Norms, language-specific meanings and schemas. A Coserian perspective on contemporary concepts in linguistics
“Jedes Ausgesprochene bildet das Unausgesprochene, oder bereitet es vor.” (Wilhelm von Humboldt)
Eugenio Coseriu’s (1921–2002) seminal contributions to linguistic theory went largely unnoticed in mainstream linguistic research in the English-speaking world. This paper discusses three concepts that have loomed large in present-day linguistics and were already addressed by Coseriu. The paper focuses on the question whether Coseriu’s contributions might be a stepping stone to establishing synergies with current strands of linguistic research, with a view to determining the topicality of Coseriu’s approach to concepts that are at the centre of ongoing discussions. The three concepts dealt with in the paper are: 1) Coseriu’s theory of language norms against the backdrop of the (Neo-)Gricean pragmatic theory of generalised conversational implicatures and default inferences; 2) Coseriu’s detailed exposition of schematic meanings in syntax and word formation as compared to cognitive approaches within the frameworks of Construction Grammar and Construction Morphology; 3) Coseriu’s elaborate account of general language-specific meanings (G. “Bedeutungen”, Fr. “signifiés”, Sp. “significados”) in light of recent developments in psycholinguistic research which focuses on the role of underspecification in meaning representations. The paper concludes that Coseriu’s work has the potential to offer solutions for a number of as yet unresolved issues in the study of pragmatics, syntax, word formation and semantics.
Eugenio Coseriu, (Neo-)Gricean pragmatics, Construction Grammar, Construction Morphology, norms, normal language use, encoded meaning (signified), sense, designation, semantic underspecification (indeterminacy), schema, construction, syntax, word formation, pragmatics.
Las contribuciones fundamentales de Eugenio Coseriu (1921-2002) a la teoría lingüística pasaron desapercibidas, en gran medida, en la línea principal de la investigación lingüística en el mundo anglosajón. En este artículo se analizan tres conceptos que han cobrado importancia en la lingüística actual y que han sido ya abordados por Coseriu. Así pues, el presente trabajo se centra en la cuestión de si las aportaciones de Coseriu podrían ser un trampolín para establecer sinergias con las corrientes actuales de la investigación lingüística, con el fin de determinar la actualidad del enfoque de Coseriu en relación con conceptos que están en el centro de las discusiones actuales. Los tres conceptos que se abordan en esta contribución son: 1) la teoría de Coseriu sobre las normas lingüísticas con el telón de fondo de la teoría pragmática (neo)griceana de las implicaturas conversacionales generalizadas y las inferencias por defecto; 2) la exposición detallada de Coseriu sobre los significados esquemáticos en la sintaxis y la formación de palabras en comparación con los enfoques cognitivos dentro de los marcos de la Gramática de construcciones y la Morfología de construcciones; 3) la detallada exposición de Coseriu sobre los significados generales específicos de la lengua (al. “Bedeutungen”, fr. “signifiés”, esp. “significados”) a la luz de los recientes desarrollos en la investigación psicolingüística que se centran en el papel de la subespecificación en las representaciones semánticas. El artículo concluye que el trabajo de Coseriu tiene el potencial de ofrecer soluciones a una serie de cuestiones aún no resueltas en el estudio de la pragmática, la sintaxis, la formación de palabras y la semántica.
Eugenio Coseriu, pragmática (neo)griceana, gramática de construcciones, morfología de construcciones, normas, uso normal de la lengua, significado, designación y saber de las cosas, subespecificación semántica (indeterminación), esquema, construcción, sintaxis, formación de palabras, pragmática.
Contemporary linguistic research is characterised by a multitude of approaches which differ in many respects: theoretical and epistemological assumptions, emphasis and dependence on data
orientation, methodology, openness towards interdisciplinarity, and so forth. Integrationist approaches emerge every now and then but do not seem to gain wide acceptance among scholars, which seems in part due to the complexity of the objects of enquiry that have been identified in modern language studies.
Coseriu’s conception of “Integral Linguistics” may be considered as one of the earliest significant, but at the same time internationally relatively lesser-known, endeavours at establishing a framework that encompasses a large number of linguistic subdisciplines under a single heading. A characteristic feature of Coseriu’s integrationist approach is that he insisted on the importance of striking a balance between empirical investigations, linguistic theory and the philosophy of language. The guiding principles of this approach, for which Coseriu draws on a comprehensive perusal of the history of the language sciences, are summarised in Willems and Munteanu (2021: § 2). One important corollary of Coseriu’s integrationist approach is that new developments in the language sciences must find anchorage in the “Integral Linguistics” approach in a principled way, even if the practical details of every new approach cannot be specified in advance.
In this paper, I confront three concepts that have loomed large in present-day linguistics with analyses found in Coseriu’s writings. The paper is accordingly organised into three major sections. My aim is to establish ways in which current debates can benefit from Coseriu’s analyses, which, as I will argue, have much to offer because of their internal consistency and their reliance on a synthesis of empirical analysis and theoretical explanation. The three concepts the paper focuses on are: Coseriu’s concept of language norms against the backdrop of the (Neo-)Gricean pragmatic theory of generalised conversational implicatures and “default inferences” (Section 2); Coseriu’s concept of schematic meanings in syntax and word formation as compared to cognitive approaches within the frameworks of Construction Grammar and Construction Morphology (Section 3); and Coseriu’s concept of general language-specific meanings or “signifieds” (G. “Bedeutungen”, Fr. “signifiés”, Sp. “significados”), which are indeterminate, in light of recent developments in experimental psycholinguistic research that focuses on the role of semantic underspecification in meaning representations (Section 4). Section 5 concludes the paper.*
2. Coseriu’s theory of “normal language use” and (Neo-)Gricean pragmatics
2.1 Coseriu was sceptical about linguistic pragmatics, specifically historical pragmatics. Kabatek (2021: 234) relates the anecdote that Coseriu congratulated his former student Brigitte Schlieben-Lange on the occasion of the publication of her book Traditionen des Sprechens Elemente einer pragmatischen Sprachgeschichtsschreibung (1983), saying that she had written an “excellent book on an issue that does not exist”. Kabatek (2021) goes on to show why “historical pragmatics” seems to be something of an inconsistency from a Coserian perspective A number of conceptual distinctions must be observed. That which is called pragmatic in the literature usually is either the unique utterance of a speaker or the whole of non-language- specific properties of language in general (designation), hence either the strictly individual or the universal but not the historical level of language-specific structures and functions. However, that which is individual or universal in language does not, in Coseriu’s theory of language, qualify as historical. Only a specific language and the corresponding language-specific knowledge a speaker possesses of a language can be said to be historical (Kabatek 2021: 234). Coseriu of course acknowledges that every individual act of discourse is historical to the extent that it partakes in the history of a language and of the discourse traditions that can be found in a linguistic community (Coseriu 2007 : ch. 2; 1985). At the same time, every individual act of discourse cannot be repeated exactly in the same form and with the same meaning.
The threefold distinction between the universal, the language-specific and the individual is one of the cornerstones of Coseriu’s “Integral Linguistics” approach. For reasons of space I cannot discuss this distinction here in detail. I refer the reader to Willems and Munteanu (2021: 7-9) for a brief presentation. I here only provide Coseriu’s matrix of linguistic competence, in which he combines the three different levels of language with three different perspectives, three kinds of meaning and three types of judgment in view of a comprehensive, integrationist approach to of linguistic competence (see Coseriu 1975 [1955–1956], 1985, 2007 , 2007 ).
Figure 1: Coseriu’s matrix of linguistic competence
For the sake of clarity, I translate the German terms “Bedeutung” and “Sinn” not as “meaning” and “sense” (cf. Coseriu 1985: xxxiv) but as “signified” (or “encoded meaning”) and “text meaning”, respectively. “Meaning” and “sense” are ubiquitous terms in various linguistic frameworks but they do not normally denote what Coseriu means with “Bedeutung” (language-specific meaning) and “Sinn” (content on text and discourse level) (see Willems and Munteanu 2021: § 6 for discussion). “Signified” is the English translation of F. “signifié”. “Signifié” was introduced by Saussure (1968: 251-276) and also Coseriu’s favourite term to refer to “language-specific meaning” (Coseriu 1987a, 2001, etc.). In this paper, I will use “encoded” to refer to the level of meaning that is part of the lexicon and grammar of a specific language, following the common use of “encoded” in Gricean pragmatics, where it contrasts with “inferred” (Levinson 2000 and Belligh and Willems 2021)1. The curved arrow in Figure 1 serves to indicate that although language-specific idiomatic knowledge undergirds the activity of speaking in a particular language, idiomatic knowledge is at the same time continuously being created in this very activity. In the same vein, the round brackets (abstracted particular language) indicate that a specific language cannot in reality be found as a “product”, but a grammar and a lexicon of a language can nevertheless approximate a specific language conceived of as érgon (Coseriu 2007 : 75).
2.2 No less important than the distinction between three levels of language and three complementary perspectives on language is another threefold distinction which must be made according to Coseriu in order to further specify the historical level of language. Coseriu’s distinction between language system (Sp. “sistema”, G. “System”), norm (Sp. “norma”, G. “Norm”) and discourse (Sp. “habla”, G. “Rede”) builds on insights from Copenhagen and Prague structuralism (Willems and Munteanu 2021, Willems and Belligh 2022).2 The basic
tenet of Coseriu’s notion of a tripartite language “structure” (Coseriu 2007 : 262-272) is that the Saussurean dichotomy between “langue” and “parole” is phenomenologically inappropriate because it is unable to capture all those aspects of language use that are neither purely individual (“parole”) nor part of the language-specific idiomatic system (“langue”) but traditional realisations of what the system provides for as language-specific possibilities.
Recognising the role that traditional, historically grounded practices play in discourse (both oral and written) is one of the merits of an approach that critically reviews and further develops structuralist thought. Coseriu’s concept of “norm” is applicable to phonology, lexicon, word formation and syntax (Coseriu 1975 : ch. 5-6; cf. 2007 : 49-55) and has proven to be particularly productive with regard to questions about semantics, which is the main focus of the ensuing discussion.
Habitual interpretations of words and word formations can be so widespread and so firmly stored in speakers’ linguistic competence that they are easily mistaken for systematically encoded signifieds. For example, in German Straßenhändler normally designates a street vendor while it could equally well be used to refer to a trader of streets. It is on the basis of extralinguistic elocutional knowledge that the latter interpretation is unlikely, but there is nothing in the word formation’s encoded semantics that bars it. Similarly, G. Goldwaage is normally used to designate a gold scale rather than a scale made of gold (Coseriu 1970, 1977b and 2007 : 109-111, 261; Dietrich 2021 provides a wealth of further examples in Spanish and French, partly also in comparison to English).
Simple lexical items, too, can only be adequately described semantically, and especially lexicologically, if the intermediate level of normal language use is taken into account. French âgé without further specification is often used to refer to ‘elderly’ or ‘old’ as in une personne
âgée ‘a senior’, but the adjective itself only indicates the years of age (compare un garçon âgé de douze ans ‘a twelve year old boy’). Similarly, lying often entails that the person who is lyingknows that what he says is false, but the verb to lie only means ‘to present as true something that one considers to be false’, irrespective whether it actually is objectively false or not (which can be beyond the speaker’s knowledge, and even if it would turn out that what someone is saying is the truth, s/he would still have been lying if s/he believed that what s/he was saying was not true; Coseriu 2000 : 27-28). In the same vein, Kabatek (2000: 197-198) points out that G. Maus ‘mouse’ does not have two systematic signifieds. The word for the small mammal and the word for the computer device are actually one and the same word with a single encoded meaning. Even if the computer mouse is the usual interpretation of Maus for many speakers nowadays, it is still true that in the German language this interpretation is based on a
transparent metaphor that has been consolidated as a conventional usage variant in German normal language use.
2.3 The distinction between language system, normal language use and discourse is of central importance to the theory of grammar. For instance, the common difference in interpretation of the subject in transitive sentences like
(1) John broke the window.
(2) A hammer broke the window.
is not specified by English syntax but a matter of normal language use. It is common for a person (John) to be the Agent of an activity such as breaking something, just as it is common to regard a hammer as the instrument of an activity or event, but these are specifications based on the inherent semantics of John and hammer, not on the subject function in (1-2), which happens to be the same in both sentences (Coseriu 1970: 107-109; Höllein 2021: 266-268). This means that (1-2) do not exclude interpretations in which, e.g., John is an instrument in (1) or in which a hammer is represented as a living being in (2). The normal interpretations of (1-2) are not tied to the grammar of English but to designation in general, and hence not to a specific language community but to a “community of experience” in general defined without recourse to any individual language (Kabatek 2000: 196). For every speaker in whatever language it is the case that you don’t normally use a person named John as an instrument to break a window.
From an “Integral Linguistics” point of view, it is important to emphasise that with analyses in terms of “thematic roles” rather than “semantic roles” (Willems 2020: 153-157) we are not concerned with the level of any particular language, but with the level of “speaking in general” and the corresponding “elocutional knowledge” that in principle all speakers / hearers share (Coseriu 1985).3 Languages differ lexically and grammatically, and what is left to inference and normal language use (viz., normal interpretation) in one language may be a matter of grammar in another language, and then the corresponding analyses must be different. For example, in an ergative language like Hindi, the animate subject of a sentence with the transitive verb ToD-naa ‘to break’ in the perfective is in the ergative case (3). A corresponding sentence with an inanimate subject like ‘a hammer’ in (4) would not only be unusual, but the ergative
case would suggest a certain volitionality attributed to the hammer normally associated with
animate Agents. Therefore, an altogether different syntactic construction (5) is usually chosen instead of (4). With the intransitive verb TuuT-naa ‘to get broken’ in (5), ‘the hammer’ is in the instrumental case (hathauDa=se) and the subject khiDakii ‘the window’ is in the unmarked nominative case (De Hoop and Narasimhan 2008: 66). Hindi can therefore be said to possess an encoded “semantic” role “Agentive”, whereas in a language like English the “thematic” role “Agent” is not grammatically encoded as “Agentive”.4
(3) LaDakii=ne khiDakii=ko toD di-yaa
boy=ERG window=ACC break give-PFV.SG.MASC
‘The boy broke the window.’
(4) ??HathauDa=ne khiDakee=ko toD di-yaa
hammer=ERG window=ACC break give-PFV.SG.MASC
‘The hammer broke the window.’
(5) KhiDakee hathauDa=se TuuT ga-yii
window-NOM hammer=INS break go-PFV.SG.FEM
≈ ‘The window broke due to the hammer.’
2.4 A similar reasoning applies to certain facts of word order, which must either be accounted for in terms of language-specific meaning (idiomatic knowledge) or extralinguistic designation (elocutional knowledge). For example, Coseriu (1987b: 253) points out that the contrast between French Elle se maria et eut un enfant (‘She married and had a child’) and Elle eut un enfant et se maria (‘She had a child and married’) is not language-specific but a contrast that concerns speaking in general, i.e. in whatever language, since it is generally the case that the order of events or actions in reality can be represented iconically by the order of clauses, no matter what language one speaks. A different analysis is required with regard to languages in which clauses regularly encode the temporal order of the sequence of events. I return to this point below (§ 2.11).
The aforementioned examples from various linguistic domains (lexicon, word formation and syntax) suffice to show that Coseriu’s account of linguistic meaning is firmly based on the insight that what is said does not coincide with what is meant, i.e. expressed and understood.
Already in the 1950s, Coseriu stresses the importance of this observation: “at every moment, what is actually said is less than what is expressed and understood”.
2.5 I now turn to present-day pragmatic research. The currently most influential strand in linguistic pragmatics is arguably the one based on the work of the British language philosopher Herbert Paul Grice. Grice’s principle of cooperation and theory of conversational maxims have been adopted by many linguists, who partly further developed and partly revised it. While Post- Gricean pragmaticists – especially Sperber and Wilson (²1995) and those adopting the premises of “Relevance theory” – use Grice’s approach as a stepping stone to develop an altogether different theory of pragmatics, Neo-Gricean pragmaticists aim to deepen Grice’s theory in a way that is fundamentally faithful to Grice’s initial conceptions. This is particularly clear in the work of Stephen Levinson (2000).
In my view, Sperber and Wilson’s Relevance theory is altogether incompatible with Coseriu’s theory of language. Virtually all of the basic assumptions of Relevance theory are inconsistent with Coseriu’s approach to language: the theory is based on a number of a- historical logical and psychological concepts, it gives pride of place to cognitive efficiency (maximizing relevance) at the detriment of linguistic creativity and identifies most types of meaning with universal, atomic concepts or pro-concepts in the sense of Fodor’s “language of thought” (or “mentalese”, Fodor 1975, 2008). The theory of language undergirding Relevance theory is at odds with Coseriu’s theory of language to such an extent that any agreement between both seems unlikely. Still, considering the worldwide success of Relevance theory and the countless studies at least partly based on the theory which continue to appear, explaining why the differences seem largely insurmountable is among the challenges that will have to be addressed in the future.
