The project builds on two basic assumptions developed in functionally oriented theories of language: a usage-based view of language, and the relevance of diachrony for the explanation of synchronic patterns. From the point of view of usage, frequency of collocations plays a crucial role in determining how constructions come into being, are grammaticalized, or fall out of usage. Frequency explains how certain changes have come about and others have not. Thus, research on constructions must be implemented through extensive corpus studies with the usage of quantitative methods. A starting point for such research is constituted by already available resources. The relevance of diachrony in the explanation of attested synchronic patterns has been pointed out by several typologists. Concerning argument structure, a number of general semantic and pragmatic motivations have been proposed for particular synchronic patterns, including e.g. alternations in argument structure and alignment types. A crucial question is whether or not these motivations really play a role in the actual diachronic processes that give rise to the relevant patterns from one language to another, which are often based on context-driven inferences, leading to the reanalysis of highly particularized source constructions, both at the syntactic and at the semantic and pragmatic level. While numerous instances of such processes have been collected within studies of language change in general, their actual implications for explanatory theories of argument structure have failed to be systematically assessed so far.

A usage based approach to language implies exploring language structures in actual usage, as attested in corpora. For our purposes, the availability of extensive electronic resources is of paramount importance.

A number of databases are planned as a result of the project. In the first place, we will produce various databases of verb argument structures extracted from texts. They will be structured using the interface provided by the partner project Index Thomisticus. Using this interface, the Homeric Dependency Lexicon (HoDeL) has recently been released. A Valency Lexicon of Old Church Slavonic, a Hittite Treebank with a data base of Hittite verbs (Inglese Towards a Hittite Treebank. Mambrini et al. eds. Proceedings of CRH Workshop, Warszawa: Polish Academy of Sciences, 2015), and a treebank of Old Irish glosses are currently being planned, as a part of the proposed project. In addition, a Latin database will be produced, based on the Proiel and Perseus treebanks, which will complement the Index Thomisticus. Depending on personnel possibly recruitable through the project, we will consider producing new resources for Gothic and Old Indo-Aryan. Argument structure constructions in the Index Thomisticus and in HoDeL have been extracted from syntactically annotated corpora (treebanks). Thus, a preliminary step for the construction of new data bases consists in assembling existing treebanks, as well other digitalized texts that can be enriched with syntactic annotation for languages for which treebanks are not available. Collections of digitalized texts are numerous and concern most Indo-European languages. Currently, we are or have been collaborating with Proiel, Perseus, Trolling, The Sanskrit Library, SLUW. The Proiel treebank has recently been extended to host treebanks for medieval stages of Romance languages: it will provide a starting point for assembling the diachronic data on historical developments in Romance. For Old Italian (Old Florentine) valency patterns will also be investigated through the TLIO/GATTO corpus. As regards Latin and Romance, we will carry out a thorough, corpus-based study of argument structure constructions particularly in relation to voice systems (Cennamo, Voice. Ledgeway & Maiden eds. The Oxford Handbook of the Romance Languages, Oxford:OUP 2016) and (in)transitive alternations, focusing on the encoding of split intransitivity (Cennamo, Late Latin pleonastic reflexives and the Unaccusative hypothesis, Transaction of the Philological Society 1999/97:103–50), argument realization, existentials and presentatives, the anticausative alternation (Cennamo & Jezek, The anticausative alternation in Italian, Massariello & Dal Masi eds. Le Interfacce, Rome:Bulzoni, 2011:809–23), factitive constructions as well as the syntax of infinitives, gerunds and past participles. Starting from a representative selection of verbs, a description of their syntactic and semantic valency will be provided, as well as of the overall system of valency rearranging and valency changing strategies. Variation in the frequency of individual patterns will be taken as indicating degrees of constructionalization, and will explain how argument structure constructions change over time. A limited section of the work on Italian will be devoted to the acquisition of transitivity in Italian as L2 by speakers with different linguistic backgrounds, based on earlier work (Jezek & Rastelli, Gradiente di inaccusatività e verbi pronominali nell'apprendimento dell'italiano come seconda lingua. Bernini et al. a cura di, Competenze lessicali e discorsive nell'apprendimento di lingue seconde. Perugia: Guerra 2008, 