Glossary

Idioconstructions carry pragmatic meaning about the communicative context (usually abbreviated as C, e.g. C_home or C_German) that they are associated with. A construction can be said to be in C-agreement with other constructions that mark (a) exactly the same type of communicative context, (b) a compatible, but more or less C-specific type, or (c) any construction that is not restricted to specific communicative contexts. For example, within a German-Swedish bilingual community, the lexical idioconstructions [hej ‘hello’ 〈C_Swedish〉] and [hej då ‘bye’ 〈C_Swedish〉] agree with each other, as do [hej ‘hello’ 〈C_Swedish〉], [tjena ‘hello’ 〈C_Swedish; informal〉] and the Polar Question Diaconstruction [v_fin¹ subj … ], whereas [hej ‘hello’ 〈C_Swedish〉] and [hallo ‘hello’ 〈C_German〉] do not agree. C-agreement often corresponds to a the pretheoretical notion of a language.

The cognitive argument (‘Languages interact in multilingual speakers’ cognition’) says that multilingual speakers’ linguistic knowledge is organized in a way that allows for all their languages to be activated jointly. Individual languages are neither stored nor processed in isolation from each other, but interact with each other. Hence, any model of linguistic knowledge that does not allow this cognitive interaction is bound to be cognitively unrealistic (cf. socio-cognitive realism).

The cognitive argument is one of six important arguments advocating a focus on multilingualism within CxG (cf. constructionist argument, factual argument, historical argument, sociolinguistic argument, variability argument).

A communicative context is in a broad sense what sociolinguists call a domain, i.e., generally speaking, any communicatively relevant constellation of interlocutors, topics, speech parameters, and sociocultural environments as well as communicative modes and intentions. Examples include ‘speaking with one’s family at home’, ‘informal communication with colleagues’, and ‘religious gatherings’. Multilingual communities tend to conventionally associate particular domains (or sets of domains) with particular languages (cf. the sociolinguistic argument). In DCxG, communicative contexts are defined as the domains or sets of domains that are relevant enough for speakers or communities to be marked by linguistic means, viz. by using idioconstructions. When describing the pragmatic meaning of an idioconstruction, the context-marking function is usually abbreviated as C_domain or C_language (e.g. C_home or C_German), referring to a specific domain or to the set of domains conventionally associated with a specific language (cf. formalization).

A community is a spatio-temporally stable-ish group of speakers, i.e. a group that is stable enough to conventionalize speech patterns that can be described in terms of a communal constructicon. Communities may overlap, and any individual speaker may belong to multiple communities at once. Communities may vary considerably in size. For example, both a family of three and the entire population of a large country can be considered communities in this sense, provided that the members share linguistic conventions with each other that they do not share with outsiders.

The constructicon consists of all the constructions entrenched in an individual speaker’s mind or conventionalized within a community. The constructicon consists of constructions that are organized into (at least) one constructional network. Multilingual speakers and communities organize their linguistic knowledge into one single multilingual constructicon rather than several language-specific ones.

The constructionist argument (‘It’s constructions all the way down’) says that if CxG claims that linguistic knowledge is organized into a constructicon consisting of constructions and interconstructional links, then the coexistence and use of multiple languages by the same speakers and within the same communities must be modelled within that constructicon – without making use of any additional conceptual tools or machinery (cf. Procrustean cone). If, on the other hand, constructionist models failed to include multilingual phenomena, this would amount to a falsification of one of the central tenets of CxG.

The constructionist argument is one of six important arguments advocating a focus on multilingualism within CxG (cf. cognitive argument, factual argument, historical argument, sociolinguistic argument, variability argument).

A counter-diasystematic change is a diachronic change that increases the proportion of idioconstructions at the expense of diaconstructions within the multilingual constructicon. From a cognitive perspective, counter-diasystematic change is less expected than its opposite, i.e. pro-diasystematic change, as it hinders rather than facilitates cognitive processing in multilinguals. However, counter-diasystematic change enables speakers to unambiguously mark communicative contexts.

Idioconstructions carry pragmatic meaning about the communicative context (usually abbreviated as C, e.g. C_home or C_German) that they are associated with. This type of meaning can be more or less specific. For example, the Swedish lexical construction [hej ‘hello’ 〈C_Swedish〉] is less C-specific than [tjena ‘hello’ 〈C_Swedish; informal〉]. When an idioconstruction instantiates a diaconstruction, C-specificity is generally lower for the diaconstruction.