2.6 Neo-Gricean pragmatics, in particular as developed by S. Levinson (2000), offers considerably better conditions for rapprochement and integration, primarily because Levinson, in contrast to Sperber and Wilson (²1995), acknowledges, if largely obliquely, the historicity of language. Levinson (2000) has developed a theory of language norms which in various respects seems compatible with Coseriu’s distinction between language system, normal language use and discourse, even if the Neo-Gricean reliance on truth-conditions and their role in constituting shared contents in communication remains a key difference between the two approaches (Belligh and Willems 2021).
Levinson (2000) borrows from Grice (1975, 1981, 1989) the insight that speakers usually mean more than they say. To capture this discrepancy, an explanatory approach must be developed that makes it possible to determine how cooperative interlocutors get from “what is said” to “what is meant”, according to Levinson. Levinson finds the key to this challenge in Grice’s notion of the cooperative principle and the associated conversational maxims, developing a comprehensive three-leveled theory of meaning which mainly focuses on sentences (but is also applicable to the lexicon). The three semantic levels Levinson (2000: 22) distinguishes are: i/ sentence-meaning, i.e. the meaning that is grammatically encoded, ii/ utterance-type-meaning, i.e. the meaning that accrues to an utterance by way of “default inference” based on normal language use, and iii/ utterance-token-meaning, i.e. the unique, one- off content of an utterance as part of a particular discourse in a specific physical, psychological and historical setting. For the present purposes, the layer of utterance-type-meaning is of particular importance. It is, like Coseriu’s “norm”, the intermediary semantic layer between the grammar and the concrete utterance: grounded in what is encoded in the grammar, utterance- type-meaning is characterised by a superimposed semantic enrichment that goes beyond “what is said” in terms of encoded meaning.
2.7 According to Levinson (2000), two of the four Gricean maxims are essential with regard to the semantic enrichment that distinguishes utterance-type-meaning from sentence meaning and utterance-token-meaning, viz. the maxim of Quantity and the maxim of Manner. Based on these two maxims, Levinson (2000: 27-42) submits three “heuristics” which interlocutors adhere to when engaging in linguistic activity. I discuss the three heuristics in turn.
The first heuristic, “What isn’t said, isn’t”, is related to Grice’s submaxim of Quantity “Make your contribution as informative as is required (for the purpose of the exchange)”. This submaxim is commonly held responsible for scalar implicatures, compare:
(6) Some of the boys came to the party.
(6) scalarly implicates ‘not all’ but is compatible with the conclusion that all boys came (compare: Some of the boys came to the party, in fact all of them came). A speaker who says (6) may not be in a position to choose the stronger alternate (Levinson 2000: 36). Other stock examples of the first heuristic are provided in (7-9). If someone says (7), then the interlocutor can assume that there were exactly three men and not four, five or more, even if more than three is not excluded by the utterance itself. Similarly, if someone says (8), then it is reasonable to
assume that the dress was not red and blue, not red and white, not red and black, and so forth. And (9) normally conveys that at least some students will pass the exam.
(7) Three men came to visit us.
(8) My sister’s dress was red.
(9) Not all of the students will pass the exam.
Revealing conclusions can thus be drawn from the quantity of what is said, which is a means to enrich an utterance with content that is not expressly communicated but nevertheless may shape normal interpretation in decisive ways. The second heuristic, “What is expressed simply is stereotypically exemplified”, is related to Grice’s submaxim of Quantity “Do not make your contribution more informative than is required”. There is no need to say what can be taken for granted. Consider the example below:
(10) Petra’s book is really good.
There are various interpretations possible on the basis of the utterance’s encoded meaning because the relationship between Petra and the book and between the book and good is neither lexically nor syntactically specified. The most plausible interpretation depends on the respective context or horizon of expectations in a given situation. For example, if Petra is my colleague, then perhaps the book that Petra borrowed from the library for me is meant, but it can also be the book Petra bought and read, or the new book by the well-known author Petra Morsbach, etc. In addition, the simple statement that a book is good or bad is normally taken without further specification, ceteris paribus, to mean that it is either well or badly written, but in an appropriate context a book can be good for an altogether different reason, e.g. because it works well as a doorstop (cf. Frisson 2015: 17, referring to a quote from Alfred Hitchcock). Similarly, Levinson (2000: 37) rightly points out that one will not find an answer in the English language system as to what the precise relationship is between the two constituents in compounds such as the following:
(11) bread knife, kitchen knife, steel knife
Specifications such as ‘knife for cutting bread’, ‘knife for using in the kitchen’ and ‘knife made of steel’ are stereotypical interpretations that are undoubtedly widely shared among interlocutors, but they are no encoded signifieds of these compound expressions. This is exactly analogue to Coseriu’s aforementioned analysis of German compounds. Levinson (2000: 37-38)
also explains various other observations in terms of the second quantitative heuristic, including the preferred interpretations of conjoined subjects (12) and word order (13) and the preference for local coreference (14):
(12) John and Jenny bought a piano.
(13) John turned the key and the engine started.
(14) John came in and he laughed.
John and Jenny did not necessarily buy a piano together, even if this is the preferred reading of (12). (13) does not encode whether there is a sequential relation, temporal or causal, between turning the key and starting the engine, again analogously to the iconicity of sequence mentioned in § 2.4, which is often a matter of inference rather than encoding. Similarly, (14) does not encode that John was the person who laughed. Finally, the third heuristic, “What is said in an abnormal way isn’t normal”, is related to
Grice’s maxim of Manner “Be perspicuous”, specifically the submaxims “Avoid obscurity of expression” and “Avoid prolixity”. Logically speaking, the following pairs of utterances are equivalent:
(15) a. It’s possible the plane will be late.
b. It’s not impossible that the plane will be late.
(16) a. Bill stopped the car.
b. Bill caused the car to stop.
In natural language, such pairs are not equivalent, neither semantically nor pragmatically (Levinson 2000: 38-39). Not only do lexical differences entail that there are differences in encoded meanings, but the choice for a more complex utterance normally signals a contrast with a less complex utterance in terms of diagrammatic iconicity (Haiman 1994). For example, the double negative not impossible (15b) might convey that the possibility is somewhat less than when the plain expression possible (15a) is used, and cause to stop (16b) is appropriate if, e.g., the action of stopping in some way deviates from what is to be expected, for example when the driver does not simply use the brakes but pulls the emergency brake.
2.8 Everything that may be “meant” in these examples but is not expressly “said”, is what
Levinson calls “conversational implicature”. Levinson distinguishes, following Grice (1981: 185), two types of implicatures, viz. “particularised” and “generalised” conversational implicatures (Levinson 2000: 16-18). If someone turns down the invitation expressed in (17) by saying (18), then the answer is based on a “particularised” implicature:
(17) Would you care to join us for dinner tonight?
(18) Tomorrow I have to get up early.
By virtue of the context of the utterance and specific contextual assumptions the reply can be interpreted as an indirect negation, even when there is no negating meaning encoded: the overt violation of a conversational maxim in (18) with regard to (17), viz. the maxim of Relevance, may be said to be a means to convey the rejection indirectly without saying “No”. This is what Grice (1975: 49) called “flouting” or “exploiting” a maxim. “Generalised” implicatures are also a means to convey a message indirectly by flouting one or more maxims, but in contrast to “particularised” implicatures they are so habitualised that a very specific context is needed for them not to obtain and be defeated, even if they too are not part of the utterance’s encoded meaning and thus, like all implicatures, “cancellable” or “defeasible” (Levinson 2000: 15-16). All default interpretations discussed so far regarding the examples (8) through (16) come about on the basis of “generalised” implicatures. Levinson’s main concern is to show that ordinary communication is to a high degree determined by such generalised implicatures. What might seem to be encoded meanings actually are habitual default inferences, i.e. parts of content that are “meant” but not “said”. A theory of normal language use based on generalised implicatures is therefore an indispensable part of a coherent theory of language, according to Levinson. From a Coserian perspective, the level of generalised implicatures represents an intermediate level between language-specific encoded signifieds on the one hand and the individual communicative intentions of speakers / hearers each time they engage in discourse, on the other. The level of generalised implicatures
is a level of systematic pragmatic inference based not on direct computations about speaker-intentions but rather on general expectations about how language is normally used. These expectations give rise to presumptions, default inferences, about both content and force […]. (Levinson 2000: 22, italics mine)
2.9 I now proceed to the question of how Coseriu’s approach to normal language use and Levinson’s Neo-Gricean approach to default inferences might complement each other. Levinson’s approach has the advantage of identifying specific maxims-based procedures that cooperative interlocutors employ in order to convey meanings implicitly. This has turned out to be a promising line of enquiry in which Grice’s cooperative principle and theory of maxims have been given pride of place. At the same time, Levinson’s approach is first and foremost concerned with universal aspects of communication and communicative competence. Grice’s maxims apply in principle everywhere and in all language communities, regardless of any specific language. A distinctive feature of almost all of Neo- and Post-Gricean pragmatics is
that while pragmatics is clearly distinguished from semantics, the different levels of meaning are conceptualised without reference to specific languages or, for that matter, linguistic diversity. Well-known Griceans – Robyn Carston, Kent Bach, Jay David Atlas, among others — do not normally take into account that systematic encoded meanings are language-specific and hence highly diverse (Belligh and Willems 2021). Levinson (2000: ch. 2) admits that semantics must also take language-specific differences into account, but he does not explain how a coherent analysis of the semantics/pragmatics interface can account for language- specific meanings. For Griceans, semantics is language-specific only to the extent that it is not — potentially universal — pragmatics.
2.10 On the latter point, Coseriu’s approach is entirely different. Coseriu readily acknowledges the role of maxims in conversation, with particular emphasis on the shared tacit assumption that interlocutors speak in a coherent and meaningful way, but he does not explicitly refer to Grice. Instead of the “cooperative principle”, Coseriu invokes what he calls the “principle of trust” (“Prinzip des Vertrauens”):
The knowledge involved in the meaningful interpretation [also of the non-sensical] is a knowledge of which maxims are used when one speaks. One assumes certain principles of speaking […], i.e. one assumes in advance that the person who speaks does it coherently and meaningfully. If the utterance is initially not coherent, then one searches for coherence. This is done because one assumes that speaking has to be coherent, as it were, and because one has confidence in others in this respect. […] Hence in interpreting what is said, the principle of trust applies.6
Coseriu moreover stresses the fact that encyclopaedic and logical categories at the universal level of designation (elocutional knowledge) cannot be derived from the historical level of language-specific idiomatic knowledge (Coseriu 1974: 64-71, 1987a: 5, 2007 : 113-116). However, following Humboldt (Di Cesare 1998: 51-56), Coseriu is equally adamant about linguistic competence being first and foremost a historical cultural – and hence continually “created” or ever-changing – reality (Coseriu 1983b). An integrationist approach to language must therefore give precedence to the activity of speaking a specific language:
Speaking is a universal activity that is general to human beings, realised individually in specific situations by individual speakers who are representatives of language communities with shared traditions of linguistic competence.7
For Coseriu, like for Humboldt, “Sprechenkönnen” necessarily involves at least one language-specific “technique of speaking” as an essential part of linguistic competence (Coseriu 1985). This means that one cannot first leave aside the knowledge speakers / hearers possess of language-specific meanings in order to address the potentially universal role of pragmatics in discourse and then afterwards take those language-specific meanings into account again with a view to complement the analysis of the ways semantics and pragmatics interact with one another. The role of both elocutional knowledge (designation) and expressive knowledge (linked to discourse and texts) must be determined from the vantage point of language-specific idiomatic knowledge, not the other way round. This entails that Coseriu’s notion of normal language use is more differentiated and, as a result, more comprehensive than the (Neo-)Gricean notion of normal usage. This has a number of consequences for empirical analyses, to which I now turn.
2.11 Recall that Levinson (2000) invokes the second heuristic, “What is expressed simply is stereotypically exemplified”, to account for the default interpretations of sentences such as (13) and (14), repeated here as (19) and (20):
(19) John turned the key and the engine started.
(20) John came in and he laughed.
It is important to specify that the default interpretations – sequential order in (19) and local coreference in (20) – are based on generalised implicatures that obtain in English but not on generalised implicatures that are universally valid. In many languages there is no grammatical encoding of sequential order and local coreference, but there are also languages in which such encoding is part of the grammar, and then no default interpretation takes place. First consider the example below from Mandarin Chinese.
(21) Zhāngsān dào túshūguăn ná shū.
Zhangsan reach library take book
‘John went to the library to get the book.’
(22) Zhāngsān ná shū dào túshūguăn.
Zhangsan take book reach library
‘John took the book to the library.’
(23) Zhāngsān dào bìngqiĕ túshūguăn ná shū.
Zhangsan reach and besides library take book
‘John went to the library and he took the book.’
In Mandarin Chinese, asyndetic coordination of clauses encodes the temporal order of the sequence of events (Tai 1985: 51). By contrast, when clauses are separated by the word bìngqiĕ ‘and besides, and moreover’, no temporal or causal order is encoded, but it may still be implicated on the basis of the elocutional principle of iconicity of sequence. (23) conveys two separate actions (Newmeyer 1992: 759, 776-777; see Tai 2002: 332-340 for discussion).Next consider the following sentences from Ewe (Niger-Kongo):
(24) Kofi be e-dzo
Kofi say he-leave.PAST
Kofii said that hei/*j left.
Ewe has a neutral third person pronoun e, which is like the English third person pronoun, but it also has a so-called logophoric third person pronoun yè, which obligatorily signals local coreference with verbs such as ‘say’, ‘believe’ etc. as in (25) (Clements 1975: 142). Again, no default interpretation takes place due a to language-specific rule of grammar, which provides for a specific signified that is at best an inference in other languages (or possibly also in the same language, as in Ewe, compare (24)).
2.12 Another factor to take into account, apart from language-specific rules that override a universal possibility of designation, is neutralisation. Neutralisation is grounded in the pervasive trait of natural language that the unmarked member of an opposition (particularly in phonology, the lexicon and syntax) can stand for the whole category, i.e. the unmarked and
marked member, in the position of neutralisation. The notion is well-known from Prague structuralism (“Aufhebung”, Trubetzkoy 1958 : 69-75; see also Jakobson 1962: 646-650 and 1971 ). It was subsequently adopted into markedness theory in early functional typology (cf. in particular Greenberg 2005 ) and it also plays an important role in Coseriu’s three-layered approach of linguistic competence, albeit in an extended sense that takes into account the three different levels of language introduced above (§ 2.1): the universal, the language-specific and the discourse or text (Coseriu 1985: xxxiv-xxxv, 2007 : ch. 2). Take, for instance, Levinson’s (2000) example of the difference between the sentences (15a)
and (15b), repeated below as (26):
(26) a. It’s possible the plane will be late.
b. It’s not impossible that the plane will be late.
Given that the equation ¬ (¬A) = A is a universal principle of logical thought, the fact that the normal interpretation of (26b) is not identical with that of (26a) entails that the logical principle is overridden in natural language. However, the language system must not be mixed up with normal language use. On the level of the English language system, not impossible is entirely adequate to express precisely the logical meaning ¬ (¬A), which is also its encoded meaning in English. The common reading in terms of an attenuated possibility is therefore necessarily a fact of normal language use. Note that the same kind of inference holds for the double negatives in (27) but not those in (28).
(27) not inconsistent, not incorrect, not infinite, not immoderate, etc.
(28) not incandescent, not infamous, not inordinate, etc.
Because the adjectives incandescent, infamous and inordinate are no antonyms to candescent, famous and ordinate in English, no default interpretation is available for the double negatives in (28) on the basis of the generalised implicature that gives rise to the default interpretations of the double negatives in (27) (Horn 1991: 91-93, 2010: 117).
2.13 On the other hand, in many languages double negation is a systematic grammatical construction, not to express that something is slightly less the case than when the positive expression is used, but to convey a single negation (Horn 2001: 269-308; 2010). Horn (2010: 111-112) terms this type of double negation “hypernegation”. In Indo-European languages it is found, for example, in Spanish (No ha visto nada ‘He hasn’t seen anything’), Italian (Non ha
visto niente / nulla), varieties of African-American English Vernacular (He di’n’t see nothin’), Afrikaans (29), several Flemish dialects, e.g. the Aarschot dialect in province of Flemish Brabant (30), and so forth:
(29) Hy het nie gelag nie. (den Besten 2012: 236)
he has not laughed not
‘He did not laugh.’