95-115). Outcomes from this part of the research will specifically be used for the implementation of guidelines for the possible application of envisaged results to language teaching (cf. section on the socio-economic impact of the project). The analysis and description of the diachronic development of argument structure constructions presupposes the availability of an extended and evidence-based inventory of argument structure constructions for Modern Italian. We will initially use the data in ValPaL for Modern Italian (Cennamo & Fabrizio Italian valency patterns. Hartmann et al. eds. ValPaL. Leipzig: Max Planck Institute for Evolutionary Anthropology 2013), and select the verbs accordingly. We will then expand our inquiry through the usage of additional resources built in collaboration with the computational linguistics community and benefitting from techniques of data extraction developed in this field: SIMPLE (Lenci et al. SIMPLE: A general framework for the development of multilingual lexicons. International Journal of Lexicography. 2000/13.4: 249-263) and T-PAS (Jezek et al. T-PAS A Resource of Typed Predicate Argument Structures for Linguistic Analysis and Semantic Processing. Calzolari et al. eds Proceedings of LREC 2014. Paris: ELRA 2014), both focusing on the semantics and syntactic properties of argument structure constructions and their interplay with the verbal lexicon. Remarkably, while the ValPaL is based on data collection relying of questionnaires, thus on the speaker’s knowledge, SIMPLE and T-PAS take a different perspective, by relying of data extracted from corpora. Within our project, we will collaborate with SIMPLE by examining the annotation schemes and linguistic specifications of both resources in view of the alignment of the two. The resulting resource will be used for the study of transitivity-related phenomena from a diachronic perspective based on corpus evidence (Jezek, Struttura argomentale dei verbi. Renzi & Salvi, a cura di Grammatica dell'Italiano Antico. Bologna: Il Mulino, 2010, 77-122). We will provide a template for classification of argument structure constructions, which will be matched with ValPaL, and made compatible with it, in order to maximize communication and data sharing with other projects. A further stage of our project foresees the semi-automatic enrichment of the resulting aligned resource based on supervised methods applied to corpora and on the Corpus Pattern Analysis (CPA) lexicographic procedure, already employed in international setting for the development of argument structures repositories for various languages (English, Spanish and Czech, cf. Hanks Lexical Analysis. Cambridge Mass.:MIT Press 2013). Argument structure constructions will be investigated in ancient I(ndo)-E(uropean) languages, taking into account case marking of nominal and pronominal arguments with finite verb forms, the possible occurrence of referential null arguments (Luraghi Omission of the direct object in Classical Latin. IF 1997/102:239-257, Sausa & Zanchi Non-accusative null objects in the Homeric Dependency Treebank. Mambrini et al. eds. Proceedings of CRH Workshop, Warszawa:Polish Academy of Sciences. 2015, 107-115 for a treebank-based study), case marking of arguments with verbal nouns, voice (including the paradigmaticization of passive voice, Luraghi The extension of the passive construction in Ancient Greek. Acta Linguistica Hafniensia 2010/42:60-74), basic valency as defined in Nichols et al. (Transitivizing and detransitivising languages. Linguistic Typology 2004/8.2:149-211, see also Luraghi Basic valency orientation in Hittite. Studies in Language 2012/36:1-32), and, for Old Irish, the use of initial mutations in argument marking (see Roma, Le mutazioni iniziali delle lingue celtiche. R. Lazzeroni et al. a cura di, Diachronica et Synchronica. Pisa: ETS, 2008:453-478). We will compare results across IE languages, and will indicate to what extent argument structure constructions and related features of argument realization can be reconstructed as belonging to Proto-IE syntax, following Barðdal (Syntax and Syntactic Reconstruction. Bowern & Evans eds. The Routledge Handbook of Historical Linguistics, London:Routledge 2015:343-373) and connecting with on-going international projects on this topic, notably EVALISA. The frequency of similar patterns in individual languages and across IE languages will be taken as evidence of earlier or later constructionalization. Argument structure constructions extracted from ancient IE languages and historical stages of Romance languages, as well as data from modern Romance varieties will then be matched against the data in ValPaL, and a classification will be proposed, based on the guidelines emerging from alignment of T-PAS with SIMPLE. The resulting information will be stored in a database, as a diachronic implementation of this tool.A complementary perspective pertains to the origins of individual argument structure patterns. Alignment systems have been shown to often originate from the reinterpretation of pre-existing constructions, though grammaticalization or other processes of form-meaning redistribution within complex expressions. For example both ergative and accusative alignment