Polish to English Translations
Polish to English Translations
The following examples of Polish structures that often resist direct, word-for-word translation into English will be briefly discussed:
1. A book of memoirs by the Polish drama producer Ludwik Heronim Morstin is entitled Opowieści o ludziach i zdarzeniach. Yet for the English-speaking reader, the word-for-word equivalent, ‘Stories of people and events’, would suggest a book devoted to historical personalities far removed from the author, as opposed to a personal memoir.
2. Polish official and legal documents typically refer to both the originator and recipient of the communication in the third person. Thus in correspondence from ZUS, the Polish Social Insurance Institute, the ‘we’ entity is referred to as zakład, and the ‘you’ entity as ubezpieczony. The equivalent is often not possible in English.
3. Impersonal expressions such as warto and należy (cognate with French il faut and cela vaut la peine) can often not be directly translated in this form into English, since the resulting expression often sounds coldly clinical, even threatening.
4. Polish press material, and also other types of text, frequently contain statements that do not originate from the author or reporter writing the article, but rather from a third-person entity within the article. For example, in an article on discontent in the ranks of the Polish police, the statement funkcjonarjuszy od dawna są przeciążeni (apparently stating that the officials have been under stress for some considerable time) is clearly not the opinion of the writer, but the unacknowledged quotation of a sentiment expressed by the police themselves. This is not usually possible in English.
In each of the above cases, the translator working into English feels the need to define a sense of ownership of the utterance, or to bring the content closer to the writer and reader. The following proposes an analysis of why this is so, and what this might have to tell us about the requirements for post-editing NMT output into English.
The model proposed here is drawn from two sources:
1) the present writer’s idea that behind all utterances in general, and the nouns in a text in particular, there is an ‘enunciative meta-statement’ that is to say the ‘utterance of the utterance’, the utterance-as-action that results in the utterance-as-product; and
2) Włodzimierz Rybarkiewicz’s concept of subject prominence and topic prominence in Polish and English.
To begin with the meta-utterance concept: every utterance by definition has an utterer, who can be seen as the enunciating ‘subject’ of an enunciating ‘verb’. That which is uttered – the entire text, in the case of a written communication – then becomes the enunciative ‘object’.
Thus the text of this paper would be the object of an enunciative meta-statement such as ‘I say that ... ’.. And each noun in the text would be backed by a statement such as ‘I say there is a ...’.
The question is, could this meta-statement possibly be different in different languages?
This is where the argument of the Polish linguist Włodzimierz Rybarkiewicz comes into the picture. He states as follows:
“According to Lehmann’s (1970) proposal, Proto-Indo-European should be regarded as a T-prominent language. Comparing modern European languages (...) the conclusion is inevitable that a drift from consistent T-prominence to highly consistent S-prominence has taken place (...). It is being suggested in this paper that the Slavonic languages are less advanced and more conservative in the T- to S-prominence drift. This general conclusion is warranted (...) by comparison of Polish and English. English is far advanced in its drift from T- to S-prominence.
In fact, elsewhere Constance Lehmann goes further, suggesting that in Proto-Indo-European (PIE), the ‘subject’ entity was not a necessary part of the sentence at all.
Accordingly, in Polish as a topic-predominant language, the subject often appears as a logical (as opposed to grammatical) subject, in an oblique case: było nam zimno, as compared with English ‘we were cold’. The shift from earlier ‘methinks’ to ‘I think’ in modern English is another example of the same phenomenon.
Now, if this idea is applied at the level of the enunciative meta-statement discussed above, the conclusion might be as follows: in English the meta-statement would be subject-prominent: ‘I say that [content of the utterance]’, and for each noun, ‘I say that there is a [...]’, whereas in Polish the meta-statement would be topic-prominent: ‘it is said (by me, or someone) that [content of the utterance]’, or perhaps ‘it seems (to me, or someone) that [content of the text]’. And for nouns within the Polish text, the meta-statement might be ‘it is said (by me or someone) that there is a [noun]’ or ‘it seems (to me or someone) that there is a [noun]’.
English utterances would therefore be backed by a verb of utterance with a mandatory, categorically and clearly defined ‘I’, whereas Polish utterances would originate from a verb with an optional, oblique subject entity.
So now let us consider the four above-mentioned translation difficulties in the light of this proposed meta-statement framework.