(30) A zal ni kom nie. (den Besten 2012: 237)
he will not come not
‘He will not come.’
In other words, in language systems such as these, the logically congruent reading of ¬ (¬A) (i.e., as equivalent to A) is neutralised in favour of an alternative reading that is logically incongruent but sanctioned by the grammar of the language as full negation: the signified ‘negation’ is part of speakers’ idiomatic knowledge which overrides the logical principle of double negation that pertains to elocutional knowledge. To treat these idiomatic double negations as an illogicality is hence misguided (Horn 1991, 2010). Moreover, negative concord (i.e., the co-occurrence of a negative indefinite pronoun and sentential negation in the same clause) is widespread and typologically common (Haspelmath 2005).
2.14 Neutralisation is also essential to account for expressions that violate elocutional knowledge or idiomatic knowledge at the level of a specific text or discourse. This third level of language relates to what Coseriu (1985, 2007 : § 1.5, 2007 : § 2.5) refers to as expressive knowledge, i.e. knowledge of discourse and text traditions, text genres, etc. Expressive knowledge only partly complies with elocutional knowledge and idiomatic knowledge. (31) and (32) below are illustrations of elocutional incongruence being neutralised in specific discourse traditions, viz. the utterance of a person arriving at the requested floor in an elevator (31) and a turn between a waitress and a client in a restaurant (32; example taken from Kennedy 2019: 248).
(31) Fifth floor – that’s me!
(32) a. This is the ham sandwich. (waitress holding up a ham sandwich)
b. I am the ham sandwich. (raising my hand and beckoning to the waitress)
It is arguably inconsistent with our world knowledge that a person would either be a floor of a building or a ham sandwich, but incongruences such as these are neutralised in specific discourse settings like those illustrated in the examples. The discourse or text induced neutralisation can also override idiomatic correctness, as in (33) and (34):
(33) Une romaine patrouille!; Bonté gracieuse!; Je pense qu’il va être l’heure, n’est-il pas?; Sûr, vous pouvez!; J’étais en dehors de mes esprits avec l’inquiétude; Il est devenu absolument noix; Vraiment, c’est!
(34) It was very hot in the train and it was very hot in the lit salon compartment. There was no breeze came through the open window.
(33) are lexical and syntactic imitations of English in the French comic book Astérix Chez Les Bretons (transl. Asterix in Britain) (Coseriu 1985: xxxv). (34) is taken from Ernest Hemingway’s short story A Canary for One (1927). The second sentence is grammatically deviant because no breeze has a double function as two different constituents in two coordinated clauses. This “apo-koinou” construction is rare in English but the creative exploitation of word order it attests to has an expressive, “aesthetic”, purpose. Obviously, what is appropriate or inappropriate in language differs widely depending on what the purpose of the discourse or text is, who the interlocutors are, the context and situation, its traditional set-up, etc.8
2.15 I close this section on pragmatics with an interim conclusion. Coseriu often stressed the need for a science that deals with the role of “knowledge of things” in speech, a so-called “skeuology” (Coseriu 1995: 191; Albrecht 2021: 172-173). Even his attention to shared maxims is geared towards how interlocutors speak in ways that are coherent and meaningful with regard to the “things” that are the subject of world knowledge and logical reasoning (Coseriu 1987a: 1-2). Conversely, Levinson’s (Neo-)Gricean theory of generalised conversational implicatures does not so much draw on factual knowledge (the subject of skeuology) but on what might be called a linguistic “praxeology” and the corresponding knowledge of how people proceed when they engage in linguistic activity. Levinson’s approach is therefore centrally concerned with the maxims of Quantity and Manner and their role in conveying default implicit meanings in communicative practices.
These different views on the kinds of knowledge that govern linguistic interaction at the same time marks out to what extent the two approaches are complementary. The (Neo-)Gricean conversational maxims that speakers respect and intentionally flout to convey implicit meanings are universal strategies: double negation, word order, syndetic and asyndetic connections between clauses, lexical periphrases, the use of anaphors, etc. are conceived as potentially universal procedures, which can of course be regulated in different ways in different languages but which are invoked as practices deployed in linguistic activity irrespective of language-specific and discourse or text-specific features. By contrast, Coseriu’s approach focuses on exactly that: the ways language-specific (idiomatic) knowledge of signifieds mould universal (elocutional) knowledge and principles of thought and in turn provides the input of text and discourse traditions that originate in speakers’ expressive knowledge. As well as being indispensable to establish language-specific constraints on the workings of universal conversational maxims, this focus makes it possible to take into account text and discourse- related competence and to determine how norms of designation (which are part of elocutional knowledge) may be neutralised by language-specific norms (which determine idiomatic knowledge), while both these norms may in turn be neutralised by norms that inform speakers’ expressive knowledge of specific text and discourse traditions. These relations of neutralisation can be represented schematically as follows (Coseriu 2007 : 179; cf. 1985: xxxv):
Figure 2: Relations of neutralisation between the three levels of language
These findings support the observation mentioned in § 2.1 that Coseriu’s “Integral Linguistics” framework is opposed to any form of a-historical pragmatics. As a matter of fact, pragmatics has to be delimited differently from a Coserian perspective depending on the level it is associated with: the universal (elocutional), the language-specific (idiomatic) or the individual (expressive). This requirement underscores the pivotal role, in linguistic analyses, of the multi-layered intentions of individual speakers. While cognisant of principles that are universally valid as standards of judgment, speakers’ linguistic competence is rooted in the historical reality of techniques and traditions of speaking: speakers not only know “a language”
3. Coseriu’s perspective on schemas and constructions in comparison with “Construction Grammar” and “Construction Morphology”
3.1 As mentioned before, Coseriu’s theory of normal language use also plays an important role in his theory of word formation. Coseriu emphasises that the relational meanings associated with particular functional templates or processes (“funktionelle Verfahren”, Coseriu 1977b: 49) are semantically general (“allgemein”, 51). This semantic generality is actually in accordance with the indeterminacy of all encoded signifieds (G. “unbestimmt”, Coseriu 1987a: 6). Only particular norms of language use or specific contextual enrichments based on designation and / or elocutional knowledge provide the principium individuationis that decides how exactly the relationship between the constituents of the word formation is to be interpreted on a specific occasion. Accordingly, designation is in principle a “further and additional determination of word-formation processes” (1977b: 49), while the meaning of the word formation template is the primary given.
Coseriu (1977b: 52) calls the relationships that exist between the constituents of word formations “grammar-like” (“grammatikähnlich”). This is an important point that deserves some clarification. Language-specific word formation templates are not merely formal templates but combine expressions with signifieds. They form their own subsystem of a language, which can be located between the subsystem of the primary lexicon and the subsystem of syntax, both of them also language-specific. The basic input of word formations are elements of the primary vocabulary, including complex elements that have already been transformed by some word formation template and subsequently become part of the lexicon.
The general relations expressed by the relational meanings that tie together the constituents of word formations result from “semantic equivalences” (“Bedeutungsäquivalenzen”) between the word formations (the “products” of processing the word formation templates) and the grammatical or syntactic constructions that correspond to their meaning (“den ihnen inhaltlich entsprechenden Konstruktionen”, 1977b: 52). It is, for example, possible to paraphrase Fr.
beauté as ‘le fait d’être beau / belle’. This paraphrase is an analytic tool in the format of a syntactic structure, the nominalised French word beauté is not in turn the product of the transformation of a syntactic structure into a word formation. The complex word is the outcome of a template that takes its lexical material from the primary vocabulary and uses this material
according to word formation principles, not syntactic principles, to build the complex word.
3.2 The notion “construction” has a long pedigree in the history of the language sciences. For instance, Port-Royal grammarians define syntax in terms of “construction des mots ensemble” (Arnauld and Lancelot 1660: ch. 24; Latin con-structio is actually a calque of Greek sýn-taxis).
The Encyclopédistes subsequently used the term “construction” to refer to word order in contrast to “syntaxe”, which refers to syntactic relations and hierarchies (Swiggers 1984: ch. 4). Coseriu (1977b) uses the term “construction” (G. “Konstruktion”) when he refers to the class of syntactic combinations that serve as paraphrases of word formations to determine the semantic equivalences between word formation and syntax. By contrast, he calls the outcome of word formation “word formation products” (G. “Wortbildungsprodukte”), not constructions. This terminological distinction is useful, in order not to blur the distinction between syntactic templates and word formation templates.
Coseriu’s observations about the relationship between the general meanings of word formation templates, which are language-specific, and the additional determinations that arise from elocutional knowledge and contextual enrichment, is consistent with a revealing passage in his early study “Sistema, norma y habla” (“System, Norm und Rede”, 1975 : § 5.2). There he discusses the question, which goes back to Saussure (1922 , 1968), of the extent to which the sentence belongs to parole or to langue. Coseriu points out that the sentence cannot belong exclusively to parole (which, by the way, Saussure himself never claimed, cf. Saussure 1968: 284-285, 292-295) because just like speech sounds and words, every actually occurring sentence corresponds to an abstract conceptual structure (“ideelle Struktur”) in the language system. Coseriu refers to such a structure as a “schema”. The sound system, the vocabulary and syntax all correspond to “schemas” in the language system, which according to Coseriu must be distinguished from the infinite variety in the realization of sounds, words and sentences in actual parole. The same applies, as we have seen, to word formation.
3.3 Coseriu was not the first scholar to point out that naturally occurring complex linguistic signs – an inflected word, a group of words, a word formation, a clause or sentence – correspond to functional structures which can be distinguished from one another with regard to both formal and semantic features at the abstract level of a particular language system. To mention just one forerunner: Hermann Paul dedicated a whole chapter in his Prinzipien der Sprachgeschichte to so-called “proportion groups” (“stofflich-formale Proportionengruppen”, Paul 1920 : § 5) under the heading “Analogy”. Paul argues that recurrent aggregations in language use lead to the representation of the general validity of specific structures (“Vorstellung einer Allgemeingültigkeit”, 111), and it is these generally valid structures that make analogous formations possible in the first place. Paul moreover considers proportion groups to be crucial
to capture speakers’ creativity in language use with regard to the interplay between reproduction and innovation (in Humboldt’s sense).
Paul refers to groups, not schemas. Coseriu adopted the term “schema” from Louis Hjelmslev (in particular Hjelmslev 1928 and 1953 , 1961 ), whose theory of language had a profound influence on Coseriu in the 1950s (Jensen 2021). It should be borne in mind though that the notion of “schema” also plays a central role in the philosophy of Immanuel Kant (in the so-called schematism chapter of the Kritik der reinen Vernunft, cf. Willems 1997: ch. 3), which is also echoed, through the intermediary of Humboldt (Coseriu 2015, II: 359), in Coseriu’s approach. For Kant, (transcendental) schemas mediate between the categories of pure reason and sense perceptions (Kant KrV, B 176-187), but whereas Kant argues that schemas are produced by imagination (“Einbildungskraft”, B 179), according to Humboldt this is achieved primarily on the basis of language (cf. Cassirer 2003 : § 3, Borsche 1981: 85-90, Trabant 1990: 31-41, 81, 106-108 and Di Cesare 1998: 22-25).
3.4 With regard to the sentence, Coseriu maintains the following complex distinction. What is expressed in a sentence (“das in ihm Ausgedrückte”) is universal and at the same time infinitely variable, but the way it is expressed (“die Art des Ausdrucks”) is language-specific in a double sense: as schema, i.e. the abstract conceptual structure of a sentence, which pertains to a specific langue, and as a realisation norm, insofar as norms too are usually different depending on the language. This point of view is analogous to Coseriu’s distinction between the abstract word with its language-specific signified and the concrete word that designates something as part of an utterance, and to his distinction between a specific word formation template that has a language-specific signified of its own and an actual word formation understood as a product that designates something, with again the level of normal language use comprising the traditional realisations and corresponding designations.
In Coseriu’s account we thus come across the terms “schema” and “construction” as well as the distinction between primary vocabulary and word formation, with an additional distinction between word formation products and “grammar-like” word formation templates. Neither “schema” and “construction” nor “construction” and “word formation process” are used as synonyms. The ensuing distinctions can be presented schematically as in Figure 3. The full single arrow indicates the gradually increasing complexity of the schematic patterning, beginning with the primary meaning bearing elements (words) before reaching the secondary word formations and concluding with sentences. The dotted double arrow indicates the mutual dependence of phonological units and signifieds in accordance with Martinet’s (1957)
observations on “double articulation” and Hockett’s (1958: 574-577) similar account of the “duality of patterning” in natural language.
Figure 3: Schemas of language-specific structuring in Coseriu
3.5 I have pointed out that Coseriu introduces the notion of “schema” with regard to the sentence while at the same time invoking the need for bringing into play the level of normal language use. In doing so, Coseriu not only anticipates an important topic in modern syntax, but he also approaches this topic in a differentiated manner that has the potential to solve a number of questions contemporary research is faced with.
The notions of “schema” and “construction” have attracted a lot of attention among scholars in modern linguistics, in particular since the 1990s and the growing success of Construction Grammar, which is an offshoot of usage-based Cognitive Linguistics. Construction Grammar has since developed in various directions and is now commonly referred to as a “family” of approaches with different but mutually entangled orientations (Hoffmann and Trousdale 2013, Ziem and Lasch 2013). In what follows, I will focus my attention on the well-established and widely adopted model of Construction Grammar as developed by A. Goldberg (1995; 2003; 2006; 2013) and subsequent work in the same vein (Hilpert 2019).
3.6 The main objective of Goldberg’s (1995) approach is to show that the semantic contribution of lexical items (in particular verbs) to sentences has to be distinguished from the semantic contribution of the syntactic patterns into which those lexical items are inserted. Patterns investigated by Goldberg (1995) under the heading of “argument structure constructions” include the passive construction, the ditransitive construction and the transitive construction with a directional adverbial. Not only lexical items but also syntactic patterns such as these are regarded as fully-fledged “form-meaning pairings”: they represent, according to
Construction Grammar, connections of a form with a meaning in the encyclopaedic “frame- semantic” understanding (I return to this conception in § 3.11). Hence, from the point of view of Construction Grammar, all linguistic signs that combine a form and a meaning may be called “constructions”. The traditional distinction between, on the one hand, lexical units (the vocabulary of a language) and, on the other hand, the rules of syntagmatic combination no longer applies. Construction Grammar instead assumes that there is a continuum between the lexicon and the grammar of a language. At every level, a language consists of “constructions”, and the knowledge of all constructions together is equated with the knowledge of the language: “The totality of our knowledge of language is captured by a network of constructions: a ‘construct-i-con’” (Goldberg 2003: 2019). The broad understanding of “construction” is illustrated below (taken from Goldberg 2003: 220; also Goldberg 2013: 436):
Figure 4: The Construction Grammar approach to “constructions”
Not only argument structures and idiomatic expressions, but also word formations processes, morphemes and words are referred to as “constructions”. At the same time, different degrees of schematicity are postulated, both in terms of form and meaning. Accordingly, the structure of constructions can be modelled in terms of a “lexicality-schematicity hierarchy” (Croft 2003, Barðdal et al. 2011). For example, the English idiom Going great guns is fully specified regarding its expression and content (‘to do something well and with a lot of enthusiasm’), while the X-er the Y-er is only semantically fully specified because the two slots for the adjectives in the comparative remain to be filled with expressions. The three-place ditransitive argument structure construction or the passive structure construction are even more abstract – more “schematic” – with respect to their expression.
3.7 How might Coseriu’s approach and the Construction Grammar approach complement each other? A comparison seems worthwhile for a number of reasons. It is an appealing feature
of Construction Grammar that not only words but also complex linguistic signs are said to correspond to form-meaning pairings (albeit in the holistic cognitive sense, which must not be mixed up with Coseriu’s conception of bilateralness outlined in Section 2). The constructionist reference to different degrees of schematicity is, moreover, akin to the notion of schemas as abstract structural templates, even if the two approaches part company on the issue of how language-specific such templates are. Lastly, it remains a challenge how the meanings of form- meaning pairings, schematic and lexically instantiated ones alike, are to be accounted for and to what extent they have to be differentiated among one another.