can develop through the reinterpretation of nominalizations. Ergative alignment can also develop as ergative markers arise from agent markers in passive-like constructions, instrumentals, cislocatives, or indexical elements, while accusative alignment can be a result of accusative markers developing from ‘take’ verbs in serial verb constructions, or topic markers. Another source of ergative alignment is an active-inactive system arising from the spread of the accusative, the case most typically used for O arguments, initially to SO arguments and finally to A arguments, the so-called extended accusative (Cennamo, Argument structure and alignment variations and changes in Late Latin, Barðdal & Chelliah eds. The Role of Semantics and Pragmatics in the Development of Case. Amsterdam: Benjamins 2009, 307–346). Cislocatives and third person pronouns give rise to inverse markers, while split intransitivity can be a result of transitive clauses with un- expressed third person agents being reinterpreted as intransitive ones, or transitive constructions with light verbs also being reanalyzed as intransitive. Valency changing (detransitivizing and transitivizing) constructions can also develop through the grammaticalization of pre-existing items, as for example, for passives, causatives, and applicatives. These processes, which have been documented in grammaticalization studies and studies of language change in general, raise a number of questions for theories of argument structure. For example, to what extent are particular argument structure patterns (particular alignment patterns, particular valency changing constructions, or any other type of argument structure pattern) motivated by the properties of specific source constructions, rather than general semantic or pragmatic principles pertaining to those patterns in themselves? To what extent does the distribution of individual patterns across different contexts (different verb classes, different semantic or pragmatic configurations) follow from such principles, and to what extent is it related to the properties of the source construction? What is the role of extension in shaping this distribution? Can specific argument structure patterns (e.g. particular alignment types or particular valency changing constructions) be related to a single underlying principle, or is each pattern an epiphenomenal result of several distinct diachronic processes, each motivated by a different principle?

While some of these issues have occasionally been addressed in the diachronically oriented typological literature (e.g. Gildea 1998 cit., Creissels Direct and indirect explanations of typological regularities: the case of alignment variations. FoLin 2008/42:1–38, Cristofaro Cognitive explanations, distributional evidence, and diachrony. Studies in Language 2012/36:645–70, Competing motivations and diachrony: what evidence for what motivations? MacWhinney et al. eds. Competing motivations in grammar and usage, Oxford:OUP, 2014:282–98), a comprehensive picture is still missing both of the possible origins of many argument structure patterns cross-linguistically, and of the general implications of the processes that give rise to these patterns for explanatory theories of argument structure. This part of the project aims to cast light on these issues by collecting cross-linguistic data on the various developmental processes that give rise to individual argument structure patterns and determine their distribution across different contexts in individual languages. This data will be analyzed in light of the theoretical questions outlined above, which will provide a new theoretical background against which to evaluate the cross-linguistic distribution of the relevant argument structure patterns, as revealed both by existing databases and databases collected within the present project. The data collected will be made available to the wider linguistic community in the form of an electronically searchable database, arranged in terms of individual argument structure patterns (and subtypes thereof), and their possible sources. This will innovate with respect to most existing databases on argument structure, which focus on the distribution of particular patterns (either at the synchronic level, that, is, from one language to another, or at the diachronic level, that is, at different stages of the same language, or in genetically related languages), rather than on the processes that give rise to these patterns. The organization of the database will also depart from many existing collections of data on grammaticalization in that the latter focus on specific source constructions and their possible evolution, rather than the various possible sources of specific patterns resulting from different processes of grammaticalization or diachronic evolution in general. This will innovate with respect to most existing databases on argument structure, which focus on the distribution of particular patterns (either at the synchronic level, that, is, from one language to another, or at the diachronic level, that is, at different stages of the same language, or in genetically related languages), rather than on the processes that give rise to these patterns.

Home