This week, I will attend the Workshop on Systematic Semantic Change,  which will take place this Friday and Saturday (April 5-6) at UT Austin.  Besides me, the speakers are: Patricia Amaral (UNC), Andrea Beltrama(Chicago), Cleo Condoravdi (Stanford), Ailis Cournane (Toronto), Östen Dahl (Stockholm), Regine Eckardt (Göttingen), Chiara Gianollo (Köln), Larry Horn (Yale), Lukasz Jedrzejowski (ZAS), Roumyana Pancheva (USC), Mike Pham (Chicago), Scott Schwenter (OSU), Julia Thomas (Chicago), Elizabeth Traugott (Stanford), and Igor Yanovich (MIT).

The  program looks very promising and I am looking forward to the workshop. I will talk about Pragmaticalization and multidimensional semantics. Here is the abstract:

Overview     We investigate the relation between the empirical phenomenon of pragmaticalization in grammaticalization and a theoretical framework of multidimensional semantics in the Pottsian tradition. In such a framework, the development from descriptive into expressive meaning can be modeled as a diachronic type shift from ordinary truth-conditional into expressive types. An intermediate stage is provided by McCready’s mixed types. Grammaticalization theory can profit from the formal perspective, while the formal semantics can profit from the additional data the diachronic development of expressive meaning provides. For instance, the need for intermediate stages, provides further evidence against Potts’s claim that there are no mixed expressives. To highlight these interrelations, we present a case study involving the development of discourse markers from subordinators in German …

Pragmaticalization      Pragmaticalization is a special subtype of grammaticalization (Diewald 2011) and can be defined as the “the development of a grammatically identifiable expression of speaker belief or speaker attitude to what is said.” (Traugott 1995: 32). A corresponding pragmaticalization path can be given as follows (Traugott 2003: 633).

(1)   propositional (> textual) > expressive meaning

Some of the best-studied cases of pragmaticalization include pejorations such as English boor ‘countryman, farmer’ > ‘crude person’ (Traugott 2003: 634) or the development of German modal particles out of adjectives or adverbs (see, amongst many others, Autenrieth 2002). While pragmaticalization as an account of semantic change is empirically and conceptually well-grounded, we think that it would nevertheless profit from an adaption of more recent developments in formal semantics in which the notions of expressive meaning and multidimensionality have received a lot of interest, especially thanks to the influential work by Potts (2005, 2007). The main hypothesis which we want to pursue in this talk is that pragmaticalization can be modeled as a (diachronic) semantic type shift in such a framework.

Multidimensional semantics     Potts framework and its extensions (Gutzmann 2011; McCready 2010) rests on two core ideas. (i) Sentences can have two meaning dimensions: a descriptive and an expressive one. (ii) The semantic types encode what kind of content is contributed by an expression and regulate the flow of information between the meaning dimension by means of dedicated application rules.

Informally, we can notate multidimensionality by prefixing the expressive dimension with the bullet “●” to separate it from the descriptive dimension.

(2)   |That damn Kaplan got promoted| = got-promoted(kaplan) ∶ tdamn(kaplan) ∶ u

Beside the standard descriptive types for e, t, s and the functional types that can be built from them by the standard definition, there is also a special type for expressive propositions, which I call u for use-conditional (following Gutzmann 2012) and which contrasts with the truth-conditional base type t. For instance, the expressive adjective damn in (2) is of type ⟨e, u⟩. Applying it to its argument yields an expressive proposition (of type u) while returning its argument unmodified.

(3)   |damn Kaplan| = kaplanedamn(kaplan) ∶ u

Diachronic type shifts     In light of this framework, pragmaticalization can be understood as the development of descriptive into expressive expressions. Formally, this means that the type of an expression shifts diachronically from a descriptive into an expressive type. Take boor again, for instance. It starts as an ordinary descriptive predicate and develops into a functional expletive expressive item that displays a negative expressive speaker attitude, labeled simply as boor-ex here.

(4)   boor ∶ ⟨e, t⟩ > boor-ex ∶ ⟨e, u

This (diachronic) type shift formally reflects the change from propositional to expressive meaning as depicted in (1). Of course, diachronically, such type shifts do not happen suddenly but evolve during complex processes and in contexts that support such changes (Traugott 2003).

Most often, pejorations like in (4) start from a conversational implicature that, given a sufficiently high frequency, become conventionalized. Given the right circumstances, this implicature then may become part of an expression’s lexical meaning. In a final stage, the original meaning may get lost, so that only the negative expressive component remains from the originally descriptive predicate. We therefore observe something like the following pattern, which complies with the so-called overlap model of grammaticalization (Heine 2003: 590).

(5)   A > A, B > B

In addition to the arguments in McCready (2010) and Gutzmann (2011), the need for such an intermediate stage add further evidence against Potts’s (2005: 7) claim that no lexical item contributes to both meaning dimensions. To model this stage, we employ McCready’s mixed expressions which consist of a descriptive and an expressive part conjoined with the diamondd “◆” and have a product type. Returning to boor, this means that after the conventionalization of the conversational implicature, the truth-conditional content is still there and we have a mixed expressive. It is only at the last stage, that the first dimension is lost and we arrive at a non-mixed expletive expressive.

(6)   boor ∶ ⟨e, t⟩ > boor boor-ex ∶ ⟨e, t⟩ × ⟨e, u⟩ > boor-ex ∶ ⟨e, u

Case study: Discourse markers in German     To put the formal approach to pragmaticalization to use, we investigate the pragmaticalization of the adversative subordinating conjunction obwohl ‘although’, as in (7), into a discourse particle marking corrections, as in (8) (Günthner 1999).

(7)   Peter ist im Kino, obwohl er keine Zeit hat. P. is at.the cinema although he no time has “Peter is at the cinema, although he has not time.”

(8) Peter ist im Kino, obwohl – er hat keine Zeit. P. is at.the cinema although he has no time “Peter is at the cinema, (correction: but wait,) he as no time.”

Beside the witnessed semantic change – in (7), there is one assertion with a descriptively relevant connection, wheras in (8), there are two speech acts where the later revises the former – the pragmaticalization of obwohl is followed by a corresponding syntactic change in word order. While the second clause exhibits subordinate word order (verb in final position) in (7), it has the root clause word order (verb-second) in (8). Given some assumption about the semantic motivation of verb movement in German (Truckenbrodt 2006) , we show that this syntactic change is a direct reflect of the semantic type shift. Since expressive expressions are arguably invisible (Potts et al. 2009) for the rules that govern verb movement in German, the presence of obwohl in (8) does not stop the verb from moving into the verb-second position that is associated with an independent speech act.


A pdf of the abstract can be obtained here. After the workshop, I will also post the slides of my talk here.