The degree of diasystematicity is the proportion of diaconstructions (as opposed to idioconstructions) within a multilingual constructicon. Structural similarities between the languages involved facilitate a higher degree of diasystematicity, but they only determine the diasystematic potential for a given pair or set of languages; the actual degree of diasystematicity has to be determined based on empirical evidence. However, the degree of diasystematicity is a conceptual notion rather than a quantifiable property. While it cannot be measured directly, different constructicons can be ranked on an ordinal scale. For example, a constructicon that has undergone pro-diasystematic change can be assumed to have a higher degree of diasystematicity than its precursor.

A diaconstruction (diacxn) is a construction that (a) does not carry pragmatic meaning that marks different communicative contexts associated with different languages (otherwise it would be an idioconstruction) or (b) carries less pragmatic meaning of this type than an idioconstruction that instatiates it. For example, in a Danish-German bilingual community, speakers cannot mark different communicative contexts by using a *German Polar Question construction instead of a *Danish one, because polar questions are marked in the same way in both languages, viz. by constructing finite-initial clauses. This is reflected in the assumption of a Polar Question diaconstruction for this community. It is important to note that a diaconstruction can be used in more contexts than an idioconstruction, but carries less information.

Diaconstructionhood is gradual and relational in the sense that a construction can be more or less C-specific. A construction that marks a communicative context that in turn is part of a larger set of communicative contexts (such as a language) can be considered a diaconstruction in its relation to more C-specific idioconstructions (that mark, say, a register of that language), while at the same time being an idioconstruction in its relation to less C-specific diaconstructions (that are language-unspecific).

DCxG is called diasystematic as a nod to Weinreich’s idea of structural systems that encompass more than one linguistic variety. Such structural systems, called diasystems, were thought to connect pre-existing individual systems with partial similarities, in which single elements such as sounds, words, or grammatical constructions could be interlingually identified with each other. The idea of systematic relations between individual elements and their interlingual equivalents as well as more abstract interlingual units correspond to horizontal and vertical diasystematic links within the multilingual constructicon. The whole constructicon could also be seen as corresponding to the diasystem. However, since structural systems in general do not play a role in usage-based CxG with its focus on individiual constructions and the networks they form, the term diasystem is hardly used in DCxG.

A diasystematically anchored innovation is an innovation (that may or may not eventually lead to diachronic change) that fills a schematic diaconstruction in a way that results in forms that are non-canonical from a monolingual perspective, but comprehensible to the relevant community. For example, Danish-German bilinguals might use an ad hoc word regnskærm ‘umbrella’ instead of the canonical form paraply based on the lexical idioconstructions [regn ‘rain’] and [skærm ‘screen’], a compound noun construction, and schematic lexical diaconstructions that capture the equivalence of Danish regn and German Regen as well as Danish skærm and German Schirm.

Multilingual speakers and communities that use two or more structurally similar languages can form diaconstructions more easily than those that use less similar languages. Hence, the diasystematic potential of structurally similar languages is higher. High diasystematic potential is a necessary, but not a sufficient condition for a high degree of diasystematicity: While speakers of, say, Japanese and Hebrew are less likely to establish a high number of diaconstructions within their muiltilingual constructicon than, say, speakers of Spanish and Portuguese, the actual degree of diasystematicity has to be determined based on empirical evidence.

While pro-diasystematic change, from a cognitive perspective, can be considered the most likely structural outcome of language contact, another expectable consequence is the preservation of an already high degree of diasystematicity in multilingual constructicons, which prevents innovations that might lead to counter-diasystematic change. This is called diasystematic stability.

As DCxG is (usage-based) vanilla CxG without the Procrustean cone, evidence for the existence of idio- and diaconstructions has to meet the same criteria as in any other usage-based approach. However, as schematic constructions tend to be more important in multilingual constructicons than in monolingual ones, evidence in favour of constructional extensibility plays a more prominent role, including not only productivity in the traditional sense but also receptivity.

