1 Overview: What is DEmA?
The Pavia Diachronic Emergence of Alignment (DEmA) database is a comprehensive open access database on the diachronic emergence of alignment patterns cross-linguistically, that is, diachronic processes that are known to have given rise to specific alignment patterns in different languages (e.g. ergative, accusative, etc.).
By alignment pattern is meant, in a maximally general sense, any possible grouping of the three argument roles A, S, and P (Dixon 1994, Comrie 1989), in terms of case marking, indexation, or other morphosyntactic phenomena. Progress in grammaticalization studies and the study of language change cross- linguistically means that a comparatively large body of data is now available on the emergence of alignment patterns in a variety of languages across different families and geographical areas (see, for example, Gildea 1998 on Carib; König 2008 on African languages; Bubenik 1998, Haig 2008 and 2017, Verbeke 2013 on Indo-Aryan). This evidence, however, has not yet been integrated into a comprehensive overview of the possible sources and developmental mechanisms that can give rise to particular alignment patterns (for example, accusative, ergative, or active) from one language to another. An early study in this direction is Harris and Campbell 1995: chap. 9, which, however, concentrates on possible mechanisms of alignment change, rather than the specific alignment patterns emerging through each mechanisms, or the source constructions that can give rise to individual patterns. Another strand of cross-linguistic research (e.g. Heine and Kuteva 2002; Kulikov 2006) has focused on the etymology of particular case markers, irrespective of the contexts and developmental mechanisms that lead to these markers evolving from particular source elements, or the consequences of this process for the alignment patterns of the language.
In general, research on the emergence of alignment patterns in individual languages has shown that individual patterns typically emerge from pre-existing constructions, through various mechanisms of constructional reinterpretation or, sometimes, processes of phonological change. Information about these processes is, however, scattered across specialized publications, and often not easily comparable from one language to another, nor accessible to non-specialists. The main goal of DEmA is to provide an expanding platform where the available evidence on these processes is integrated in a typologically informed framework that makes it possible to compare different processes from one language to another, so as to obtain data both on the emergence of alignment patterns in particular languages, and on the possible sources and developmental processes leading to the emergence of particular alignment patterns cross-linguistically. This type of data can be used to address different research questions about the diachronic origins of alignment (Harris and Campbell 1995; Gildea 1998; Mithun 2005; Creissels 2008; Cristofaro 2012, 2013, 2014, among others):
· What source constructions give rise to particular alignment patterns cross- linguistically?
· What developmental mechanisms lead from particular source constructions to particular alignment patterns?
· What is the relationship between the properties of particular source constructions and developmental mechanisms and the properties of the resulting alignment pattern, in terms for example of what argument roles are or are not encoded in the same way, or the distribution of the pattern across different contexts (NP-based and TAM-based alignment splits, or other types of splits)?
· The same alignment patterns (for example, ergative or accusative alignment) originate from different source constructions and through different developmental mechanisms in different cases. Can individual patterns be explained in terms of some overarching principle that applies to all instances of the pattern, or should different instances of the pattern be explained in terms of different principles depending on the developmental processes involved?
2 The organization of DEmA
In DEmA, each entry is a process that has led to the development of a new alignment pattern in some language, as described in published sources. The current release features data on monotransitive alignment (i.e. alignment of one- and two- place verbs) only. Languages are identified by means of their Glottocode, linking each entry in DEmA to the Glottolog database. We also indicate the genealogical affiliation of each language (we only note the macro-family classification, e.g. Indo-European or Sino-Tibetan; genealogical information is taken from Glottolog). In line with a number of cross-linguistically oriented accounts (see, for example, Harris and Campbell 1995: Chap. 9), the development of a new alignment pattern is conceived as a process that takes place within particular constructions, for example through the reinterpretation of the argument structure of these constructions, or through the development of a new marker for A, P, or S arguments as a result of grammaticalization. This process will lead to the development of a particular alignment pattern for the construction in question, and may have different effects depending on the original alignment pattern of the language in the relevant grammatical domain. For example, the development of a new ergative pattern in perfective constructions may lead to a TAM based split if non-perfective constructions use a non-ergative pattern. If these constructions have ergative alignment, however, the language will remain consistently ergative.
The most innovative feature of DEmA is that it allows for a fine-grained research of the various components involved in the emergence of new alignment pattern (source constructions, original alignment pattern of the language, developmental mechanisms, alignment in the resulting constructions, global effects for alignment patterns in the language). In particular, DEmA is structured so as to provide information about three different grammatical domains:
· The initial situation in the language, including both the original alignment pattern of the language and a detailed description of the source construction involved in the emergence of the new alignment pattern.
· Developmental mechanisms, that is, the nature and dynamics of the change from the original to the new alignment pattern.
· The effects of the process of change, including the alignment pattern that develops in the construction undergoing the change and the effects of this development on the global alignment pattern of the language.
