Pragmatic frames as a tertium comparationis in translation

Semantic frames, in the tradition of Charles Fillmore (Fillmore 1982), provide an approximation to a tertium comparationis in contrastive linguistics and translation (Boas 2013). A frame is a structured set of interrelated concepts, modelling the cognitive backdrop against which speakers understand word meanings. The primacy of frame model of trans­lation (Czulo 2017), drawing on frame semantics and construction grammar, establishes the principle of maximum frame comparability between source and target text. The principle of choosing the semantically closest translation can be overridden by pragmatic factors relating to the function of the expressions at hand, e.g. considerations of information structure.

However, also the functional dimension of meaning can be represented by frames and in this way be integrated into the model. Pragmatic frames (Czulo, Ziem, and Torrent 2020) model “how we conceptualize what is going on between the speaker and the hearer, or between the author and the reader” (Fillmore 1982, 117), assuming a conventionalized relationship between a function and its formal realization. So far, this dimension is only marginally present in the frame semantic lexicographic resources.

This talk will present preliminary results of an ongoing PhD project that aims at developing new pragmatic frames to be included in the German FrameNet (GFN 2025) and using them in translation analysis. A bilingual corpus of German and English texts in two registers with a parallel and a comparable subcorpus is used as a testing ground to develop new frames and showcase their application in cross-linguistic analysis. The following questions are addressed: How is a communicative function, described in terms of a pragmatic frame, realized in two languages and translations between them? Do we find patterns or tendencies comparing original vs. translated texts? 

The communicative functions studied include metacommunicative comments adding information how the speaker wants their utterance to be taken, as in example (1), marking contrast, as in (2), as well as other phenomena relating to speaker-hearer interaction, text and information structuring and stance-taking. In (1) we find differences in wording between the German original and the two English translations, the human translation leaving out the reference to addressee and speaker. Yet, all three versions function as a Metacommunicative­_comment on the question that follows. In (2), the contrast is lexically marked and thus frame-evoking only in the first German translation. The translator chose to split up the long English sentence and insert a contrastive connective, possibly following the preference for a more detailed and ‘didactic’ manner of presentation in German popular scientific texts (Böttger and Probst 2001). Indeed, in the parallel subcorpus, the Contrast-frame is evoked more frequently in German originals and translations.

Within functional approaches to translation theory, models developed for describing the pragmatically relevant features of a text and its situational setting (House 1997; Nord 2009) propose to derive individual translation choices from the functional characterisation of the text and the communication situation as a whole; the text being considered the primary unit of translation. The frame approach, in contrast, addresses pragmatic phenomena that are not necessarily derivable from the text level. The application of frames as tertium comparationis in (machine) translation evaluation (Czulo et al. 2019), benefits from integrating the functional-pragmatic dimension. Adding pragmatic frames to FrameNets thus further improves the potential that these resources have for this and other computational applications.
 

Examples

1. a.  [_PropositionAlso etwa die Frage] – [_{Metalinguistic_comment}lassen Sie es mich ruhig ad personam sagen]^Target: [_PropositionWoher kommt denn dieser Herr Barroso?]  (European Parliament, Gerald Häfner, 22/11/2012, German original)

b.  [_PropositionMy question] – [_{Metalinguistic_comment}and it relates to one specific person]^Target – [_Propositionis this. Where did this Mr Barroso come from?] (English translation)

c. [_PropositionSo, for example, the question] – [_{Metalingusitic_comment}let me put it bluntly]^Target: [_Propositionwhere does this Mr Barroso come from?] (English machine translation (DeepL))

2. a. They said the study didn't consider warm sea-surface temperatures in Mexico, which Trenberth claimed added significant moisture to the atmosphere, increasing the total precipitation. (Lois Parshley: Blame Game. Scientific American 06/2023, English original)

b. Wie sie bemängelten, hatte die Studie nicht berücksichtigt, dass [_{State_of_affairs}die Meeresoberfläche vor der Westküste Mexikos ungewöhnlich warm gewesen war]. [Doch]^Target [_{Contrasted_state_of_affairs}gerade deshalb habe sich deutlich mehr Feuchtigkeit in der Atmosphäre befunden, wodurch die Menge an Niederschlag stieg]. (German translation)

c. Sie argumentierten, die Studie habe die warmen Meeresoberflächen­temperaturen in Mexiko nicht berücksichtigt, die laut Trenberth der Atmosphäre erhebliche Feuchtigkeit zuführten und damit die Gesamtniederschlagsmenge erhöhten.  (German machine translation (DeepL))