In addition to the six core arguments advocating a focus on multilingualism within CxG (cf. cognitive argument, constructionist argument, factual argument, historical argument, sociolinguistic argument, variability argument), there is also the evolutionary argument (‘Language evolved in multilingual settings’). This argument says that since early human populations consisted of small exogamous hunter-gatherer groups whose members were exposed to different languages or varieties, humans’ capacity for language must have evolved in multilingual settings from its earliest beginnings. Hence, the ability to store, process, and use different communicative systems is, and has always been, central to human language.

The extensibility of a schematic construction is its ability to be used in combination with new slot-fillers, resulting in the extension of a pre-existing schema to new types. The type of extensibility that is best known and easiest to study is productivity, i.e. extensibility in production. An example for the productive extensibility of a schematic diaconstruction is the production of diasystematically anchored innovations (e.g. non-canonical Danish regnskærm ‘umbrella, literally: rain screen’, which bilinguals might produce based on, among other things, German Regenschirm). However, because speakers (can) process an incomparably greater number of linguistic utterances receptively than they produce themselves, extensibility in reception (receptivity) is at least as important, even though it is harder to study. An example for the receptive extensibility of a schematic diaconstruction is the comprehension of novel technical terms by multilingual speakers (e.g. Norwegian diakonstruksjon, as a regular equivalent to, say, English diaconstruction or Afrikaans diakonstruksie, Italian diacostruzione, or Portuguese diaconstrução).

The factual argument (‘Language contact is everywhere’) says that, measured on a global and historical scale, most speakers and communities are multilingual (or multilectal) at least to some extent. Hence, multilingual knowledge and the use of different languages within the same individual or community are the rule rather than an exception, and no usage-based approach can exclude multilingualism or view it as somehow peripheral.

The factual argument is one of six important arguments advocating a focus on multilingualism within CxG (cf. cognitive argument, constructionist argument, historical argument, sociolinguistic argument, variability argument).

In DCxG, diaconstructions in particular can be formally ‘empty’ – or maximally schematic – in the sense that they do not contain any lexical or phonological material on the formal side. Formally empty constructions capture the fact that a community uses different idioconstructions in different languages to express the same language-unspecific concept, e.g. a lexical concept such as ‘Tuesday’ or a grammatical concept such as the T-V-distinction (e.g. German du vs. Sie, Danish du vs. De or French tu vs. vous). However, formally ‘empty’ constructions do contain formal information. For example, the Danish-German Tuesday diaconstruction not only captures the lexical concept ‘Tuesday’, but also the fact the this concept is expressed using single phonological words in both languages.

As a usage-based constructionist approach, DCxG is generally flexible in terms of formalization. However, formalization is often seen as useful, for example as a convenient tool for shorthand notation or as a means of achieving consistency and precision in one’s reasoning. Although the following conventions are not obligatory, they can be used as guidelines:

• constructions are notated in boxes or square brackets
• formal information must be given, functional information can be omitted: [kaffe ’coffee’], [kaffe]
  • referential meaning is indicated by single quotation marks: [kaffe ’coffee’]
  • non-referential meaning (e.g. grammatical or pragmatic properties) is indicated by an approximate description in angle brackets: [v-er 〈agentive noun〉], [tjena 〈C_Swedish; informal〉]
• filled elements (e.g. word forms) are italicized: [coffee], [v-ing]
  • phonological material is not italicized, but indicated by slashes (phonemic notation), square brackets (phonetic transcription), or other symbols commonly used by phonologists and phoneticians: [/ˈkɒfi/], [[ˈkʰɒfi]], [ˈσσ_ω]
  • formally empty elements are indicated by underscores: [__ω]
• slots are indicated by small capitals: [subj v]
  • additional requirements for slots (e.g. morphological forms) can be indicated in subscript: [subj_nom v_fin]
  • unspecified slots are indicated by a small capital x: [x-ish]
  • alternatives can be given in curly brackets: [{adj, adv}-er]
  • Latin-script variables (i, j, k, …) can be used to index (co-)referentiality: [subj_i write obj_j ‘a person_i writes a text_j’]
  • Greek-script variables (α, β, γ, …) can be used to indicate (identical) attribute values: [n_number:α adj_number:α]
• lexical paradigms are indicated by italicized small capitals: [bite the dust]
• additional components (compulsory or optional) of a construction (left out in the description) are indicated by ellipsis marks: [sneeze obj …]
• components of a construction are separated by commas: [v_fin, subj]
  • adjacent elements of constructions can be separated by spaces or, within words, hyphens: [drink coffee], [v-ing coffee]
  • non-segmentable elements are joined by a backslash: [v\past]
  • the linear order of elements can be indicated in superscript: [v_fin¹, subj²], [adv¹, v_fin², subj^≤4], [v_fin^final]
• regular expression (regex) syntax can be used to mark more complex patterns including alternatives and optionality: [adj|adv] (adjectives or adverbs), [art adj* n] (an article, zero or more adjectives, and a noun), [art_def adj_def+ n_def] (a definite article, one or more definite adjectives, a definite noun)