1) Opowieści o ludziach i zdarzeniach: in terms of the model being proposed here, the problem with a word-for-word translation into English as ‘stories of people and events’ is that the clearly defined ‘I’ on the utterer side of the equation tends to imply ‘non-I’ content, i.e. clearly distinct from ‘I’, on the utterance side. The English-speaking reader might therefore get the impression of a work about, say, Alexander the Great or Winston Churchill, as opposed to the author’s own experience, ‘my’ people, and ‘my’ events.
Conversely, the less clearly defined ‘I’ uttering the Polish enunciation allows a less clearly ‘non-I’ enunciative object.
A preferable title in English might therefore be something along the lines of ‘My life and times in the Polish theatre’. This overcomes the ‘non-I’ implication in two ways: by explicitly stating the first-person pronoun, and with a cliche expression, “life and times”, creating a sense of community and shared linguistic experience between the parties to the communication.
2) Zakład / ubezpieczony: This pair of terms used in ZUS correspondence looks like ‘the institute’ and ‘the insured’, but those English terms, given the strong, categorical ‘I’ utterer behind an English text (with the resulting implication of ‘non-I’ reference), would suggest that ‘I’ am not the institute, and ‘you’ are not the insured. This is patently absurd – so zakład could well have to be translated as ‘we’, and ubezpieczony would definitely have to be translated as ‘you’.
In Polish, on the other hand, no such exclusion of reference to ‘me’ and ‘you’ would apply, because of the less clearly defined ‘I’ entity standing behind the text.
Conversely, a contract in English beginning the terms and definitions section by saying “‘We’ means XYZ Ltd, and ‘you’ means the user of the service’”, when professionally translated into Polish, is likely to read (back-translated into English), as “‘the company’ means XYX Ltd, and ‘the user’ means the user of the service”.
3) Warto and similar expressions: in this case, if the direct equivalent English phrase ‘it is useful’ is used, the categorical ‘I’ behind the statement on the utterer side implies some sort of universal knowledge of an objective truth. This creates a cold, clinical, impersonal impression, as if the ‘I’ backing the statement can somehow absolutely know this to be the case, and lay it down without question.
This rather daunting situation can be rectified in the English translation by saying something like ‘at this point it may be useful to ...’, which diminishes the force of the statement in terms of certainty (“may”) and scope (“at this point”). No real meaning is added, but an appropriate level of consensus between the parties to the communication is restored.
4) Funkcjonariusze od dawna są przeciążeni: the enunciative meta-statement idea is also clearly applicable to this ownership of utterances problem. ‘The officials have been under stress for smoe considerable time’ would be unsatisfactory as a translation here precisely because that English statement could only be uttered by an ‘I’ entity, whereas the Polish was clearly rather based on a meta-statement such as ‘it is said (by them) that ... ‘. The solution is to restore the rightful ownership of the utterance: ‘The police say they have been under stress for some considerable time.
There is nothing new about this sort of solution to these and similar translation problems, but my hope is that the meta-statement model will help to clarify the nature of the problem, and validate its solution in this way.
To illustrate the potential benefits from this approach, the following short case study is offered, in the form of a ‘think aloud protocol’, following the present writer’s encounter with an impressive monument in a park in Szczecin, called the Pomnik czynu Polaków. The English translations provided were “Monument of the Deed of Poles”, in the tourist guide material available, and “Monument to Polish Endeavour”, on the monument itself.
The writer’s reconstructed TAP ran something like this: “Well, you certainly can’t say ‘the deed of Poles’, since that would suggest one deed in the past (as opposed to the intended reference to multiple deeds in the past, present and future), and would carry the implication ‘I (the creator of the monument) am not a Pole’, which is ridiculous. Using the adjective ‘Polish’ is much better, since it is not subject to this ‘I vs. the Poles’ opposition. So that will be part of the solution. Now, what about ‘deed’? ‘Endeavour’ sounds a bit laboured for what we need here. So what are the generic connotations of the Polish word czyn? To my ear it sounds lofty, noble, heroic, rather than anything routine or run-of-the-mill. Aha, that’s it – I will use ‘Monument to Polish heroism’, i.e. the abstract noun from the adjective implied by czyn.”