In Coseriu’s approach, the meanings of the primary vocabulary of a language have to be distinguished from the meanings of word formation templates and the meanings of syntactic structures (see §§ 3.1-3.4). While all three types of meanings are encoded, i.e. constitute language-specific unitary signifieds that are moreover functionally delimited by contrasts (“oppositions”) based on paradigmatic relations, the nature of the meanings involved is different. Lexical meanings are quidditates that are intuitively apprehended (Coseriu 1987a: 8; Willems and Munteanu 2021: 33-44 for a succinct presentation).9 Conversely, the meanings of word formations are “grammar-like” elaborations of primary lexical items, more specifically modifications, developments and compounds (Coseriu 1977b). Finally, syntactic meanings are meanings that are part of so-called combinatorial paradigms of a language (“Kombinationsparadigmen”, Coseriu 1987a: 143). There are two kinds of combinatorial meanings, according to Coseriu (1987a: 150): syntactic meanings and ontic meanings.10
3.8 Let us first consider syntactic meanings like number, voice and tense. A syntactic meaning presupposes a syntactic combination (whence its name), for instance number requires a noun, verb, pronoun etc. in order to be realised, voice and tense require a verb lexeme, and so forth. Regarding their paradigmaticity, syntactic meanings are no different from lexical items. To take a simple example: in a language in which the category of number encompasses singular,
plural and dual, the three corresponding syntactic meanings must be distinguished from one another in a paradigmatic manner that is different from, e.g., English with its binary number system. Other subsystems of syntactic meanings are more complex. The great variety of voice systems in the languages of the world is a case in point. Active and passive contrast paradigmatically with one another in a two-term voice system like English in a way that is different from the contrast between active and antipassive in a two-term ergative voice system such as Basque or the contrast between active, passive and anti-passive in a three-term ergative voice system such as Kaqchikel (a Mayan language). If one acknowledges that diathesis encompasses much more constructions than active, passive and anti-passive, it is clear that the language-specific paradigmatic contrasts between syntactic meanings in the domain of voice are extremely diverse – irrespective of the fact that certain typological tendencies can be observed and, accordingly, crosslinguistic generalisations can be stated, in particular regarding the number of participant roles, degrees of transitivity, etc. (Palmer 1994, Kulikov 2011, Heaton 2017). Moreover, in establishing the relation between encoded syntactic meaning and crosslinguistic generalisations, the level of normal language use plays a crucial role. I will return to this point below (§§ 3.11-3.15).
3.9 The same reasoning applies to ontic meanings, i.e. propositional or sentence-type meanings. A language system in which, e.g., ‘assertion’ and ‘question’ are encoded as language-specific meanings differs from systems that encode other series of ontic meanings, e.g., ‘assertion’, ‘question’, ‘wish’ and ‘request’, or no specific ontic meaning at all. In the latter case, ontic content would be entirely a matter of inference and hence context-dependent. In most languages, at least some ontic meanings seem to be grammatically encoded. For instance, in German, the ontic meaning ‘question’ is encoded in a specific prosodic construction but not by means of a specific word order, compare:
(35) a. Der Arzt arbeitet nicht mehr für das Krankenhaus.
‘The doctor no longer works for the hospital.’
b. Der Arzt arbeitet nicht mehr für das Krankenhaus?
‘The doctor no longer works for the hospital?’
c. Arbeitet der Arzt nicht mehr für das Krankenhaus?
‘Does the doctor no longer work for the hospital?’
(35b) is an interrogative (i.e., an encoded question) solely by virtue of the question prosody, inversion as in (35c) is an option but not necessary to produce a question in German
(even if it is much preferred in ordinary discourse). Inversion is moreover not necessarily associated with a question. It is also used, e.g., in protases:
(36)Arbeitet der Arzt nicht mehr für das Krankenhaus, dann hat er nur noch seine APrivatpraxis.
‘If the doctor does no longer work for the hospital, he only has his private praxis left.’
3.10 The fact that syntactic and ontic meanings are dependent on combinations with other lexical meanings or categorial meanings such as pronouns (Coseriu 1987a: 149), distinguishes them as secondary meanings in the language system, whereas the units of lexical paradigms constitute primary meanings. Recall that language-specific semantic encoding and the corresponding paradigmaticity of signifieds is not restricted to the lexicon. In fact, the distinction between lexicon and grammar, on the one hand, and semantic encoding and paradigmaticity, on the other, is orthogonal: encoding and paradigmaticity are as foundational in grammar as they are in the lexicon. Importantly, knowledge of a language’s basic vocabulary (words) is not, from a Coserian perspective, knowledge of “constructions”. On the contrary, the words of the primary vocabulary form the “substance” for word formation templates and syntactic constructions without being themselves constructions or any other kind of linguistic units with combinatorial meanings. The words of the primary vocabulary are a language’s basic units (cf. Humboldt 1963 : 15, 1963 : 37-38, 1998 : §§ 13-14). They provide speakers, by virtue of their language-specific encoded meanings, with the principal building blocks in terms of an “intuitive capturing of being as a possibility” in a language (“intuitive Erfassung einer Möglichkeit des Seins”, Coseriu 1992: 18).11
3.11 The question arises as to what may be the deeper reason among construction grammarians for considering all levels of linguistic structuring, from morphemes over words and idioms to sentence patterns, invariably as “constructions”. The answer lies in the theory of meaning Construction Grammar subscribes to, viz. Frame Semantics (Fillmore 1982, 1985, 1988, Fillmore, Kay and O’Connor 1988, Fillmore and Atkins 1992). According to this theory, linguistic units only have “meaning” in the sense of the usage-based perspective of cognitive linguistics: linguistic units, on any level of structuring, do not strictly speaking have meaning,
meaning accrues to linguistic units by virtue of designation in language use. At the same time, designation is not conceived from the perspective of language-specific signifieds, but from the point of view of world knowledge, factual knowledge as well as traditional practical knowledge. Accordingly, any distinction between structurally encoded signifieds (Coseriu’s “Bedeutung”) and the realm of designation (“Bezeichnung”) is rejected (or, better still, considered intractable) from the outset:
a word’s meaning can be understood only with reference to a structured background of experience, beliefs, or practices, constituting a kind of conceptual prerequisite for understanding meaning.(Fillmore and Atkins 1992: 76-77; cf. Fillmore 1985: 223).
This theory of meaning has proven popular among many linguists in the last three decades, undoubtedly in large part because it offers the promise to dispense with the often cumbersome rigidities of having to tease apart – to use Jakobson’s famous words – what a language must convey and what it may convey (Jakobson 1959: 236). However, it also has a number of drawbacks, both regarding language description and linguistic explanations. I have dealt with these at some length in an article (Willems 2011), in which I explore the idiosyncratic interpretation of Saussure’s theory of the linguistic sign from the perspective of Cognitive Grammar (Langacker 1987, Taylor 2002, among others). I refer the reader to this article for further discussion. Here, I focus on one aspect which is of particular interest to the issue at hand, viz. the potential synergy resulting from considering Coseriu’s conception of constructions and schemas to address some unresolved issues in Construction Grammar.
3.12 Goldberg (1995) adopts Fillmore’s analysis of sentences in terms of universal thematic roles, which are not differentiated vis-à-vis encoded semantic roles (see § 2.3). This entails that the meanings of argument structure constructions are defined in terms of traditional usages, hence habitualised designations.12 For example, Goldberg (1995: 2-3, 12-13, 37-38, 141-151) defines the construction meaning of the ditransitive Double Object Construction in English as follows: ‘Agent causes recipient to receive patient’ (“caused possession”), e.g.:
(37) Joe handed his mother a letter.
The construction in (37) is contrasted with an alternative construction which includes a prepositional phrase and which expresses “caused motion” (‘Agent causes patient to move to a goal’) instead of “caused possession”, e.g.:
(38) Joe kicked the bottle into the yard.
The alternation between sentences like (37) and (39) is traditionally referred to as “dative alternation”.13 Goldberg (1995: 89- 95 and ch. 7) assumes that (39) and similar sentences with other trivalent verbs involve a metaphor: ‘transfer of ownership as physical transfer’. (39) accordingly instantiates the “transfer-caused-motion” construction, which “allows the caused- motion construction to be used to encode the transfer of possession” (Goldberg 1995: 90). According to Goldberg (1995: 91- 95), (37) and (39) are “semantically” synonymous but not pragmatically. The recipient is considered less focused in (37) as compared to (39). Ditransitive sentences like (40), which contain atypical (since two-place instead of three- place) verbs, are analysed accordingly:
(40) Chris baked Mary a cake.
According to Goldberg, (40) can only mean that Chris baked a cake with the intention of giving it to Mary (cf. Goldberg 2013: 437). If no intention of ‘causing to receive’ is involved, an alternative beneficial construction with a prepositional phrase must be chosen instead of the ditransitive construction, e.g.:
(41) Sue baked a cake for Chris.
This analysis is shared by most construction grammarians (Croft 2003). One exception is Paul Kay, one of the founding fathers of Construction Grammar, who pointed out that sentence (42) does not necessarily mean that I will peel on orange with the intention to give it to you. It can also mean that I want to show you how well a peeling machine works (Kay 1996: 11-12):
(42) I’ll peel you an orange.
From a Coserian perspective – as well as a (Neo-)Gricean perspective along the lines of Levinson (2000) —, the common assertion that the construction meaning of the ditransitive Double Object Construction in English is ‘Agent causes recipient to receive patient’ is a claim about the most obvious and probably most common use (i.e. interpretation) of the construction, which crucially relies on inference (what is meant), not just encoded meaning (what is said). By contrast, in Goldberg’s (1995) framework ‘Agent causes recipient to receive patient’ is the “central sense” of the construction (Goldberg 1995: 38). Recall that the holistic, usage-based account dispenses with the distinction between the language system and normal language use or the corresponding distinction between language-specific encoded meaning and designation (cf. Taylor 1999). Instead, the central sense of the construction, which can also be classified as salient because an archetypal relationship between people is involved, is said to be the prototype. All other uses with less prototypical “senses” — including ‘Agent causes recipient not to receive patient’, ‘Agent intends to cause recipient to receive patient’, ‘Agent enables recipient to receive patient’ etc. — are considered “extensions” from the prototype linked to it by polysemy links, and the construction as a whole is said to be a polysemous category which can be represented schematically as follows:
Figure 5: The polysemy network of the English ditransitive Double Object Construction according to
Goldberg (1995: 38)
I will not further elaborate here on the conclusion that an argument structure construction is a polysemous category on the ground that it can host various verbs whose meanings alter the construction’s central sense (represented by the prototypical ditransitive verb give, Goldberg 1995: 40). This conclusion is a direct consequence of lumping together the construction’s encoded signified and its array of different but related uses. Figure 5 seems convoluted, given that no extended senses can reasonably be said to accrue to the construction if these senses are supplied by the lexical meanings of the verbs that instantiate the construction’s verb slot (Croft 2003 and Coene and Willems 2006 for discussion). The question what the structurally encoded syntactic meaning of the construction is, in the sense of a “schema” that pertains to a combinatorial paradigm, is not even posed.
3.13 I now turn to the consequences of a perspective on syntactic constructions in which construction meaning is equated with the structured set of uses the construction serves to express, irrespective of the encoded properties it may have of its own. I will draw on extensive
empirical research on the ditransitive construction in German (De Vaere 2020, De Vaere et al. (2021a, 2021b) that adopts Coseriu’s layered approach to meaning. It provides convincing evidence for why it is indispensable to distinguish different levels of meaning in order to capture the function and uses of a construction in both a descriptively and explanatory coherent way.
A first observation concerns the fact that the alternation between German sentences such as (43a) and (43b), taken from De Vaere et al. (2021a: 74-75), has been commonly referred to as an alternation between a Double Object Construction and an alternative construction involving a prepositional phrase, just like in English (e.g., Sabel 2002, Du 2009, Røreng 2011).
(43) a. Meine Lebensgefährtin gab mir eine Notspritze.
‘My partner gave me an emergency injection.’
b. Striffler hat sein Werk an das Archiv des Deutschen Architekturmuseums in Frankfurt gegeben.
‘Striffler has given his work to the “Archiv des Deutschen Architekturmuseums” in Frankfurt.’
This is however a mistake with regard to (43a). The grammatical alignment in German is different from that in English, given that the two objects are differentiated by means of morphological case (dative and accusative, respectively) in German but not in English (Malchukov et al. 2010, De Vaere et al 2021a: 77-78).14 The common terminological confusion regarding the ditransitive alternation in German may be indicative of the currently dominant focus on universal properties of constructions to the detriment of language-specific properties. Such a focus facilitates, without much further consideration, the transfer of conclusions about a well-studied language like English to other languages which have similar constructions (cf. the observations on the differences between “functional grammar and “relational grammar” in Coseriu 1987a: 145).
A second important observation is that the contrast between “caused possession” and “caused motion”, which is traditionally invoked to explain the alternation in English, is also commonly adopted with regard to alternating sentences such as (43) in German (Adler 2011, Welke 2019, Proost 2015, among others; Barðdal et al. 2011: 66, 99 who also point out the
danger of analysing the ditransitive alternation in various Germanic languages along the lines of what we know from English). The adoption of the contrast between “caused possession” and “caused motion”, too, bears testimony to the pervasive presence of the universal level of elocutional knowledge in current (valency as well as constructionist) approaches to syntax. However, such a universalist account misrepresents the alternation in German in a number of respects, which I briefly summarise next.
3.14 In a critical confrontation with the aforementioned universalist perspective and based on a large-scale corpus study of the two alternants in German with over a dozen of different transfer verbs, De Vaere (2020), De Vaere et al. (2021a, 2021b) propose an alternative account of the ditransitive alternation in German. They moreover provide evidence that Coseriu’s distinction between a construction with an encoded, paradigmatically defined syntactic meaning and traditional instantiations in normal language use with corresponding types of enrichment through designation is instrumental in accommodating the broad spectrum of semantic variation found in the data.
In particular, it is shown that the classification of both the argument structure with the dative and accusative object (43a) and the argument structure with the accusative object and the prepositional object (43b) as fully-fledged constructions in their own right crucially depends on the supposition that the content of a construction is to be accounted for in terms of designation rather than constituting a language-specific encoded signified. In contemporary German, the ditransitive accusative-dative construction and the ditransitive prepositional object construction have no independent syntactic meanings. For this reason, they cannot be considered as two “constructions” in the grammar of present-day German. Many three-place verbs occur in both argument structures, and based on an analysis of nearly 10,000 sentences, De Vaere (2020) shows that in present-day German the two argument structures, which often alternate with each other, represent variants of a single construction. These variants are called “allostructions”, in analogy to allophones and allomorphs in phonology and morphology (De Vaere et al. 2020 for discussion). Both variants are found, in naturally occurring texts, with partly the same, partly different habitualised senses in normal language use. Variables such as pronominality and animacy of the objects, the length of the constituents, the specific sense of the verb in context, etc. play a role in the alternation and turn out to be in large part verb-specific rather than characteristic of one of the two alternants.
With regard to the superordinate level of the construction, a language-specific, “schematic” meaning can be paraphrased in paradigmatic contrast to other argument structure constructions. The scaffold of the schematic syntactic meaning of the ditransitive construction
is provided by a number of components: i) a general verbal transfer meaning, ii) three argument roles (to be distinguished from the three participant roles of three-place verbs) and iii) the indeterminacy (or underspecification) of the argument roles, which are labelled Agent-like, Theme-like and Goal (Bickel 2011: 403, Willems 2020: 155). The findings regarding the argument roles of the German ditransitive construction are particularly worth noting. They dovetail with the observation in § 2.3 concerning the difference between universal thematic roles and encoded semantic roles. In the German ditransitive construction, the semantic role of the Subject is no “Agentive” (cf. 44) and the semantic role of the Indirect Object is no “Receptive” (cf. 45). Agent and Recipient are denotational categories that are only specified in contexts of use, not at the level of the construction; compare (examples taken from De Vaere 2020):
(44) a. Der Sieg hat mir viel Selbstvertrauen gegeben.
‘The victory gave me a lot of confidence.’
b. Vielfach gaben vielfältige Tätigkeiten den Schülern regelrecht Grund zur Freude.
‘A great diversity of activities often gave the students a real reason to be happy.’
c. Es gibt Gutachten, die klare Empfehlungen an die Stadt geben.
‘There are reports that make clear recommendations to the city.’
(45) a. Sie haben der Klinik Reha-Patienten geschickt.
‘They sent rehab patients to the clinic.’
b. Kunst verleiht der Fantasie der Betrachter Flügel.
‘Art gives wings to the viewers’ imagination.’
Furthermore, the formal marking of the Indirect Object by means of a noun phrase in the dative case or a prepositional phrase (with an, with some verbs also with zu) is not a feature of the construction either, but distinguishes the two alternants in normal language use. The expression of the Goal role in the German ditransitive construction is not specified with regard to the distinction between a case marked noun phrase and a prepositional phrase (which is also case marked): the Goal role in the construction is “schematic” with regard to the distinction between the two (strongly related) syntactic functions of morphological case and preposition.
