According to the results of Helati and Juhász, in English into Hungarian translations personal pronouns change to zero pronouns (Heltai & Juhász, 2002) and the English into Hungarian direction favours implicitation instead of explicitation.

Investigations on lexical cohesion in English into Hungarian and Hungarian into English translation, based on Hoey’s model of lexical repetition, found differences between novice and expert translators: experts favoured simple repetition, and their texts had stronger ties formed by lexical repetition (Klaudy & Károly, 2000).

Bakti (2020b) investigated shifts of grammatical cohesion in Hungarian target texts translated and simultaneously interpreted from the same English source language text. Her results show that demonstrative references and conjunctions were better preserved from the source text in translated target texts in contrast to simultaneously interpreted target texts, in other words, more demonstratives and conjunctives were omitted during SI than during translation. In addition, results show that during the process of SI more cohesive devices were added than during translation, in other words, in the simultaneously interpreted target texts there were more conjunctions and referential ties for 100 words of target text than in the translated target texts. A possible explanation can be that simultaneous interpreters add more cohesive ties to the target texts as a result of the linearity constraint in SI, that is because they work with unfinished sentences. Another possible explanation is that, because of the concurrent, active analysis of the source text, unfinished sentences and repetitions characterize the target language output of simultaneous interpreters, leading to the proliferation of demonstratives. These words, together with other function words, are the most frequently repeated elements among disfluencies (Gyarmathy, 2009).

Bakti (2018) compared the use of cohesive devices in consecutively interpreted and sight translated Hungarian target language texts. Results show that more cohesive devices were omitted during consecutive interpreting than during sight translation. The high proportion of omissions during CI is in line with previous results (Gumul, 2012). As concerns cohesive ties per 100 words of texts, in the consecutively interpreted target texts the proportion of both referential and conjunction ties were higher than in the source language texts or in the sight translated target texts. This might be due to the fact that in consecutive interpreting, interpreters actively analyze the ST and then recreate and strengthen grammatical cohesion ties during target text production.

This leads to the following research questions: