Clitics and translation effects: the case study of the Czech pronominal dative clitic mu and its correlates in Polish parallel texts

1. Introduction 

The paper addresses the hypothesis of translation universals (Chesterman 2000, 2010) applied to the syntax of Slavic pronominal clitics. The translation effects regarding clitic placement are rather understudied from the contemporary perspective, but the diachronic studies on Czech of Kosek et al. 2022 derive the variation of clitic placement in the Bible translations into Czech from the influence of Vulgate as a source language. In contrast to this investigation the approach taken here is synchronic and oriented towards cross-Slavic investigations. 

In general, modern Czech pronominal clitics are permanent Wackernagel enclitics (see Uhlířová, et al. 2017) with rare deviations related to, for instance, to colloquial. Polish unstressed pronominal forms pose a bigger challenge, as their behavior is less uniform. They have more freedom in the word order and seem to be subject to a grammatical change (Jung & Migdalski 2021). Kulik (2018) is the first, who examines the quantitative distribution of the dative pronominal form mu in the spoken variety. His results show considerable distributional variation, with almost 60% preference for Wackernagel position and nearly 30% for the third position in the clause. 

The starting point of the present study are therefore the Czech dative pronominal clitic mu (3sg masculine or neuter) and the corresponding Polish dative pronouns: weak mu and strong jemu as translation equivalents. The exact research questions are: 

What are the distributional properties of the Polish personal pronouns mu and jemu in translation, in comparison to the distributional properties of the corresponding Czech pronominal clitic mu in the source text, and in comparison to the distributional properties of the Polish personal pronouns mu and jemu in the Polish literary and spoken texts? Can I observe any translation effects similar to the processes known from language contact (pattern or matter replication Matras 2009)? Can I trace other translation effects? 

2. Material and method 

To approach the problem, I use parallel corpus InterCorp (Čermák & Rosen 2012), from which I extracted the parallel occurrences of 3SG.M|N.DAT in Czech-Polish texts, where Czech is the language of the original. Following the approach of Kosek et al. 2022, I study the following factors: absolute position of the dative pronoun in the clause and the governor (predicate) related position. Additionally, the frequencies are compared to the baseline frequencies computed from the samples of the written and spoken Polish. 

3. Preliminary results 

The analysis of 3SG.M|N.DAT translations of four Czech novels shows several kinds of translation effects. In quantitative terms, the form mu is overrepresented in translated texts in comparison to the baseline frequency of mu in the Polish National Corpus (PNC). The qualitative inspection of data shows that translators try to reduce this effect, for instance in multiply compound structures. This process can be explained by various translation universals, among them, reduction of repetition (Baker 1993) and the shift towards reader in coherence and cohesion (Blum-Kulka 1986). 

As for placement, while the Czech form mu is mostly placed on the second position in clause, the Polish form mu shows variation being rather placed directly after the predicate in positions two, three, and four. This corresponds mostly to the empirical distribution expected for the placement in literary Polish. The translation of Hašek‘s novel diverges slightly from this pattern, as the second-position effect is here stronger. Since the distribution approaches here the distribution obtained from the spoken Polish of Kulik (2018), this effect can be better explained as a stylistic variation corresponding to the spoken style of the original novel than as a direct translation effect.