On the other hand, the semantic difference between the prepositions an and zu, which are both found with certain verbs (compare 46 below, taken from Proost 2015: 171), is accounted for in
terms of Coseriu’s category of “instrumental meaning” (Coseriu 1987a: 149), which specifies functional elements such as prepositions in view of their semantic contribution to a syntactic combination (prepositions are hence not themselves part of a grammatical combinatorial paradigm). Whereas the prepositional phrase an X turns out to be a prepositional variant of a dative-marked noun phrase within the German ditransitive construction and has no spatial instrumental meaning in present-day German, the prepositional phrase zu X remains a carrier of a spatial instrumental meaning.
(46) a. Er faxt seinem Kollegen die Nachricht.
≈ ‘He faxes the message to his colleague.’
b. Er faxt die Nachricht an seinen / zu seinem Kollegen.
‘He faxes the message to his colleague.’
3.15 The above brief outline cannot do full justice to the intricacies of the investigations required to establish the formal and semantic properties of a construction like the ditransitive argument structure construction in a single language, due to the plethora of factors that have to be considered and the theoretical distinctions that require careful attention in the analysis. It may nevertheless suffice to show that sentence schemas can be determined uniformly with regard to their language-specific formal and semantic properties and that it is possible to differentiate between language-specific encoded constructions and pragmatically motivated variants (or “subconstructions”, as they are often called in the literature), even if it requires taking a certain degree of indeterminacy regarding the construction’s encoded form and meaning into account. It also becomes clear that the distinctions between linguistic system and normal language use and between encoded meaning and designation are key to determining the language-specific differences between similar syntactic structures in different languages.
Coseriu’s notion of schema furthermore provides an interesting avenue of research to pursue in terms of an alternative perspective on corpus linguistics than the one that is currently predominant. As well as dispensing with the distinction between the linguistic system and normal language use, the “usage-based” perspective of most current constructionist approaches to grammar does not distinguish between a corpus of naturally occurring discourse and the activity of speaking. A corpus is always érgon, whereas language understood as enérgeia is the dynamic creation of language, which entails that language-specific idiomatic knowledge, dýnamis, is not only applied (“put to use”) but itself created (see Coseriu 1974 , 1975 ; cf. Trabant 2021 for a succinct presentation of this aspect of Coseriu’s theory of
language). Thus, while the focus of the usage-based perspective is limited to what has already been realised in language use, to the extent that it can be found in a corpus, it remains dependent on designation but does not capture the virtual content of language-specific encoding. By contrast, a layered perspective such as the one advocated by Coseriu (1985, 2007 ) is able to account for linguistic competence in its full extent.
3.16 Construction Morphology is a theory of word formation that developed out of Construction Grammar. It has been gaining ground in the past two decades as a constructionist perspective that complements the earlier focus on argument structure constructions, cf. in particular Croft (2001), Fried and Östman (2004), Goldberg (2006), Booij (2010). The most elaborate account of the theory to date is found in Booij’s Construction Morphology (2010), on which I will draw in the subsequent discussion.
Incidentally, Booij (2010: 171) points out that Coseriu’s distinction between “system” and “norm” is important for a comprehensive description of the vocabulary of a language, including word formations. Booij does not elaborate on Coseriu’s distinction, nor does he systematically consider the difference between encoded meaning and designation in his account of compounds and derivations. Booij’s analyses of word formation templates and word formation products are nevertheless more differentiated than what can be found in previous accounts, which hardly paid any attention to the difference between systematic meanings and frequently occurring senses in normal language use, which in turn can also be language-specific (Coseriu 1975 ). For this reason, Booij’s account of word formation provides for an interesting test case on how insights from Coseriu’s theory of word formation (see § 2.2 and § 3.1) might complement recent usage-based approaches with partly similar objectives. In what follows I limit myself to one case study.
3.17 Booij (2010: 77-80) points out that in Dutch, deverbal nouns with the suffix -er have a range of interpretations, as has long been acknowledged in word formation research. The same is true for the corresponding word formation pattern with -er in English and German. Booij proposes a number of “roles” to account for the traditional interpretations of complex words such as the following:
(47) a. bakk-er ‘baker’ (animate Agent)
b. wijz-er ‘pointer’ (non-animate Agent)
c. maai-er ‘mower’ (Instrument)
d. voor-lad-er ‘front-loader’ (Object)
e. treff-er ‘hit’ (Event)
f. gill-er ‘what makes you scream’ (Causer)
Many of these examples are not limited to a single interpretation but can be interpreted in the sense of more than one thematic role. For example, D. tell-er may denote someone who counts (agent) or an instrument. To account for this multifunctionality, referred to as “the problem of polysemy in word formation” (Booij 2010: 77), Booij assumes both a general word formation pattern (“general schema”) and several subordinate patterns (“subschemas”):
Figure 6: The schematic hierarchy of Dutch deverbal -er nouns according to Booij (2010: 80)
The upper level of the schematic hierarchy captures the relational meaning between -er as head and the meaning of any verb that can fill the V slot, for the rest the pattern is semantically “underspecified” (2010: 84). The meaning at the abstract level of the schema [V i- er]Nj is in no contradiction to the finding that the schema occurs with a series of habitualised interpretations. It is therefore warranted to maintain that the “polysemy” of the various subschemas goes hand in hand with the “monosemy” of the general schema (Booij 2010: 77). Booij’s constructionist focus on different levels of schematicity is a major step forward from many previous word formation accounts. It has long been customary to either simply list the various traditional interpretations of word formations or to restrict the alleged “meaning” of a word formation to one — particularly salient — sense, without paying attention to any general level of meaning that can be attributed to the word formation pattern. Instead, the notion of a prototype-based radial category was often invoked to represent the patterns’ structured polysemy organised around a central sense (similar to the approach illustrated in § 3.12 and
Figure 5 with regard to a specific argument structure construction).15
3.18 Booij (2010: 78-81) goes on to explain that the general, semantically underspecified schema [Vi-er]Nj cannot account for the relationship between the various roles and their hierarchy as represented in Figure 6. According to Booij (2010: 80), “the driving forces behind this polysemy pattern” are the “semantic-conceptual extension mechanisms” that take their starting point in “the (prototypical) Agent interpretation”. On the one hand, loanwords like computer and printer are to be analysed in a metaphorical sense as “nomina agentis”. On the other hand, words like D. kurken-trekk-er ‘cork-pull-er, cork screw’ and veg-er ‘sweep-er, brush’ denote instruments but because there is no path from the human agentive prototype to an instrument, kurken-trekk-er and veg-er are metonymies, they designate “participants in the event denoted by the base verb” (2010: 79).
From the point of view of Coseriu’s theory of word formation, this analysis raises a number of questions. First, it seems analytically arbitrary to claim that words like computer and printer should be analysed as “Agent nouns with a personified Agent because a computer computes something, and a printer prints something” (2010: 78), given that ‘animate Agent’ is not an encoded semantic feature of the word formation pattern (see Figure 6). It is furthermore not plausible that one imagines a person doing something when designating a computer or a printer. Nor is it plausible to maintain that words like kurken-trekk-er and veg-er are metaphors or metonymies on the grounds that, e.g., a brush does not wipe the floor itself, but that someone wipes the floor with it (2010: 79). The crucial point is that words that instantiate the schema [Vi-er]Nj do not signify any designations, but share the encoded meaning that makes such word formations and their traditional designations possible in the first place.
Second, Booij’s analysis shows the limitations of a word formation analysis that is not consistently semantic, but jumps back and forth between expression-related and meaning- related criteria. This is the reason why in Coseriu’s theory of word formation the schema [Vi- er]Nj is not considered a derivational template at all, but a template of a compound, more precisely a generic compound. Two types of compounding must not be confused. Lexical composition involves at least two lexemes (e.g., greenhouse, hatchet job, train crew, street vendor, waterproof, babysit). Generic composition is only formally “derivation” but functionally composition. For instance, in the schema [Vi-er]Nj-er fulfils the function of a so- called “prolexeme” which is a general substantive-pronominal element (“ein allgemeines substantivisch-pronominales Element”, Coseriu 1977b: 54) whose meaning is not ‘someone’
but indeterminately ‘someone or something’ (Dietrich 2021: 288). Generic composition is not dependent on metaphorical or metonymic extensions (which, in Coseriu’s text linguistics are defined differently altogether, but this is not an issue in this context, cf. Coseriu 1979 , 1987a: 109, 2000  and see Faur 2021 for discussion).
Finally, the notion of a schematic structure, to the extent it specifies a word formation template, loses much of its original meaning if both the general encoded meaning and the various types of designation are indiscriminately referred to as schemas. This watering down tendency is indicative of the cognitive-linguistic assumption underpinning Construction Morphology that the superordinate schema represents nothing else than an abstraction from one or more subordinate schemas. There is hence no qualitative difference between the different schemas but only a difference in the degree of schematization. This contrasts with Coseriu’s approach, in which schemas of language-specific structuring are specified according to what is structurally encoded meaning (“Bedeutung”) in the language system and therefore distinct from designation (“Bezeichnung”) in individual discourse and normal language use.
3.19 Booij (2010: 82-84) goes one step further and assumes, above the level of the schema [Vi-er]Nj, yet another schema [Xi-er]Nj with an even more abstract schematic meaning. The only categorial specification is that X can be instantiated not only as a verb but also as a noun, the combination of a quantifier and a noun, or a numeral, compare:
(48) a. apothek-er ‘pharmacist’, Amsterdamm-er ‘inhabitant of Amsterdam’
b. drie-wiel-er ‘three-wheel-er, tricycle’
c. dertig-er ‘person in his thirties’
At this point it is difficult to ignore that an analysis which starts out by postulating a human Agent as prototype and assumes different superordinate schemas via levels of increasing schematisation, runs into trouble. Multifunctionality is the hallmark of a word formation template, and this is particularly clear with generic compounds. In German, too, generic compounds are used with a plethora of referential functions: Berlin-er can designate a person, a living being in general, something to eat, and so forth, Fünfzig-er is not only used to refer to a person in his fifties but also a banknote, etc. It may be feasible to capture such a wide usage range in terms of a human prototype and subsequent stages of schematisation, but it is doubtful that the outcome of such a description reflects the knowledge competent speakers have of the encoded meaning and the conventional uses of the word formation template.
3.20 Let me draw an interim conclusion regarding constructionist approaches to language. Construction Grammar(s) and Construction Morphology have been instrumental in promoting and developing a perspective on syntax and word formation that overcomes some of the limitations of projectionist approaches which have been the mainstay in these subfields in the second half of the 20 th century. The claim that both sentences and word formations instantiate schematic patterns is not new. Coseriu is among the scholars who already in the 1950s emphasised the importance of schemas in linguistic competence, from the interplay between sound schemas and word schemas over the role of grammar-like word formation schemas to sentence schemas or “constructions” (cf. Figure 3, § 3.4). Importantly, the notion of schemahas more than mainly a heuristic meaning in Coseriu’s framework. In particular, word formation schemas and sentence schemas are bilaterally associated with meanings that are language-specific. Their traditional, habitualised instantiations in normal language use, which are equally part of the speakers’ linguistic competence, are not to be confused with what they encode semantically (G. “bedeuten”) (see Section 2 above). The holistic view of meaning in the usage-based, cognitive framework of Construction Grammar seems ill-equipped to account for this difference. In this framework, form is directly associated with interpretation:
In order to interpret language, speakers need to assign semantic interpretations to the overtly expressed formal patterns that they witness; conversely, in order to produce language, speakers need to choose formal patterns to express meanings they want to convey. In this way, correlations between surface form and interpretation constitute the basis of our knowledge of language (Goldberg 2013: 435).
Under this view, there is no need to assume a systematic linguistic level of language- specifically established associations of expressions and encoded meanings (see Taylor 2002: ch. 6, 2003: ch. 5). From a Coserian perspective, this is a misapprehension – and perhaps proof of the indirect pervasiveness of Ogden and Richard’s (1923: 11) famous yet wildly misleading “semiotic triangle”, which equates thought with reference but is oblivious of encoded meaning. When engaging in linguistic activity, people do not go about assigning interpretations to formal patterns, nor do they label interpretations by means of formal patterns. They interpret the encoded pairings of schematic expression and content they know with respect to designation and they produce such encoded pairings in order to designate something in the world of experience (Willems 2011 for discussion). Accordingly, in Coseriu’s framework not only words are paradigmatically anchored in a language’s langue (the subject matter of “lexematics”, cf. Coseriu 2001: 215-410), but syntactic meanings and sentence-type (“ontic”) meanings, too, form paradigms of language-specific semantic encoding. Constructions at sentence level
constitute combinatorial paradigms which provide for the patterns (G. “Muster” 2007 : 237) speakers realise in discourse. By contrast, the primary lexicon consists not of constructions but of units (“gemachte Einheiten”, including already lexicalised complex units, Coseriu 2007 : 253-255) that are available (“vorgegeben”) to speakers as the principal components for combinations at the higher levels of word formation and syntax. Importantly, schematic word formation patterns and schematic syntactic patterns are themselves no designations but invariably techniques (“Verfahren”, 253) of encoded meanings: they are the language-specific linguistic resources which enable speakers to designate things, events, processes and activities in the extralinguistic world of experience.
4. Coseriu’s notion of “Bedeutung” (signified, encoded meaning) and semantic “underspecification” in psycholinguistic research
4.1 In the last section of this paper, I focus on some recent developments in the study of meaning from a psycholinguistic perspective. As with pragmatics, Coseriu was sceptic about psycholinguistics, for largely the same reason. Like pragmatics (as understood in the modern, twentieth-century sense), psycholinguistics is not part of linguistics as a cultural science: their objects of enquiry are not amenable to an analysis, or a methodology, that treats language as a historical object, according to Coseriu.
It is true that psycholinguistic research often disregards the cultural reality of language entirely. With its timeless focus on language processing (in particular reception and comprehension, much less production), almost exclusive reliance on quantitative methods (including priming) and common disregard of both naturally occurring discourse and the intuitive knowledge speakers share with regard to language as a cultural artefact (compare, e.g., the discussion in Branigan and Pickering 2017), psycholinguistics often targets issues that to many linguists are, disturbingly, trivial. The problem seems compounded by the habit to confuse methods with “methodology” (along with the theoretical and epistemological assumptions methodological choices presuppose) and to use common linguistic terms that are accorded altogether different meanings. The disconnection from mainstream linguistics may be one of the reasons why a minority of linguists have paid attention to psycholinguistics in the past. However, linguistics cannot afford to ignore that part of psycholinguistic research that is explicitly concerned with “real” language (and not with an object of enquiry that only remotely resembles it) and with questions that relate to speakers’ “knowledge” of language, in particular the mental lexicon and mental grammar. To conclude this paper, I therefore turn my attention
to an interesting research topic in current psycholinguistics that aligns with the areas of interest already discussed in the previous sections.
4.2 In the two previous case studies on pragmatics and Construction Grammar, we saw that an important linguistic insight consists not only in recognising that language-specific encoded meanings cannot be reasoned away or put aside in linguistic analyses, but also that their conceptual unity implies a form of indeterminacy, or underspecification, that distinguishes them as discrete categories of an intrinsically functional nature vis-à-vis designation. The importance of semantic underspecification has long been recognised in Neo- and Post-Gricean pragmatics (albeit accounted for in various ways, cf. Atlas 1989, 2005, Bach 1994, 2010, Carston 2002, 2012, Levinson 2000, among others). We also saw that Construction Grammar and Construction Morphology offer avenues for the development of accounts of meaning that pay heed to its many layers of schematicity, interpretation and representation. Conversely many scholars in other fields of the language sciences are still struggling with how to incorporate underspecification into a comprehensive theory of meaning. It is noteworthy that semantic underspecification has also been a subject of contemporary psycholinguistic research.Several psycholinguistic studies since the 1990s have focused on the processing of homonymy and polysemy on the basis of experiments, including lexical decision tasks (priming), sensicality judgment experiments and eye tracking. Some of the studies found that polysemous and homonymous words are processed in the same way: it apparently does not matter whether words have completely different encoded meanings but an identical expression or whether one expression corresponds to several but related contents. For instance, Klein and Murphy (2001, 2002) found that there is no difference in language processing when subjects interpret sentences with homonymous words like (49) or with polysemous words like (50):
(49) a. bank ‘building or institution for financial matters’, ‘a slope’, ‘a series of objects arranged in a line’
b. pupil ‘schoolchild’ or ‘minor under supervision of a guardian’, ‘black hole in the centre of the eye’
(50) book, newspaper, report, etc.