The historical argument (‘Language contact changes grammar’) says that languages that are used in multilingual communities tend to undergo changes that typically lead to gradual structural convergence of the languages involved, potentially resulting in (or at least approximating) construction-per-construction translatability. Such changes can be explained by reorganizational processes that facilitate a simplification of the multilingual constructicon (cf. pro-diasystematic change), even if complexifying changes (cf. counter-diasystematic change) also occur.

The historical argument is one of six important arguments advocating a focus on multilingualism within CxG (cf. cognitive argument, constructionist argument, factual argument, sociolinguistic argument, variability argument).

An idioconstruction (idiocxn) is a construction that (a) carries pragmatic meaning that marks different communicative context associated with different language (otherwise it would be a diaconstruction) or (b) carries more pragmatic meaning of that type than a diaconstruction that it instantiates. For example, in a Danish-German bilingual community, speakers can mark different communicative contexts by using the German word Dienstag instead of its Danish equivalent tirsdag (‘Tuesday’). This is reflected in the assumption of both a German Tuesday idioconstruction and a Danish Tuesday idioconstruction for this community, potentially linked through a formally empty Tuesday diaconstruction (cf. formal emptiness). It is important to note that an idioconstruction can be used in fewer contexts than a diaconstruction, but carries more information.

Idioconstructionhood is gradual and relational in the sense that a construction can be more or less C-specific. A construction that marks a communicative context that in turn is part of a larger set of communicative contexts (such as a language) can be considered a diaconstruction in its relation to more C-specific idioconstructions (that mark, say, a register of that language), while at the same time being an idioconstruction in its relation to less C-specific diaconstructions (that are language-unspecific).

Diaconstructions can emerge in different ways. One possibility is interlingual identification: Multilingual speakers identify different idioconstructions as variants of the same form-function pair, based on (partial) similarities on the formal side, the functional side, or both sides. Interlingual identification can be thought of, in constructionist terms, as the establishment of a horizontal link between idioconstructions, which may then form the basis for the formation of a (typically more schematic) diaconstruction that is instantiated by and vertically linked to the idioconstructions.

DCxG assumes that languages are not a priori relevant in the cognitive organization of linguistic knowledge. Firstly, languages in the traditional (vague) sense are not distinguishable (on purely linguistic grounds) from other types of varieties such as dialects or registers (cf. the variability argument). Secondly, in a broad usage-based sense, languages can be defined as the sets of constructions that are co-occur within the same communicative contexts, including – in multilingual communities – both idioconstructions used to mark these contexts and diaconstructions without this type of pragmatic meaning. Language-specificity is an optional part of the pragmatic meaning of a construction.

Schematicity is usually thought of as lexical schematicity: A schematic construction contains a slot (or several slots) that is filled with lexical (or morphological) material. For example, the v slot in a [subj v_fin] construction can be filled with any finite verb, and the v slot in [v-er] is filled with a verbal stem (e.g. walk in walker). Especially from a DCxG perspective, however, phonological schematicity is also important.

Multilectalism is essentially the same as multilingualism, but at the level of varieties (of one language), i.e. the coexistence of more than one variety within the same speaker, community, society, or region or within constructicons, texts, conversations, utterances, and parts of utterances.

Since languages are not distinguishable (on purely linguistic grounds) from other types of varieties such as dialects or registers (cf. the variability argument), multilectalism cannot (and, hence, should not) be distinguished from multilingualism.

The concept of multilectalism implies the term monolectalism as a counterpart. However, as far as we know, no speaker or community only masters a single variety of a language and uses it indiscriminately in all communicative contexts: Monolectalism is unattested (cf. the factual argument).