For each of these domains, DEmA provides multiple searchable fields, which are described in detail below. The type of data is also specified which provide the basis for the proposed historical scenario, e.g. historical corpus data or internal reconstruction. The reason for doing so is that different types of data may entail a different degree of reliability of the proposed historical scenario (for instance, diachronic studies based on historical corpus data may be more reliable than those based on internal or comparative reconstruction only).
3 The initial situation in the language
For this domain, two distinct fields are provided:
· Original alignment pattern: This refers to the original alignment patterns attested in the language, along with any constraints in the distribution of this pattern, e.g. accusative, ergative, TAM or NP based splits, and the like. Only the alignment pattern pertaining to the grammatical domain involved in the process of change is taken into account. For example, if a process of change involves alignment in indexation, only the alignment pattern originally found for indexation in the language (and not, for example, case marking alignment) is taken into account.
· Source construction: This refers to the construction that serves as the basis for the development of the new alignment pattern, with particular focus on the specific elements that undergo change in the development of the new alignment pattern (for example, particular lexical items that undergo grammaticalization, particular adpositions or case affixes that undergo a change in their grammatical function). While we try to standardize the terminology used in the description of different source constructions cross-linguistically, this field contains highly heterogenous and language-specific descriptions, e.g. ‘constructions of the type ‘take X (and) VERB (X)”, where the verb ‘take’ and some other verb share a P argument’. This is due to the fact that, for each language, different semantic, pragmatic or morphosyntactic properties of the source construction must be taken into account that play a role in the development of the new alignment pattern.
The distinction between original alignment patterns of the language and source constructions is motivated by the fact that (i) the processes that give rise to a new alignment pattern take place within particular constructions, and may be independent of the alignment patterns previously attested in the language, but (ii) the global effects of individual processes in the language will depend on these patterns. For example, ergative patterns have been shown to develop as intransitive resultative constructions with an oblique NP are reinterpreted as transitive ones, so that the S argument in the intransitive construction becomes a P argument, whereas the oblique NP becomes an A argument (‘X is VERBed by Y > ‘Y ERG VERBed Y’: Gildea 1998, among others). This process will give rise to ergative alignment for resultative constructions, and is independent of the original alignment of S arguments, for example whether they are aligned with A (accusative alignment) or P (ergative alignment). This alignment, however, will determine the global effects of the process in the language. If S arguments were originally aligned with A arguments, for example, this alignment will be retained for non-resultative constructions, leading to a split between accusative alignment in non-resultative constructions and ergative alignment in resultative ones. If S arguments were originally aligned with P arguments, the process will only lead to the development of an additional ergative pattern in the language, specialized for resultative constructions.
4 Developmental mechanisms
For this domain, we provide a number of fields pertaining to various aspects of the processes whereby the source construction gives rise to a new alignment pattern:
· Developmental mechanism: This is a detailed description of mechanisms whereby the source construction gives rise to the new alignment pattern.
· Intermediate stages: This is an optional field that is used in case the historical scenario can be described as unfolding in a number of distinct steps. In some cases, for example, a new alignment pattern initially develops in particular constructions, and is subsequently extended to other constructions.
· Type of change: This field provides a typological classification of different types of developmental mechanisms. While this classification involves abstracting away from the details of individual processes of change (for which the user is referred to the relevant sources), it aims to relate these processes to general mechanisms of change traditionally discussed in grammaticalization studies and historical linguistics. The following types of change are identified (note that multiple such mechanisms may be at play for individual types of change):
– Reinterpretation of argument structure: A change in the argument structure of an existing construction, for example an intransitive construction with an oblique or possessor NP is reinterpreted as a transitive one, so that the S argument and the oblique or possessor NP become the A and P argument of the transitive construction.
– Grammaticalization: An element not originally used to encode grammatical relations (e.g. a verb form, a demonstrative, a topic marker) grammaticalizes into a marker for A, S, or P arguments.
– Extension: the markers used for particular argument roles are extended to other roles (e.g from A to S) or the same roles in other contexts (e.g. from the S arguments of particular intransitive verbs to the S arguments of other intransitive verbs).
– Phonological change, which may lead to the development of specialized forms for particular argument roles (e.g., nouns undergo phonological reduction when used as S and P arguments, leading to a new form for these arguments: Haspelmath and the APiCS Consortium 2013) or the loss of existing specialized forms (e.g., the distinction between A/S and P arguments is lost due to phonological reduction: Blake 2001).
– Loss: This refers to cases where an existing form for some argument role was lost in the language, but there is no clear evidence that this was due to phonological change.
5 The effects of the process of change
For this domain, a number of fields are provided that describe the effects of the process of change leading to the development of the new alignment pattern:
· Resulting construction: this field is similar to the ‘Source construction’ field in that it features a description of the construction resulting from the process of change. Descriptions in this field are schematic but are meant to provide all the details relevant to the understanding of the new alignment pattern, e.g. ‘Transitive construction with an overtly marked P argument’.