Book, newspaper, report can refer both to a physical object (a bound book, a soaked newspaper) and its content (a scary book, a liberal newspaper). Klein and Murphy conclude that speakers select the different meanings or senses of polysemous words exactly like they select the meanings or senses of homonymous words. They also found that correct
interpretations in either case depend on whether a sense or meaning is either the most common one overall or the most plausible one in a specific context. This would mean, then, that there is no experimental psycholinguistic evidence for the well-established linguistic distinction between homonymy and polysemy.
4.3 Other psycholinguists subsequently critically assessed the experiments carried out by Klein and Murphy (2001, 2002) along with their conclusions. They replicated a number of the experiments but also conducted new ones with items that are better controlled for regarding relevant factors. They also formulated research questions that are better attuned to the data and used more sophisticated methods (see Frisson 2009, 2015, Frisson and Pickering 2001a, 2001b and 2007, Frisson and Frazier 2005, among others). In what follows, I will focus my attention on the findings reported by Frisson (2009), in which the author provides a summary of the research to date and a general discussion.
Frisson (2009: 114) found that frequency indeed plays a crucial role in the language processing of the different meanings of homonymous words. For example, if the context does not disambiguate which of two homonymous words is meant in a particular sentence, then speakers usually activate the meaning that occurs most frequently in a corpus of natural language utterances (e.g., bank ‘institution or building’). By contrast, the frequency of the different usages of polysemous words does not play a role, in contrast to homonymous words.
On the contrary, Frisson (2009: 114-115) found that speakers who interpret sentences with words such as book and newspaper initially do not commit themselves to any specific use in the process of language comprehension. They instead first activate a general, underspecified meaning. The exact interpretation is moreover only construed or accessed if necessary. For instance, a highly polysemous word like school is not given an exact interpretation until required in the context of a particular sentence (examples taken from Frisson 2009: 112):
(51) a. Jocelyn walked to the school. (‘building’)
b. The concerned mother talked to the school. (‘school board, admissions office, etc.’)
c. The school won the match in the last minute. (‘school team’)
d. School’s out! (‘school time period’) etc.
The same applies to metaphorical and metonymic interpretations of words, e.g. of proper names (meet Dickens / read Dickens). In English, it is common to use Vietnam not only to refer to the country but also to refer metonymically to the Vietnam War. Again, speakers only
commit themselves to the precise interpretation (the “homing-in stage”, Frisson 2009: 117) when it is necessary to determine what exactly the utterance refers to (for further experimental evidence, see Frisson 2015 for ample illustration and additional experimental evidence).
4.4 Why should linguists care about findings such as these? To be sure, they neither verify nor falsify the traditional linguistic distinction between homonymy and polysemy. This distinction refers to an objective fact about language, irrespective of whether or not speakers’ knowledge of it can additionally be measured by means of reaction times, eye-movements, and so forth (Itkonen 2003: § 18 for a brief discussion of the object of psycholinguistics). However, while taking psycholinguistic findings into account does not necessarily make traditional semantics more coherent, a dialogue between both disciplines may arguably benefit both sides: there is the possibility to make concepts more articulate and to arrive at a firmer evaluation of the import of conceptual distinctions that are central to semantics overall. I would like to draw attention to three aspects that seem particularly worthwhile in this regard.
4.5 First, psycholinguistic studies of semantic underspecification such as the ones mentioned show that interpreting concrete language use in discourse and texts does not merely amount to interpreting encoded meanings. The indiscriminate use of the term “meaning” is bound to obscure this important insight, because “meaning” is a potentially misleading cover term for a heterogeneous collection of content types. As a matter of fact, the psycholinguistic findings reported by Frisson (2009) dovetail with the main outcome of much research in linguistic pragmatics (cf. Section 2). The observation that speakers usually mean more than they say is but one of many that underscores the need to determine linguistic meaning in terms of a layered approach which pays heed to the difference between meaning and interpretation.
A layered approach can readily accommodate the complementarity of different content types in language use, viz. universal, language-specific and discourse- or text-specific contents.16 The psycholinguistic findings about the processing of homonyms and polysemes reported by Frisson (2009, 2015) make it possible to determine in real time, by applying suitable methods and measures, what kind of content language users initially access when interpreting an utterance. Frisson (2009: 113) makes the important terminological distinction between meanings of homonymous words and senses of polysemous words, emphasising the qualitative difference between the types of content involved. People actually choose one of the meanings
of homonymous words at the access stage in language comprehension, whereas there is no such selection of senses with polysemes like school, newspaper and book. The view that the specific senses of polysemous words are constructed in a given context on the basis of a general, underspecified meaning, is consonant with an analysis along the lines of Coseriu’s layered approach to meaning. Coseriu’s distinction between signified (G.“Bedeutung”) and designation (“Bezeichnung”), too, involves the difference between drawing on available language-specific lexical and grammatical items and constructing, on the basis of these items, a content that goes beyond the input: language users enrich meaning in discourse and texts with regard to designation, in particular through the elocutional knowledge that we activate along with the idiomatic knowledge of the language. Importantly, Coseriu’s intermediate level of normal language use ensures that the role of traditional — or, in psycholinguistic parlance: stored — senses of words, word formations and constructions is readily accounted for as well: not all enrichments of encoded meaning occur on the spot but are routinely retrieved from memory or, better still, from the “historical” linguistic experiences language users keep updating and sharing with one another all the time.
4.6 Second, Coseriu’s theory of meaning can provide an avenue for resolving some of the outstanding issues which emerge from the psycholinguistic analyses of semantic underspecification. It is no coincidence that Frisson, after establishing the major difference between the underlying underspecified meaning and the multiplicity of senses in the interpretation of polysemous words in context, poses the question:
while the idea of underspecification was put forward to account for the experimental data, not much attention has gone to explaining what an underspecified meaning actually is. For example, which semantic information is contained in such an abstract meaning and which is not stored at this level is unclear. A number of proposals have been offered, though they have almost exclusively been theoretical in nature, without much experimental support. (Frisson 2009: 121)
Frisson refers to a number of widely differing proposals, all centred around the psychological concept of “semantic representation” but all found wanting in some respect. Frisson also mentions, in this context, the Generative Lexicon Theory, which however posits enriched semantic representations directly in the “lexicon” to account for polysemy, thus abrogating the difference between structurally encoded meanings and senses in language use (cf. Willems 2013). Frisson does not consider a proposal along the lines of Coseriu’s theory of meaning, which entails that he does not indicate that underspecified meanings are language- specific and encoded, as opposed to senses, nor that underspecification corresponds to a
conceptual unity in the Coserian sense. The latter point in particular lends support to semantic underspecification not as a purely theoretical notion that can only be approached per viam negationis, but as a positive notion (Coseriu’s “unitary meaning”, cf. Willems and Munteanu 2021: 31-34 and 42-43) which at the same time encompasses its delimitation by the functional contrasts in the primary vocabulary of a given language (Coseriu 1978 and 1983a). It is precisely this conception of indeterminate encoded meanings that seems promising in the present context, because it implies that the difference between encoded meaning and designation is a difference in kind rather than degree, contrary to what psycholinguists (but also pragmaticists and construction grammarians) generally assume. As Coseriu explains regarding the primary vocabulary of natural languages (which he opposes to the “secondary” vocabulary of terminologies and nomenclatures):
For the words of historical languages, at least those of the primary and purely linguistic lexicon, name – in an immediate manner – not ‘things’ but intuitions of quidditates that are intuitively apprehended. Each primary expression of ‘natural’ languages originally corresponds to a νόησις τῶν ἀδιαιρέτων (apprehensio simplex or indivisibilium intelligentia) and not to a clearly delimited class of objects or facts. In these languages one does not name already classified objects; on the contrary, in these languages objects (and ‘facts’) are classified with reference to meanings. 17
This explanation may sound philosophical, and it is true that much transpires in it from Coseriu’s interpretation of Aristotle, Humboldt and Hegel (Coseriu 2015, I: § 6, II: §§ 11-12), but it is an important part in helping us to account for the kind of underspecified meaning processed in the initial access stage referred to above. Encoded meaning is not some kind of representation of an object or a fact or of some prototypical entity (Coseriu 2000 ), but rather the language-specific source and means for such a representation, which in turn requires a non-linguistic principium individuationis that only a specific text or discourse can provide.18 Compared to semantic underspecification, any subsequent specification of content, by speakers as well as hearers, is not a means but an end. This conclusion, too, aligns with Frisson’s (2009:
112) assertion that his findings contradict the common hypothesis of full incrementality, i.e., the assumption that the language processor always attains a “specific interpretation” before continuing in a text.
4.7 Finally, Coseriu’s theory of language norms may also be instrumental in further explaining the psycholinguistic findings on the difference between homonymous and polysemous words. The alleged polysemy of words such as book, newspaper and school does not involve different meanings, as Frisson (2009, 2015) rightly concludes, but the senses of these polysemes are not the outcome of “homing-in” processes in individual texts or discourses across the board either. Senses represent, in large part, traditional contents, habitualised designations in normal language use, which means that they are stored in memory and readily available to speakers and hearers because of their ease of identification. This is an important qualification of the observation that language users commit themselves to a precise interpretation in a “homing-in stage” only when necessary, because the specification with regard to context and situation (including the specific “universe of discourse” or text) does not only involve a process of fleshing out an underlying unitary but indeterminate signified on the basis of further information available to the language user, but a further process of selection as well: the fleshing out can coincide with selecting a traditional sense. Many of the examples discussed in previous sections are cases in point: a good book, steel knife, Bill caused the car to stop, G. Fünfziger – all these utterances combine expressions (themselves phonetically variable to a considerable extent) with indeterminate encoded meanings, but at the same time they are associated with a fairly circumscribed number of conventional interpretations that are both common and frequent within certain boundaries of discourse and texts. In many instances, constructing a specific interpretation and selecting a conventional one go hand in hand – if a homing-in stage is at all required. The latter remark is particularly noteworthy. It is well- established in the psycholinguistic literature that language users often do not arrive at a fully specified interpretation of an utterance but make do with “shallow processing” or “good-enough representations” (Sanford and Sturt 2002, Ferreira and Patson 2007, among others).
4.8 To conclude this section on underspecification, I wish to point out that so far semantic underspecification has been almost exclusively associated with language comprehension in psycholinguistic investigations. Language production is largely disregarded or only included indirectly, through the bias of language comprehension. This is a major shortcoming, which is due to the research design and approach. Psycholinguistic experiments have hitherto required that linguistic input is administered to informants, so that the production of language by informants, and hence the creative potential and use of language, is at best inferred obliquely.
However, if we consider the findings of the two previous sections (on pragmatics and normal language use and on schematic syntactic and word formation templates), it is clear that semantic underspecification plays a central role in language production as well: the language-specific resources speakers draw upon when formulating and conveying a message are themselves indeterminate vis-à-vis the specifications that characterise the message’s content. Again, these resources encompass both infinitely flexible (“schematic”) signifieds and habitualised (“normal”) contents. “Homing-in” entails an enrichment of language that goes beyond that which is available in the linguistic system and normal language use, for hearers and speakers alike.
5. General conclusion
In this paper, I have pursued two main objectives. On the one hand, I have probed into three concepts that have attracted a lot of attention in various branches of present-day linguistics, viz. default inferences based on generalised conversational implicatures in (Neo-)Gricean pragmatics, the notions of construction and schema in Construction Grammar and Construction Morphology, and semantic underspecification in meaning representations in psycholinguistic research. On the other hand, considering the fact that all three concepts were already addressed in one way or another in the scholarly work of Eugenio Coseriu, I have focused on the similarities and differences between the present-day accounts and Coseriu’s Integral Linguistics approach, posing the question whether Coseriu’s contributions can help us provide solutions to certain theoretical and methodological challenges that have arisen, and occasionally identified, in recent research.
In view of the second objective, I have deliberately refrained from elaborating on the epistemological differences between the three contemporary frameworks and Coseriu’s framework. These differences should not be underestimated, but a focus on potential synergies between partly contrasting perspectives on language is bound to benefit more from adopting a “best practices approach” that tries to move beyond differing theoretical and methodological assumptions than concentrating on obstacles which may impede the development of an integrative approach to issues such as those discussed in this paper. One specifically Coserian stance towards the concepts dealt with, which I have also prioritised in the preceding sections, is the emphasis on the need to introduce conceptual distinctions when we are facing complex objects of enquiry. With regard to default inferences based on generalised conversational implicatures, schematic templates and constructions, and
semantic underspecification this specific stance has proven to be useful in order to arrive at an explanatory level that makes reference to existing research but at the same time tries to go beyond it, offering an alternative approach that takes more factors into account and, in so doing, prepares the ground for cogent generalisations.
Under this view, contemporary investigations in the domains of pragmatics, constructionist approaches and psycholinguistics have served in this paper as three docking sites for binding claims about the nature and function of linguistic signs put forward by Coseriu since the 1950s. These claims have in common that they complement the currently dominant universal perspective in pragmatic, cognitive and psycholinguistic research (“universal” in the sense conceived from a Coserian perspective, cf. Willems 2016) with a perspective on language-specific idiomatic knowledge, along with the expressive knowledge associated with specific texts and discourses (cf. Figure 1). Among other things, this differentiation makes it possible to determine the role of traditional, habitualised realisations of language-specific resources while at the same time acknowledging their creative possibilities.
This view has a number of implications, which I have already identified in the interim conclusions to the three sections of this paper. To conclude, I would like to emphasise in general terms that from a Coserian perspective the structurally encoded meanings of words, word formation templates and syntactic constructions are “schematic” in an original sense, viz. schemas do not, as is often thought, represent “generalised designations” (Coseriu 1987a: 7).
In this regard, Coseriu’s perspective also echoes Humboldt’s emphasis on linguistics creativity (i.e., language as enérgeia) as opposed to the still prevailing opinion that the meanings of word, word formations or constructions eventually coincide with the “sum” of their uses. On the other hand, language is neither only systematic nor only usage but also “norm”, with the important addition that normal language use must be considered on all three levels of language, i.e. the universal, the language-specific and the level of text and discourse. With regard to the role of normal language use as an intermediary level between the linguistic system (“langue”) and text and disourse (“parole”) within a particular language, “norm” has an important role to play in a comprehensive layered theory of meaning, which assigns not just encoded signifieds but also pragmatics, combinatorial paradigms and processing to different but complementary levels of language. There is no contradiction between the assumption of an underspecified uniform signified at the level of the language system, the existence (and general cognisance) of different traditional contents at the level of normal language use (“polysemy”) and the ultimately unique content of every act of discourse. On the contrary, these three levels of meaning can be mutually demarcated and coherently accounted for vis-à-vis one another. Differences in usage frequency
Adler, Julia (2011): Dative alternations in German. The argument realization options of transfer verbs. PhD Hebrew University: Jerusalem.
Albrecht, Jörn (2021): “Eugenio Coseriu and the primacy of history”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu. Past, present and future. Berlin / Boston: De Gruyter, 165–179.
Arnauld, Antoine / Claude Lancelot (1660): Grammaire générale et raisonnée (‘Grammaire générale et raisonnée de Port-Royal’). Paris: Pierre le Petit.
Atlas, David (1989): Philosophy without ambiguity. Oxford: Clarendon Press.
Atlas, David (2005): Logic, meaning, and conversation. Semantical underdeterminancy implicature, and their interface. Oxford: Oxford University Press.
Bach, Kent (1994): “Conversational impliciture”, Mind and Language 9 (2), 124–162.
Bach, Kent (2010): “Impliciture vs. explicature: what’s the difference?”, in: Belén Soria / Esther Romero (eds.), Explicit communication: Robyn Carston’s pragmatics. Basingstoke: Palgrave Macmillan, 126–137.
Barðdal, Jóhanna / Kristian Emil Kristoffersen / Andreas Sveen (2011): “West Scandinavian
ditransitives as a family of constructions: With a special attention to the Norwegian ‘V-REFL-NP’ construction”, Linguistics 49, 53–104.
Belligh, Thomas / Klaas Willems (2021): “What’s in a code? The code-inference distinction in Neo-Gricean Pragmatics, Relevance Theory, and Integral Linguistics”, Language Sciences 83. https://doi.org/10.1016/j.langsci.2020.101310
Bickel, Balthasar (2011): “Grammatical relations typology”, in: Jae Jung Song (ed.), The Oxford handbook of linguistic typology. Oxford: Oxford University Press, 399–444.
Booij, Geert (2010): Construction Morphology. Oxford: Oxford University Press.
Borsche, Tilman (1981): Sprachansichten. Der Begriff der menschlichen Rede in der Sprachphilosophie Wilhelm von Humboldts. Stuttgart: Klett-Cotta.
Branigan, Holly P. / Martin J. Pickering (2017): “Structural priming and the representation of language”, Behavioral and Brain Sciences 40, 1–61. [e313]. https://doi.org/10.1017/ S0140525X17001212.
