Tartu

Non-clausal elements affecting the word order in Estonian


This paper discusses the status of two Estonian words: 'äkki' and 'tegelikult'~'tegelt'. Both of them are adverbs in their literal meaning, but additionally they have other usages which are similar to discourse particles and which are very common especially in spoken Estonian:

äkki: lit. 'suddenly', manner adverb

Äkki        teatas                       Igwe,   et     on           tekkinud      väike probleem.

Suddenlyannounce-IPF-SG3 Igwe    that  be-SG3   arise-PTCL little   problem

'Suddenly Igwe announced that a little problem has been arosen.' (AJAE1990\ee0371)

Other usages: 'maybe', 'I hope'. It is often used in propositions.

A:no       siis   teeme      `nii  et     tulete           äkki `homme       tagasi. (CONV)

    PART then do-PL-1   so    that  come-PL-2 äkki tomorrow    back

    'we will do so: you will come back tomorrow'


tegelikult: lit. 'actually' ('practically', 'really')

/... /  et    ta         tegelikult  ei    `ole         mees. (.) (CONV)

        that he/she  actually    not  be-NEG  man

'that she is not actually a man.'

Other usages: contrasts the following clause to the preceding clause

K:ühel        `heal            päeval     avastab=et               $ tegelikult    võiks

    one-ADE good-ADE  day-ADEfind out-SG-3=that     tegelikult    shall-KND-SG-3

    ikka    `naine  süia          teha.           $  heh (CONV)

    PART wife     food-PRT make-INF       heh

    'one nice day it occurs to him that the wife should cook'

An important question is the treatment of such elements: are they immediate constituents of the clause or not?

        Sometimes such elements are called sentence adverbials; Longman Grammar of Spoken and Written English (Biber et al. 1999) calls them stance or linking adverbials. In Estonian linguistics, they are not regarded adverbials and therefore they are not the constituents of the clause (EKG: 99-102). I will call them non-clausal elements.

        These elements may occur in the initial, middle or final position of the clause and usually they do not affect the word order. But sometimes it may happen that in the initial position of the clause they may cause the inversion of the subject and predicate ( similarly to other adverbials):


A:aga  `äkki   oled         sa    nüüd    kuidagi:   sest      `välja  kasvand. (.) (CONV)

            äkki     V             S

    but  maybe be-SG-2  you now out      grow-PTCL

    'but maybe you have grown out from this by now'


        There is also another question of terminology. When talking about the order of subject, object, predicate and other elements in the clause, usually two terms are used: 'word order' and in recent studies 'constituent order'. An Estonian linguist Peep Nemvalts has suggested a term 'phrase order' (Nemvalts 2002: 190). In this case, the older term 'word order' is probably the most appropriate.




Biber, D., S. Johannson, G. Leech, S. Conrad, E. Finegan 1999. Longman Grammar of Spoken and Written English.

EKG =Erelt, M., R. Kasik, H. Metslang, H. Rajandi, K. Ross, H. Saari, K. Tael, S. Vare 1993. Eesti keele grammatika II. Süntaks. Lisa: Kiri. Tallinn 1993.

Nemvalts, Peep 2002. Keeleteaduse teaduskeelest. In: R. Pajusalu, I. Tragel, T. Hennoste & H. Õim (eds.). Teoreetiline keeleteadus Eestis. Tartu Ülikooli üldkeeleteaduse õppetooli toimetised 4. Pp. 179-192.

Kadri Muischnek
University of Tartu

Verb argument or a part of a multi-word verb?

The aim of this paper is to describe the fuzzy border between a nominal component of a multi-word verb and a noun belonging to the argument structure of a verb in Estonian. There are three main types of multi-word verbs in Estonian:

1) verb + adverb combinations (phrasal verbs)
2) verb + noun combinations
3) verb + verb combinations, the second verb is infinite

In addition to this, some serial verb constructions have also been described (Tragel 2003).