· Alignment in the resulting construction: This is the alignment pattern in the construction resulting from the process of change. For example, if an intransitive resultative construction of the type ‘X is VERBed by Y’ is reinterpreted as a transitive one ‘Y VERBed X’, this will give rise to ergative alignment, because X becomes a P argument and is encoded in the same way as the S argument from which is derived, whereas Y becomes an A argument with dedicated marking, because it retains the marking used for the oblique NP from which it is derived.
· Global alignment pattern following the change: This is meant to capture possible differences between the alignment pattern in the construction resulting from the change and the global alignment pattern of the language, as determined by (i) the alignment pattern of the construction resulting from the change and (ii) the alignment pattern of other constructions within the same grammatical domain. For example, some processes of change may give rise to new perfective constructions with ergative alignment. If non-perfective constructions have other alignment patterns, however, the language will end up with a TAM-based alignment split, rather than a global ergative alignment pattern. The labels used to indicate global alignment patterns are standard ones in the literature on alignment, for example ‘nominative-accusative’, ‘TAM-based split ergativity’, or ‘split intransitivity’ (see e.g. Dixon 1994).
· Constraints: This field provides further specification about the alignment splits resulting from the process of change. If there is a TAM or NP based split, for example, the field will specify the exact properties of the split (e.g. perfective constructions vs. nonperfective ones, pronouns vs. nouns, inanimate nouns vs. other NP types).
· Grammatical domain: This refers to the grammatical domains involved in the process of change, for example case marking, indexation, or word order. Particular processes of change may involve multiple grammatical domains, e.g. both case marking and indexation.
· Symmetry: This refers to the morphosyntactic encoding of arguments in the resulting constructions. Symmetric encoding means that all roles are encoded through the same strategy (overt case marking, overt indexation), whereas asymmetric encoding means that different roles are encoded through different strategies (zero vs. overt case marking, zero vs. overt marking in indexation).
References
Blake, B. J. (2001). Case. 2nd edition. Cambridge: Cambridge University Press.
Bubenik, V. (1998). A historical syntax of late middle Indo-Aryan (Apabrahms´a). Amsterdam and Philadelphia: John Benjamins. ˙
Comrie, B. (1989). Language universals and linguistic typology. 2nd edition. Oxford: Basil Blackwell.
Creissels, D. (2008). Direct and indirect explanations of typological regularities: the case of alignment variations. Folia Linguistica 42, 1–38.
Cristofaro, S. (2012). Cognitive explanations, distributional evidence, and diachrony. Studies in Language 36, 645–670.
Cristofaro, S. (2013). The referential hierarchy: Reviewing the evidence in diachronic perspective. In D. Bakker and M. Haspelmath (Eds.), Languages across Boundaries: Studies in the Memory of Anna Siewierska, pp. 69–93. Berlin and New York: Mouton de Gruyter.
Cristofaro, S. (2014). Competing motivations and diachrony: what evidence for what motivations? In B. MacWhinney, A. Malchukov, and E. Moravcsik (Eds.), Competing motivations in grammar and usage, pp. 282–98. Oxford: Oxford University Press.
Dixon, R. M. W. (1994). Ergativity. Cambridge: Cambridge University Press.
Gildea, S. (1998). On reconstructing grammar: Comparative Cariban morphosyntax. Oxford: Oxford University Press.
Haig, G. (2008). Alignment Change in Iranian Languages; A Construction Grammar Approach. Berlin and New York: Mouton de Gruyter.
Haig, G. (2017). Deconstructing Iranian Ergativity. In J. Coon, D. Massam, and L. D. Travis (Eds.), The Oxford Handbook of Ergativity, pp. 465–500. Ox- ford: Oxford University Press.
Harris, A. C. and L. Campbell (1995). Historical syntax in cross-linguistic perspective. Cambridge: Cambridge University Press.
Haspelmath, M. and the APiCS Consortium (2013). Alignment of case marking of personal pronouns. In S. M. Michaelis, P. Maurer, M. Haspelmath, and M. Huber (Eds.), Atlas of Pidgin and Creole Language Structures Online. Oxford: Oxford University Press.
Heine, B. and T. Kuteva (2002). Word Lexicon of Grammaticalization. Cam- bridge: Cambridge University Press.
König, C. (2008). Case in Africa. Oxford: Oxford University Press.
Kulikov, L. (2006). Case systems in a diachronic perspective. In L. Kulikov A. Malchukov, and P. de Swart (Eds.), Case, valency and transitivity, pp. 23–47. Amsterdam and Philadelphia: John Benjamins.
Mithun, M. (2005). Ergativity and language contact on the Oregon Coast: Alsea, Siuslaw, and Coosan. In Proceedings of the Berkeley Linguistic Society, pp. 77–95.
Verbeke, S. (2013). Alignment and Ergativity in New Indo-Aryan Languages. Berlin and New York: Mouton de Gruyter.