Bühler, Karl (1934): Sprachtheorie: Die Darstellungsfunktion der Sprache. Jena: Gustav Fischer. (Reprints Stuttgart: UTB 1982, 1999.) English translation: Karl Bühler (1990):
The theory of language: the representational function of language. Amsterdam / Philadelphia: John Benjamins.
Carston, Robyn (2002): Thoughts and utterances. The pragmatics of explicit communication. Malden: Blackwell.
Carston, Robyn (2012): “Word meaning and concept expressed”, The Linguistic Review 29, 607–623.
Cassirer, Ernst (2003 ): “Die Kantischen Elemente in Wilhelm von Humboldts Sprachphilosophie”, in: E. Cassirer, Gesammelte Werke. Hamburger Ausgabe, vol. 16: Aufsätze und kleine Schriften (1922-1926). Hamburg: Meiner, 105–133.
Clements, Nick (1975): “The logophoric pronoun in Ewe: Its role in discourse”, Journal of West African Languages 10, 141–177.
Coene, Ann / Klaas Willems (2006): “Konstruktionelle Bedeutungen: Kritische Anmerkungen zu. Adele Goldbergs konstruktionsgrammatischer Bedeutungstheorie”, Sprachtheorie und germanistische Linguistik 16, 1–35.
Coseriu, Eugenio (1970): “Bedeutung und Bezeichnung im Lichte der strukturellen Semantik”, in: Peter Hartmann / Henri Vernay (eds.), Sprachwissenschaft und Übersetzen. München: Hueber, 104–121.
Coseriu, Eugenio (1974): “Les universaux linguistiques (et les autres)”, in: Luigi Heilmann (ed.), Proceedings of the eleventh International Congress of Linguists, vol. I. Bologna: Il Mulino, 47–73.
Coseriu, Eugenio (1974 ): Synchronie, Diachronie und Geschichte. Das Problem des Sprachwandels. München: Fink.
Coseriu, Eugenio (1975 ): “System, Norm und Rede” (Spanish original 1952), in: E. Coseriu, Sprachtheorie und allgemeine Sprachwissenschaft. 5 Studien. München: Wilhelm Fink, 11–101.
Coseriu, Eugenio (1975 [1955–1956]): “Determinierung und Umfeld. Zwei Probleme einer Linguistik des Sprechens” (Spanish original 1955–1956), in: E. Coseriu, Sprachtheorie und allgemeine Sprachwissenschaft. 5 Studien. München: Wilhelm Fink, 253–290.
Coseriu, Eugenio (1975 ): Sprachtheorie und allgemeine Sprachwissenschaft. 5 Studien. München: Fink.
Coseriu, Eugenio (1977a): “Zu Hegels Semantik”, Kwartalnik Neofilologiczny 24, 2–3 [= Gedenkschrift L. Zabrocki], 183–193.
Coseriu, Eugenio (1977b): “Inhaltliche Wortbildungslehre am Beispiel des Typs coupe- papier”, in: Herbert E. Brekle / Dieter Kastovky (eds.), Perspektiven der Juli 1976. Bonn: Grundmann, 48–61.
Coseriu, Eugenio (1978): “Einführung in die strukturelle Betrachtung des Wortschatzes”, in: Horst Geckeler (ed.), Strukturelle Bedeutungslehre. Darmstadt: Wissenschaftliche Buchgesellschaft, 193–238.
Coseriu, Eugenio (1979 ): “Die Metaphernschöpfung in der Sprache” (Spanish original 1956), in: Eugenio Coseriu, Sprache. Strukturen und Funktionen. Tübingen: Narr, 15– 44.
Coseriu, Eugenio (1979): “‘Historische Sprache’ und ‘Dialekt’”, in: Joachim Göschel, Pavle Ivić / Kurt Kehr (eds.), Dialekt und Dialektologie. Ergebnisse des Internationalen Symposions “Zur Theorie des Dialekts”, Marburg / Lahn, 5–10 Sept. of 1977. Wiesbaden: Steiner, 106–122.
Coseriu, Eugenio (1983a): “Pour et contre l’analyse sémique”, in: Shirô Hattori / Kazuko Inoue (eds.), Proceedings of the XIIIth International Congress of Linguists. Tokio: CIPL, 137– 148.
Coseriu, Eugenio (1983b): “Linguistic change does not exist”, Linguistica nuova ed antica. Rivista di linguistica classica medioevale e moderna 1, 51–63.
Coseriu, Eugenio (1985): “Linguistic competence: What is it really?”, The Modern Language Review 80 (4), xxv–xxxv.
Coseriu, Eugenio (1987a): Formen und Funktionen. Studien zur Grammatik. Tübingen: Niemeyer.
Coseriu, Eugenio (1987b): “L’ordre des mots au Colloque de Gand. Bilan et perspectives”, Travaux de Linguistique 14 / 15, 249–261.
Coseriu, Eugenio (1992): Einführung in die allgemeine Sprachwissenschaft (second edition). Tübingen: Francke.
Coseriu, Eugenio (1992): “Zeichen, Symbol, Wort”, in: Tilmann Borsche / Werner Stegmaier (eds.), Zur Philosophie des Zeichens. Berlin / New York: De Gruyter, 3–27.
Coseriu, Eugenio (1995): “My Saussure”, in: Tullio de Mauro / Shigeaki Sugeta (eds.), Saussure and linguistics today. Roma: Bulzoni, 187–191.
Coseriu, Eugenio (2000 ): “Structural semantics and ‘cognitive’ semantics” [Spanish original 1990], Logos and Language I (1), 19–42.
Coseriu, Eugenio (2001): L’homme et son langage. Louvain: Peeters.
Coseriu, Eugenio (2007 ): Textlinguistik. Eine Einführung (fourth edition). Tübingen: Narr.
Coseriu, Eugenio (2007 ): Sprachkompetenz. Grundzüge der Theorie des Sprechens (second edition). Tübingen: Narr.
Coseriu, Eugenio (2015): Geschichte der Sprachphilosophie, vol. I: Von Heraklit bis Rousseau. Vol. II: Von Herder bis Humboldt. Tübingen: Narr–Francke–Attempto.
Croft, William (2001): Radical Construction Grammar. Oxford: Oxford University Press.
Croft, William (2003): “Lexical rules vs. constructions: a false dichotomy”, in: Hubert
Cuyckens / Thomas Berg / René Dirven / Klaus-Uwe Panther (eds), Motivation in language. Amsterdam / Philadelphia: John Benjamins, 49–68.
Davidson, Donald (2001): Inquiries into truth and interpretation (second edition). Oxford: Clarendon Press.
De Hoop, Helen / Bhuvana Narasimhan (2008): “Ergative case marking in Hindi”, in: Helen De Hoop / Peter De Swart (eds.), Differential subject marking. Dordrecht: Springer, 63–
Den Besten, Hans (2012): Roots of Afrikaans. Selected writings of Hans den Besten. Amsterdam / Philadelphia: John Benjamins.
De Vaere, Hilde (2020): The ditransitive alternation in present-day German. PhD Ghent University: Ghent.
De Vaere, Hilde, Julia Kolkmann / Thomas Belligh (2020): “Allostructions revisited”, Journal of Pragmatics 170, 96–111.
De Vaere, Hilde, Ludovic De Cuypere / Klaas Willems (2021a): “Alternating constructions with ditransitive geben in present-day German”, Corpus Linguistics and Linguistic Theory 17 (1), 73–107.
De Vaere, Hilde, Ludovic De Cuypere / Klaas Willems (2021b): “Constructional variation with two near-synonymous verbs: the case of schicken and senden in present-day German”, Language Sciences 83, doi.org/10.1016/j.langsci.2020.101313.
Di Cesare, Donatella (1998): “Einleitung”, in: Wilhelm von Humboldt, Über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluss auf die geistige Entwicklung des Menschengeschlechts, edited by Donatella Di Cesare. Paderborn: Schöningh, 11–128.
Dietrich, Wolf (2021): “Coseriu’s approach to word formation as an illustration of his theory of meaning”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu. Past, present and future. Berlin / Boston: De Gruyter, 279–293.
Du, Rong (2009): Zur Alternation von Doppelobjekt- und Präpositionalkonstruktion bei Besitzwechselverben im Deutschen und Chinesischen. Eine kontrastive Untersuchung. Frankfurt am Main: Peter Lang.
Faur, Elena (2021): “The crux of metaphor: between Cognitive and Integral Linguistics”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu. Past, present and future. Berlin / Boston: De Gruyter, 311–326.
Ferreira, Fernanda / Nikole D. Patson (2007): “The ‘good enough’ approach to language comprehension”, Language and Linguistics Compass 1, 71–83.
Fillmore, Charles (1982): “Frame semantics”, in: Linguistic Society of Korea (ed.), Linguistics in the morning calm. Seoul: Hanshin Publishing, 111–137.
Fillmore, Charles (1985): “Frames and the semantics of understanding”, Quaderni di semantica 6, 222–254.
Fillmore, Charles (1988): “The mechanisms of ‘Construction Grammar’”, in: Shelley Axmaker, Annie Jaisser / Helen Singmaster (eds.), Proceedings of the fourteenth annual meeting of the Berkeley Linguistics Society. Berkeley, CA: Berkeley Linguistics Society, 35–55.
Fillmore, Charles / Beryl Atkins (1992): “Toward a Frame-based Lexicon: The semantics of RISK and its neighbors”, in: Adrienne Lehrer / Eva Feder Kittay (eds.), Frames, fields, and contrasts: New essays in semantic and lexical organization. Hillsdale: Erlbaum, 75– 102.
Fillmore, Charles / Paul Kay / Mary O’Connor (1988): “Regularity and idiomaticity in grammatical constructions: the case of let alone”, Language 64, 501–538.
Fodor, Jerry (1975): The Language of Thought. Cambridge (MA): Harvard University Press.
Fodor, Jerry (2008): LOT2: The Language of Thought revisited. Oxford: Oxford University Press.
Fried, Mirjam / Jan-Ola Östman (2004): “Construction Grammar: a thumbnail sketch”, in: Mirjam Fried / Jan-Ola Östman (eds.), Construction Grammar in a cross-language perspective. Amsterdam / Philadelphia: John Benjamins, 11–86.
Frisson, Steven (2009): “Semantic underspecification in language processing”, Language and
Linguistics Compass 3, 111–127.
Frisson, Steven (2015): “About bound and scary books: The processing of book polysemies”, Lingua 157, 17–35.
Frisson, Steven / Lyn Frazier (2005): “Carving up word meaning: portioning and grinding”, Journal of Memory and Language 53, 277-291.
Frisson, Steven / Martin Pickering (2001a): “Obtaining a figurative interpretation of a word: support for underspecification”, Metaphor and Symbol 16, 71–99.
Frisson, Steven / Martin Pickering (2001b): “Figurative language processing in the underspecification model”, Metaphor and Symbol 16, 149–171.
Frisson, Steven / Martin Pickering (2007): “The processing of familiar and novel senses of a word: why reading Dickens is easy but reading Needham can be hard”, Language and Cognitive Processes 22, 595–613.
Goldberg, Adele (1995): Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press.
Goldberg, Adele (2003): “Constructions: a new theoretical approach to language”, Trends in Cognitive Sciences 7, 219–224.
Goldberg, Adele (2006): Constructions at work: The nature of generalization in language. Oxford: Oxford University Press.
Goldberg, Adele (2013): “Argument structure constructions versus lexical rules or derivational verb templates”, Mind and Language 28(4), 435–465.
Green, Georgia (1974): Semantics and syntactic regularity. Bloomington: Indiana University Press.
Greenberg, Joseph (2005 ): Language universals: with special reference to feature hierarchies. Berlin / New York: Mouton de Gruyter.
Grice, H. Paul (1975): “Logic and conversation”, in: Peter Cole / Jerry L. Morgan (eds.), Syntax and Semantics, vol. 3: Speech acts. New York: Academic Press, 41–58.
Grice, H. Paul (1981): “Presupposition and conversational implicature”, in: Peter Cole (ed.), Radical Pragmatics. New York: Academic Press, 183–198.
Grice, H. Paul (1989): Studies in the Way of Words. Harvard: Harvard University Press.
Gruber, Jeffrey (1976): Lexical structures in syntax and semantics. Amsterdam: North Holland.
Haiman, John (1994): “Iconicity”, in: Ronald E. Asher (ed.), The encyclopedia of language and linguistics, vol. 3. Oxford: Pergamon Press, 1629–1633.
Haspelmath, Martin (2005): “Negative indefinite pronouns and predicate negation”, in: Martin Haspelmath / Matthew Dryer, David Gil / Bernard Comrie (eds.), The World Atlas ofLanguage Structures. Oxford: Oxford University Press, 466–469.
Heaton, Raina (2017): A typology of antipassives, with special reference to Mayan. PhD University of Hawaiʻi at Mānoa.
Hilpert, Martin (2019): Construction Grammar and its application to English (second edition). Edinburgh: Edinburgh University Press.
Hjelmslev, Louis (1928): Principes de grammaire générale. Copenhagen: Bianco Lunos.
Hjelmslev, Louis (1953 ): Prolegomena to a theory of language (Danish original 1943). Baltimore: Waverly Press.
Hjelmslev, Louis (1961 ): Prolegomena to a theory of language (second edition) (Danish original 1943). Madison: The University of Wisconsin Press.
Hockett, Charles (1958): A course in modern linguistics. New York: Macmillan.
Hoffmann, Thomas / Graeme Trousdale (2013): “Construction grammar: Introduction”, in: Thomas Hoffmann / Graeme Trousdale (eds.), The Oxford handbook of Construction Grammar. Oxford: Oxford University Press, 1–12.
Höllein, Dagobert (2021): “Coseriu, significative semantics and a new system of semantic roles”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu: past, present and future. Berlin / Boston: De Gruyter, 261–278.
Horn, Laurence R. (1991): “Duplex negatio affirmat. The economy of double negation”. Chicago Linguistic Society 27(2), 80–106.
Horn, Laurence R. (2001): A natural history of negation. Stanford (Calif.): CSLI Publications.
Horn, Laurence R. (2010): “Multiple negation in English and other languages”, in: Laurence R. Horn (ed.), The expression of negation. Berlin / New York: Mouton de Gruyter, 111– 148.
Humboldt, Wilhem von (1963 ): “Über das vergleichende Sprachstudium in Beziehung auf die verschiedenen Epochen der Sprachentwicklung”, in: Wilhelm von Humboldt, Schriften zur Sprachphilosophie (Werke in fünf Bänden, vol. III), edited by Andreas Flitner / Klaus Giel. Darmstadt: Wissenschaftliche Buchgesellschaft, 1–25.
Humboldt, Wilhem von (1963 ): “Über das Entstehen der grammatischen Formen, und ihren Einfluss auf die Ideenentwicklung”, in: Wilhelm von Humboldt, Schriften zur Sprachphilosophie (Werke in fünf Bänden, vol. III), edited by Andreas Flitner / Klaus Giel. Darmstadt: Wissenschaftliche Buchgesellschaft, 31–63.
Humboldt, Wilhem von (2017 ): “Grundzüge des allgemeinen Sprachtypus. Als Einleitung zu ausführlichen Untersuchungen über die Amerikanischen Sprachen”, in: Wilhelm von Humboldt, Von den Sprachen Amerikas zum allgemeinen Typus der Sprache (Schriften zur Sprachwissenschaft, vol. IV, 1), edited by Bettina Lindorfer / Jürgen Trabant. Paderborn: Schöningh, 151–276.
Humboldt, Wilhem von (1998 ): Über die Verschiedenheit des menschlichen Sprachbaues und ihren Einfluss auf die geistige Entwicklung des Menschengeschlechts, edited by Donatella Di Cesare. Paderborn: Schöningh.
Itkonen, Esa (2003): What is language? A Study in the Philosophy of Linguistics. Turku: University of Turku.
Jakobson, Roman (1959): “On linguistic aspects of translation”, in: Reuben A. Brower (ed.), On Translation. Cambridge MA: Harvard University Press, 232–239.On Translation. Cambridge MA Harvard University Press, 232–239.
Jakobson, Roman (1962): “Retrospect”, in: Roman Jakobson, Selected Writings, vol. I: Phonological studies. The Hague: Mouton, 631–658.
Jakobson, Roman (1971 ): “Signe zéro”, in: Roman Jakobson, Selected Writings, vol. II: Word and Language. The Hague: Mouton, 211–219.
Jensen, Viggo Bank (2021): “Coseriu’s Hjelmslev”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu. Past, present and future. Berlin / Boston: De Gruyter, 97–114.