In my poster I will concentrate on the Estonian multi-word verbs consisting of a verb and a noun, and call this combination an expression verb. A prototypical expression verb is an idiom, so its meaning is not the sum of the meanings of its components. For example the Estonian expression verb jalga laskma consists of the noun jalg 'leg' in the partitive case and the verb laskma meaning 'to let' as well as 'to shoot'. But the combination as a whole has the meaning 'to make off, to run away'.

Besides such prototypical expression verbs there are various types of multiword expressions that share at least some common features with the expression verbs. Several authors, e.g. Howarth (1998) and Nenonen (2002) have described a continuum model that has an idiom at the one end of an axis and a free combination at the other end of it. In the case of the verb + noun combination the continuum would be something like this: expression verb (e.g. an idiomatic noun + verb combination) -- support verb + noun combination -- collocation -- free combination.

In order to describe the distribution of various verb + noun combinations on this axis, one needs to define the main features of a prototypical expression verb. Based on Rätsep (1978) and Moon (1998), the following criteria are used to define a prototypical expression verb:

1. The expression verb is (synchronically) structurally opaque. So the syntactic analysis of such constructions doesn't make any sense. For example, in the expression verb

aru saama
mind-PART get
'to understand'

one can't analyze the noun aru as an object (although it formally has all the qualities of an object); the semantic object of this expression verb as a whole should be in the elative case.

2. The combination is unique, only one noun can belong to this particular verb and only one verb can be used with this particular noun, no synonyms can act as substitutes.

3. Some lexicogrammatical fixedness, restrictions on aspect, mood, voice etc have also been listed among the qualities of a multi-word expression (cf. ). In Estonian there seem to exist the limits on aspect only.

4. The morphological parallel forms of a noun can't be used as substitutes in an expression verb. E.g. the noun silm 'eye' has 2 variants of an illative case, but in the expression verb silma torkama 'to attract attention' only one of them can occur.

5. The noun in the expression verb can?t be modified, e.g.

*laskis vasakut jalga
let/shoot-3.SG.PAST left-PART leg-PART

In my poster I will focus on describing the existence and gradual loss of these features along the expression verb -- free combination axis.


Howarth, P. (1998) The Phraseology of Learner's Academic Writing. In Phraseology: Theory, Analysis and Applications. Ed. By A. P. Cowie. Oxford: Oxford University Press
Moon, R. (1998) Frequencies and Forms of Phrasal Lexemes in English. In Phraseology: Theory, Analysis and Applications. Ed. By A. P. Cowie. Oxford: Oxford University Press
Nenonen, M. (2002) Idiomit ja leksikko. Lausekeidiomien syntaktisia, semanttisia ja morfologisia piirteitä suomen kielessä. Joensuun Yliopiston Humanistisia Julkaisuja N:o 29. Joensuu: Joensuun Yliopisto
Rätsep, H. (1978) Eesti keele lihtlause tüübid. Tallinn: Valgus Tragel, I. (2003) Eesti keele tuumverbid. Dissertationes Linguisticae Universitatis Tartuensis 3. Tartu: Tartu ülikooli Kirjastus

Séverine Vienney

University of Franche-Comté

A method to find the borderlines between “the core” and “the periphery” in French sentences - application to an automatic grammar correcting system


The aim of this paper is to present a method to find the borderlines between the "core" and the "periphery" of a French sentence. The purpose is to simplify the sentence to the maximum in order to keep only the obligatory constituents and to facilitate its analysis. This analysis is situated at the syntactic level. The grammatical categories of the elements of the sentence are studied in order to detect which ones must be separated. Firstly the method and its concepts are described. The system implementing the method carries out a partial analysis of the sentence. It is argued that such an analysis is completely satisfactory to solve limited problems, as is in the case of the specific application which is then examined, this being the automatic grammar checking/correcting of the past participle or the verb in general. The principal goal consists in fact in finding the subject (and in some cases the object) of the verb. This thus requires finding the "central" elements of the sentence. Perspectives of the method are discussed as to whether it could be applicable to other problems than automatic grammar correcting or to other languages than French.

Back to the symposium main page