Insofar as the concept of language can be defined at all in a socio-cognitively realistic way, multilingualism is the coexistence of more than one language within the same speaker, community, society, or region. Consequently, individual and communial constructicons as well as texts, conversations, utterances, and parts of utterances can also be said to be multilingual.

Since languages are not distinguishable (on purely linguistic grounds) from other types of varieties such as dialects or registers (cf. the variability argument), multilingualism cannot (and, hence, should not) be distinguished from multilectalism. Likewise, multilingualism is not categorically different from more specific notions such as bilingualism, trilingualism, and quadrilingualism. DCxG makes no theoretical distinction between these concepts either and mostly uses multilingualism as an umbrella term.

Unlike traditionally assumed in linguistics, multilingualism plays an important role in human communication; this is reflected in six key arguments that suggest a focus on multilingualism within CxG (cf. variability argument, factual argument, historical argument, cognitive argument, sociolinguistic argument, constructionist argument).

Phonological language markers (PLMs) are phonologically schematic idioconstructions that consist of submorphemic sounds, sound sequences, or suprasegmentals on the formal side and mark the communicative context as part of their pragmatic meaning. For example, a speaker of Swedish who is infrequently exposed to Danish might notice a very frequent non-word-initial approximant [ð̞ˠ] in Danish speech and store this sound as a marker of Danish contexts, without having any more substantial knowledge. This would correspond to a PLM [[ð̞ˠ] 〈C_Danish〉] or, formalized more precisely, [x+[ð̞ˠ]x* 〈C_Danish〉].

PLMs are key in modelling regular interlingual sound correspondences in DCxG. For example, Danish-Swedish bilinguals might identify Danish [u] as a regular equivalent to Swedish [ʉ] in cognate words (in certain positions). In this case, both sounds qualify as PLMs ([u] 〈C_Danish〉] and [[ʉ] 〈C_Swedish〉]); additionally, the correspondence between both PLMs is captured by a phonologically schematic diaconstruction (PSD) that consists of a slot that can alternatively be filled with either PLM and that marks the communicative context as either Danish or Swedish, i.e. [[u]_i|[ʉ]_j 〈C_Danish;i|C_Swedish;j〉]. PLMs and PSDs interact with schematic lexical diaconstructions (SLDs) that specify the shared word form and contain phonological slots, e.g. in [[hV_u|ʉːs] ‘house’], which contains a vowel slot that can be filled by either the Danish or the Swedish PLM, resulting in Danish hus [huːˀs] and Swedish hus [hʉːs].

Obviously, the formalization of PLMs, PSDs, and SLDs tends to be complex and awkward. However, this is of secondary importance – phonologically schematic constructions can also be (and often are) described in a more informal manner.

Schematicity is usually thought of as lexical schematicity. Especially from a DCxG perspective, however, phonological schematicity is also important, since phonologically schematic idioconstructions such as phonological language markers as well as phonologically schematic diaconstructions and phonologically schematic lexical diaconstructions are needed to model regular interlingual sound correspondences.

Whether such phonologically schematic constructions are socio-cognitively real, is an empirical question that has to be answered based on evidence. It is important to note, however, that the need for CxG to deal with phonological schematicity is not restricted to multilingual phenomena; other examples include constructional models of phonaesthemes (e.g. English gl- ‘vision, light’ as in glitter, gleam, glisten etc.) as well as non-concatenative morphology (e.g. German tr-x-nk– ‘drink’ as in trinken (infinitive), trank (indicative preterite 3rd person singular), getrunken (past participle) etc.).

Phonologically schematic diaconstructions (PSDs) are diaconstructions that are needed to model regular interlingual sound correspondences, together with schematic lexical diaconstructions (SLDs) and phonological language markers (PLMs; see there for more details).

In DCxG, pragmatic meaning is understood as the conventionalized pragmatic functions associated with a construction. While referential meaning is typically written in quotation marks (e.g. ‘bird’), non-referential meaning, including pragmatic meaning, is usually given using angle brackets (e.g. 〈informal〉). In particular, marking communicative contexts is part of the pragmatic meaning of idioconstructions.

DCxG is vanilla construction grammar applied to language contact and multilingualism, i.e. it does not require any additional conceptual tools or machines on top of what (usage-based) CxG needs anyway. In particular, the analysis of language contact phenomena is not analysed in terms of interactions between separate languages, but in a radically bottom-up manner that does not take the existence of languages for granted. In extending the icecream metaphor, the rejection of the individual language as an a priori concept in grammar is described as getting rid of the Procrustean cone (or cornet) around vanilla construction grammar.