Kabatek, Johannes (2000): “Einheitlichkeit der Bedeutung, Designat und Integrale Linguistik”, in: Bruno Staib (ed.), Linguistica romanica et indiana. Festschrift für Wolf Dietrich zum 60. Geburtstag. Tübingen: Narr, 187–205.
Kabatek, Johannes (2020): “Linguistic norm in the linguistic theory of Eugenio Coseriu”, in:
Franz Lebsanft / Felix Tacke (eds.), Manual of standardization in the Romancelanguagesl, 127–144. Berlin / Boston: De Gruyter.
Kabatek, Johannes (2021): “Eugenio Coseriu on immediacy, distance and discourse traditions”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu. Past, present and future. Berlin / Boston: De Gruyter, 227–244.
Kant, Immanuel (1956 [1781/1787]): Kritik der reinen Vernunft. Hamburg: Meiner.
Kay, Paul (1996): Argument structure: Causative ABC constructions. University of California, Berkeley, ms. [online: https://www1.icsi.berkeley.edu/~kay/bcg/5/lec05.html, last access: 19.09.2022]
Kennedy, Christopher (2019): “Ambiguity and vagueness: An overview”, in: Claudia Maienborn / Klaus von Heusinger / Paul Portner (eds.), Semantics: Lexical structures and adjectives. Berlin / Boston: De Gruyter, 236–271.
Klein, Devorah / Gregory Murphy (2001): “The representation of polysemous words”, Journal of Memory and Language 45, 259–282.
Klein, Devorah / Gregory Murphy (2002): “Paper has been my ruin: conceptual relations of polysemous senses”, Journal of Memory and Language 47, 548–570.
Kulikov, Leonid (2011): “Voice typology”, in: Jae Jung Song (ed.), The Oxford handbook of linguistic typology. Oxford: Oxford University Press, 368–398.
Langacker, Ronald W. (1987): Foundations of Cognitive Grammar, vol. 1: Theoretical prerequisites. Stanford: Stanford University Press.
Lebsanft, Franz (2005): “Kommunikationsprinzipien, Texttraditionen, Geschichte”, in: Angela Schrott / Harald Völker (eds.), Historische Pragmatik und historische Varietätenlinguistik in den romanischen Sprachen. Göttingen: Universitätsverlag Göttingen, 25–43.
Levinson, Stephen (2000): Presumptive meanings: The theory of generalized conversational implicature. Cambridge: MIT.
Malchukov, Andrej / Martin Haspelmath / Bernard Comrie (2010): “Ditransitive constructions: A typological overview”, in: Andrej Malchukov / Martin Haspelmath / Bernard Comrie (eds.), Studies in ditransitive constructions. A comparative handbook. Berlin: De Gruyter, 1–64.
Martinet, André (1957): “Arbitraire linguistique et double articulation”, Cahiers Ferdinand de Saussure 15, 105–116.
Newmeyer, Frederick (1992): Iconicity and generative grammar, Language 68, 756–796.
Ogden, Charles K. / Ivor A. Richards (1923): The meaning of meaning. A study of the influence of language upon thought and of the science of symbolism. New York / London: Harcourt, Brace and Co. / Kegan Paul, Trench, Trubner and Co.
Palmer, Frank (1994): Grammatical roles and relations. Cambridge: Cambridge University Press.
Paul, Hermann (1920 ): Prinzipien der Sprachgeschichte (fifth edition). Halle: Niemeyer.
Proost, Kristel (2015): “Verbbedeutung, Konstruktionsbedeutung oder beides? Zur Bedeutung deutscher Ditransitivstrukturen und ihrer präpositionalen Varianten”, in: Stefan
Engelberg / Meike Meliss / Kristel Proost / Edeltraud Winkler (eds.), Argumentstruktur zwischen Valenz und Konstruktion. Tübingen: Narr, 157–176.
Reichling, Anton (1963): Das Problem der Bedeutung in der Sprachwissenschaft. Innsbruck: Sprachwissenschaftliches Institut der Leopold-Franzens-Universität.
Røreng, Anita (2011): Die deutsche Doppelobjektkonstruktion. Eine korpusbasierte Untersuchung zur relativen Abfolge nominaler Akkusativ- und Dativobjekte im geschriebenen Deutsch. PhD Universitetet i Tromsø:Tromsø.
Sabel, Joachim (2002): “Die Doppelobjekt-Konstruktion im Deutschen”. Linguistische Berichte 190, 229–244.
Sanford, Anthony / Patrick Sturt (2002): “Depth of processing in language comprehension: not noticing the evidence”, Trends in Cognitive Sciences 6, 382–386.
Saussure, Ferdinand de (1922 ): Cours de linguistique générale, edited by Charles Bally / Albert Sechehaye (third edition, 1931). Paris: Payot.
Saussure, Ferdinand de (1968): Cours de linguistique générale, critical edition by Rudolf Engler, vol. I. Wiesbaden: Harrassowitz.
Schlieben-Lange, Brigitte (1983): Traditionen des Sprechens. Elemente einer pragmatischen Sprachgeschichtsschreibung. Stuttgart: Kohlhammer.
Sperber, Dan / Deirdre Wilson (1995): Relevance: Communication and cognition (second edition). Oxford: Blackwell.
Stefanowitsch, Anatol (2011): “Argument structure: Item- based or distributed?”. Zeitschrift für Anglistik und Amerikanistik 59 (4), 369–386.
Swiggers, Pierre (1984): Les conceptions linguistiques des Encyclopédistes. Etude sur la constitution d’une théorie de la grammaire au siècle des Lumières. Heidelberg: Groos & Leuven: University Press.
Tai, James (1985): “Temporal sequence and Chinese word order”, in: John Haiman (ed.),
Iconicity in Syntax. Amsterdam / Philadelphia: John Benjamins, 49–72.
Tai, James (2002): “Temporal sequence in Chinese: A rejoinder”, in: Lily I-wen Su, Chinfa Lien / Kawai Chui (eds.), Form and function: Linguistic studies in honor of Shuanfan Huang. Taipei: Crane, 331–351.
Taylor, John (1999): “Cognitive semantics and structural semantics”, in: Andreas Blank / Peter Koch (eds.), Historical semantics and cognition. Berlin: Mouton de Gruyter, 17–48.
Taylor, John (2002): Cognitive Grammar. Oxford: Oxford University Press.
Taylor, John (2003): Linguistic categorization (third edition). Oxford: Oxford University Press.
Trabant, Jürgen (1990): Traditionen Humboldts. Frankfurt am Main: Suhrkamp.
Trabant, Jürgen (2021): “The essence of language: on Coseriu’s philosophy of language”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu. Past, present and future. Berlin / Boston: De Gruyter, 47–61.
Trubetzkoy, Nikolaj S. (1958 ): Grundzüge der Phonologie (reprint). Göttingen: Vandenhoeck and Ruprecht.
Welke, Klaus (2019): Konstruktionsgrammatik des Deutschen: Ein sprachgebrauchsbezogener Ansatz. Berlin / Boston: De Gruyter.
Willems, Klaas (1997): Kasus, grammatische Bedeutung und kognitive Linguistik. Ein Beitrag zur allgemeinen Sprachwissenschaft. Tübingen: Narr.
Willems, Klaas (2011): “Meaning and interpretation: the semiotic similarities and differences between Cognitive Grammar and European structural linguistics”, Semiotica 185, 1–50.
Willems, Klaas (2013): “The linguistic sign at the lexicon-syntax interface: Assumptions and implications of the Generative Lexicon Theory”, Semiotica 193, 233–287.
Willems, Klaas (2016): “The universality of categories and meaning: a Coserian perspective”. Acta Linguistica Hafniensia 48 (1), 110–133.
Willems, Klaas (2020): “Remarks on the ditransitive construction in German”, Sprachwissenschaft 45 (2), 141–180.
Willems, Klaas / Thomas Belligh (2022): “The legacy of structuralism. Structuralist notions in contemporary linguistics”, in: Lorenzo Cigana / Frans Gregersen (eds.), Studies on structuralisms. Copenhagen: Danish Royal Academy of Sciences and Letters.
Willems, Klaas / Cristinel Munteanu (2021): “Introduction”, in: Klaas Willems / Cristinel Munteanu (eds.), Eugenio Coseriu. Past, present and future. Berlin / Boston: De Gruyter, 1–44.
Ziem, Alexander / Alexander Lasch (2013): Konstruktionsgrammatik: Konzepte und Grundlagen gebrauchsbasierter Ansätze. Berlin / Boston: Mouton de Gruyter.
* This paper is the expanded English version of the presentation I gave on 17 June 2021 at the Coseriu 100 International Conference “Coseriu’s linguistics – origin and actuality” organised at the University of Zürich. I thank the organisers for the invitation and two anonymous reviewers for their comments on an earlier version of this paper. The usual disclaimers apply.
1 It must be kept in mind, though, that natural language is a historical cultural artefact created by speakers in a community, to be neatly distinguished from artificial “codes” such as traffic signs, mathematical notations, musical notations, etc. (Bühler 1934: § 3, Reichling 1963, Coseriu 1992 for discussion).
2 The first elaborate presentation of this distinction is provided in Coseriu (1975 ). Because the term “norm” is used in modern linguistics to refer to different phenomena, including prescriptive norms rather than descriptive abstractions, I henceforth use “normal language use” to refer to Coseriu’s descriptive level of shared norms in language use (Kabatek 2020 for a succinct discussion of Coseriu’s concept of norm).
3 The importance of the level of elocutional knowledge and its role in the activity of speaking is duly acknowledged, mutatis mutandis and under various labels, in many contemporary schools of thought, including analytic philosophy, cf. W. Quine’s ‘principle of charity’, D. Davidson’s ‘principle of rational accommodation’, among others (see Davidson 2001).
4 I adopt the terminological distinction between the language-specific “Agentive” role, which designates an encoded “semantic” role, and the universal non-encoded “Agent” role, which designates a “thematic” role, from Coseriu (1979: 45). In (1-2) the thematic role Agent corresponds to an underspecified semantic role “Handlungsträger” (‘carrier of the action’, Coseriu 1987a: 37; Höllein 2021: 272).
5 “in jedem Augenblick ist das wirklich Gesagte weniger als das Ausgedrückte und Verstandene” (Coseriu 1975 [1955¬–1956]: 276, my translation). This crucial difference is already highlighted by Humboldt, compare Humboldt (1963 ), (1963 ) and (1998 ): ch. 20).
6 “Das Wissen bei der sinnvollen Interpretation [auch des Sinnwidrigen] ist ein Wissen darüber, mit welchen Maximen man spricht. Man nimmt bestimmte Grundlagen des Sprechens an […] d.h. man nimmt im voraus an, dass derjenige, der spricht, es mit Kohärenz und sinnvoll tut. Wenn der Ausdruck auf den ersten Blick nicht kohärent ist, dann sucht man nach einer Kohärenz. Man macht dies deswegen, weil man annimmt, dass das Sprechen sozusagen kohärent zu sein hat und weil man in dieser Hinsicht Vertrauen zu den anderen hat. […] Bei der Interpretation des Gesagten gilt also das Prinzip des Vertrauens” (Coseriu 2007 : 95-96, my translation). Cf. Lebsanft (2005: 26-27) for discussion.
7 “Das Sprechen ist eine universelle allgemein-menschliche Tätigkeit, die jeweils von individuellen Sprechern als Vertretern von Sprachgemeinschaften mit gemeinschaftlichen Traditionen des Sprechenkönnens individuell in bestimmten Situationen realisiert wird” (Coseriu 2007 : 70, my translation).
8 Compare Coseriu (1975 [1955–1956]: § 3) and (2007 : 111-137) for an overview of the settings that can be taken into consideration in a comprehensive model of text-linguistics.
9 Coseriu’s explanation of the differences between the meaning of words and the meaning of grammatical items and constructions is based on Humboldt’s account of the relationship between the lexicon and grammar in terms of the Aristotelian distinction between “matter” and “form” and Humboldt’s new conception of the word between “sign” and “image” (or “symbol”), see in particular Humboldt (1963 ), (2017 : 204-237) and (1998 : 220-240); cf. also Di Cesare (1998: 46-51 and 66-73).
10 In his outline of a functional syntax that is centred on grammatical paradigms (“grammatische Paradigmatik”, Coseriu 1987a: 145), Coseriu distinguishes between five types of meaning. These five types are: lexical meaning (“lexikalische Bedeutung“), categorial meaning (“kategorielle Bedeutung”, i.e. the meanings of parts of speech), instrumental meaning (“instrumentelle Bedeutung”, e.g. the meaning contributed by word order, conjunctions, prepositions, prosody, etc.), syntactic meaning (“syntaktische Bedeutung”) and ontic meaning (“ontische Bedeutung”) (Coseriu 1987a: 149-150). In order not to complicate matters, I here focus on the last two types of meaning.
11 It is worth recalling that, for Coseriu, this characterisation of lexical meanings does not preclude their analysis based on semantic features (Coseriu 2001: 355-369). It is also evident that combinations can themselves become part of the lexicon, e.g. conventionalised word formations, phrasal idioms, etc. (see Coseriu 1978: § 4 and 1992 : ch. 11).
12 Goldberg initially proposed a narrow definition of “construction” (Goldberg 1995) which she later extended to a broader definition (e.g., in Goldberg 2006). The extension concerns the second sentence in the following definition and introduces a component of storage related to frequency of use (Goldberg 2006: 5):
Any linguistic pattern is recognized as a construction as long as some aspect of its form or function is not strictly predictable from its component parts or from other constructions recognized to exist. In addition, patterns are stored as constructions even if they are fully predictable as long as they occur with sufficient frequency.
The broad definition has remained controversial as it poses a number of theoretical and methodological problems which thus far only have been dealt with occasionally (Stefanowitsch 2011).
13 The English dative alternation has been a research topic in English linguistics for over half a century, in particular since Green (1974). The contrast between “possession” and “motion”, too, has a long pedigree in the research literature on the alternation, at least since Gruber (1976).
14 The Double Object Construction also exists in German, but as a gradually obsolete constructional pattern which is used only with a handful of verbs that require two accusative objects, viz. lehren ‘teach’, abfragen, abhören ‘test’ and bitten ‘ask’) (Willems 2020: 158).
15 The common practice to present one salient interpretation as the meaning or central sense of a word formation can be found, e.g., in Taylor’s (2003: 172-176) treatment of the diminutive in Italian. According to Taylor, the diminutive is a polysemous category which lacks a “common meaning core” but whose various distinct meanings are linked to a “central sense”, i.e. ‘smallness in physical space’ (2003: 175). Among the many diminutive suffixes that exist in Italian, Taylor mentions the verbal suffixes -icchiare and -ucchiare and provides the example parlare ‘speak’ > parlucchiare ‘speak (a foreign language) badly’. This paraphrase is incorrect because it does not capture the meaning of the diminutive suffix but only one of its manifold uses. Parlucchiare can be used in a variety of senses, including il bambino già parlucchia ‘the child already speaks a little’, io parlucchio il cinese ‘I speak a little Chinese, but only badly’, ne abbiamo parlucchiato ‘we briefly talked about it, but without going into detail’. The general encoded meaning of parlucchiare can be paraphrased as ‘speak without performing the activity completely’ (I thank Claudia Crocco for providing me with the examples).
16 Here I disregard, in order not to complicate matters, the no less important insight that truth conditions are no part of semantics proper, contrary to what is often assumed in the work of analytic philosophers and pragmaticists (Coseriu 2015, I: 71-92, with reference to Aristotle).
17 “Denn die Wörter der historischen Sprachen, zumindest die des primären und rein sprachlichen Wortschatzes, nennen – auf unmittelbare Weise – nicht die ‘Dinge’, sondern Vorstellungen, intuitiv erfasste quidditates. Jeder Primärausdruck der ‘natürlichen’ Sprachen entspricht ursprünglich einer νόησις τῶν ἀδιαιρέτων (einer apprehensio simplex bzw. indivisibilium intelligentia), und keiner fest umrissenen Klasse von Gegenständen oder Tatsachen. In diesen Sprachen benennt man keine schon klassifizierten Gegenstände; im Gegenteil, man klassifiziert dort die Gegenstände (und die ‘Tatsachen’) unter Bezugnahme auf die Bedeutungen” (Coseriu 1987a: 8, my translation; italics in the original).
18 For a cogent account of this stance, I refer in particular to Coseriu’s lengthy discussion of Hegel’s philosophical conception of linguistic signs and natural language, to which Coseriu is deeply indebted (Coseriu 2015, II: 333-350; cf. also 1977a).
Copyright (c) 2022 Klaas Willems
This work is licensed under a Creative Commons Attribution 4.0 International License.