Rather than providing any fixed rule for translation from Polish to English, the suggestion is that with practice, the meta-statement model and its implications can become second nature for the translator, instinctively prompting a preferred translation in the manner illustrated above. The model could then be applied more consciously at the stage of validating the translation after the fact.
The meta-statement model could also cast light on some other situations regularly confronted by translators working into English from other European languages. The first of these is the translation of człowiek, German Mensch and other cognates.
For example, a Polish monograph on Kierkegaard has as part of its title the phrase Dyskurs o człowieku, glossed in English in the table of contents as “Discourse on the human being”. In the English speaker’s mind this conjures up unwanted associations of a group of aliens engaged in a discussion of our species, again because of the ‘I’ utterer versus ‘non-I’ utterance dichotomy. Specifically, the meta-narrative in English is something like ‘there is a human being’, with the additional implication of ‘I am not a human being’. The rule of thumb in such cases is to make the noun phrase more familiar, shared or generic in some way, such as ‘Discourse on the nature of man’. (In the same way, from the English speaker’s perspective, the French question quel temps fait-il? clearly is not asking ‘what weather’ we are having, but ‘what sort of weather’, ‘what is the weather like’.)
English translators also have great difficulty with Nietzsche’s letzter Mensch in Also sprach Zarathustra. They tend to oscillate between “the last man” and “the ultimate man” – but the latter is an artificial and virtually meaningless phrase, and there is a twofold problem with the former: 1) it tends to imply that the speaker is not a man (human being); and 2) the reference is to a flesh-and-blood individual, hiding behind this or that bush.
This again is quite consistent with the meta-statement model – a more definite ‘I’ tends to stand opposite a more concrete ‘it’. In this case, it could again be fruitful to took towards a cliche expression such as ‘the lowest of the low’ or ‘the dregs of humanity’.
Another translation issue potentially explained by this model is the fact that words that appear to mean ‘must’ in Continental European languages, such as French doit/doivent, are frequently better translated into English as ‘should’. English ‘must’ can have unwanted connotations of coercive, physically asserted authority, and accordingly the expression of obligation in English is often downgraded to ‘should’. This would again be because of the more clearly defined ‘I’ entity standing behind the utterance in English.
Conversely, ‘should’ is best upgraded to, say, doit when translating in the other direction, from English into Continental European languages.
This Jamieson-Rybarkiewicz model might also provide a neat explanation of some key difficulties faced by learners of English as a foreign language – in particular the use of the definite article and distinctions between verbal tenses such as ‘I do’ vs. ‘I am doing’, ‘I did’ vs. ‘I have done’.
Both of these somewhat idiosyncratic aspects of English can be seen as arising from the need of the utterer to situate him/herself with regard to the uttered content.
But the less categorical ‘I’ utterer in, say, Polish, would obviate the need for all this care and attention.
The upshot of all this is a curious enunciative paradox. In English, the direct, confronting nature of the implicit meta-statement underpinning the content requires a degree of softening and personalisation of the explicit textual content. In Polish (and numerous other European languages), the less assertive underlying meta-statement allows more assertive textual content, so that less of the above kind of manipulation is required in the explicit content.
So as readers, we experience not just the text, but the presence of the author encoded within it. In an English text in particular, the explicit content defines a constantly changing, clearly encoded authorial presence implicit within the text. Conversely, in a text in Polish, the authorial stance is less clearly encoded, so the changes in the authorial stance can be left for us as the reader to deduce for ourselves.
Now, all this has some important implications for the post-editing of NMT output into English, since the computer programme reads the words on the page, but lacks the ability to adjust the author-text, or author-text-reader, relationship in the English output text. The post-editor will probably have to carry out precisely the sorts of manipulation described in this paper, that is to say in most cases to make the content more generic, familiar and shared between writer and reader. Since the mechanical errors of syntax in MT output have now largely been ironed out, it becomes a matter of creating an appropriate narrative perspective and creating the desired impression in the reader’s mind.
Strange as it may seem, therefore, the formerly clinical task of post-editing is now coming rather close to a transcreation process!
But if we can identify the different authorial meta-statements that underpin texts in different languages, and recognise this as the missing component in NMT output, this could provide a valuable basis for the training of post-editors, particularly those working into English, and possibly those editing output into other languages as well. Indeed, it may even be possible to apply the meta-statement concept in the ongoing quest for better NMT output in the first place.