A pro-diasystematic change is a diachronic change that increases the proportion of diaconstructions at the expense of idioconstructions within the multilingual constructicon. From a cognitive perspective, pro-diasystematic change is more likely than its opposite, i.e. counter-diasystematic change, as it facilitates cognitive processing in multilinguals.

Productivity is productive extensibility, i.e. extensibility in language production (as opposed to receptivity). An example for the productivity of a schematic diaconstruction is the production of diasystematically anchored innovations (e.g. non-canonical Danish regnskærm ‘umbrella, literally: rain screen’, which bilinguals might produce based on, among other things, German Regenschirm).

Receptivity is receptive extensibility, i.e. extensibility in language reception (as opposed to productivity). An example for the receptive extensibility of a schematic diaconstruction is the comprehension of novel technical terms by multilingual speakers (e.g. Norwegian diakonstruksjon, as a regular equivalent to, say, English diaconstruction or Afrikaans diakonstruksie, Italian diacostruzione, or Portuguese diaconstrução).

Schematic lexical diaconstructions (SLDs) are diaconstructions that are needed to model regular interlingual sound correspondences, together with phonologically schematic diaconstructions (PSDs) and phonological language markers (PLMs; see there for more details).

Construction grammar is a non-modular approach that views all form-function pairs as forming a continuum. This continuum has often been called the lexicon-syntax continuum, as it includes both lexically filled and lexically variable constructions. However, while lexical and syntactic constructions differ in their degree of schematicity, they do not form the poles of the continuum, which arguably includes, for example, utterance-level and discourse-level constructions as well as submorphemic and sublexical units (such as phonological language markers); moreover, lexical constructions can be phonologically schematic as well (e.g. lexical schematic diaconstructions). As such types of constructions are of particular importance in DCxG, it seems more appropriate to use the term ‘schematicity continuum’ rather than ‘lexicon-syntax continuum’.

DCxG aims to model language contact phenomena in a way that is consistent with what is known about multilingual speakers and communities from research in contact linguistics, sociolinguistics, and usage-based linguistics. In particular, it assumes that multilingualism does not merely mean the (non-functional) coexistence of two or more languages, but rather that these languages are used in a socially meaningful way within a community (cf. the sociolinguistic argument) and that they interact in individual speakers’ cognition (cf. the cognitive argument).

Hence, the analysis of a language system in a multilingual environment includes constructions from all languages or varieties involved, and the social establishment and individual acquisition of such a system is seen as inherently multilingual.

The sociolinguistic argument (‘Multilingualism is functional’) says that different languages (or varieties) tend to be used for different purposes in multilingual (or multilectal) communities rather than be chosen freely. This social convention is reflected in individual speakers’ linguistic knowledge as an association of individual constructions with specific (sets of) communicative contexts as part of the constructions’ pragmatic meaning.

The sociolinguistic argument is one of six important arguments advocating a focus on multilingualism within CxG (cf. cognitive argument, constructionist argument, factual argument, historical argument, variability argument).

DCxG is ordinary construction grammar applied to language contact and multilingualism, without additional conceptual tools or machines on top of what (usage-based) CxG needs anyway. Hence, DCxG is what Croft has called ‘vanilla construction grammar’. Extending the icecream metaphor, one might describe DCxG as being vanilla construction grammar without the Procrustean cone.

The variability argument (‘We can’t distinguish languages from other types of varieties’) says that ‘language’ is a notoriously ill-defined concept in linguistics. In particular, languages are not distinguishable (on purely linguistic grounds) from other types of varieties such as dialects or registers.

Hence, even if linguists were to doubt the ubiquity on multilingualism (at least on a global or historical scale (cf. the factual argument, as far as we know, no speaker or community only masters a single variety of a language and uses it indiscriminately in all communicative contexts: Monolectalism is unattested (cf. multilectalism). Consequently, usage-based approaches to language can neither distinguish multilingualism and monolectalism nor view them as somehow peripheral.

The variability argument is one of six important arguments advocating a focus on multilingualism within CxG (cf. cognitive argument, constructionist argument, factual argument, historical argument, sociolinguistic argument).