THE THEORY OF TRANSLATION
3. Translation theory and psycholinguistics
To draw conclusions about the process of translation not solely based on a com
parison of the source language and target language texts but also taking into account the translator and the processes in his/her mind, i.e. in the "black box", psychology and psycholinguistics also need to be involved in translation theory.
3.1. Perception and production in translation
Psycholinguistics, studying the psychological processes accompanying speech activity, entered a period of rapid growth in the 1960s at the same time as socio
linguistics (Miller 1965, Adams 1972, Greene 1972). Since translation is also a speech activity, but with the difference that it is conducted in two languages, every
thing that psycholinguistics has stated about speech perception, memory opera
tion, and speech production can be used to get closer to understanding the bilin
gual speech activity of translators. Since speech perception can be interpreted in both a narrow and a broad sense, it is important to note that here perception will be used as a cover term which involves not only speech perception but also speech comprehension. In addition, it has to be stated that perception is used to refer to the comprehension of both written and spoken texts, and production refers to the creation of both written and spoken texts. Let us now review the differences between monolingual speakers and translators in terms of perception and production.
(1) In the case of the monolingual speaker, perception and production take place in the same language, whereas in the case of the translator, perception and production do not happen in the same language, and there is a transcoding stage between them.
(2) In the case of the monolingual speaker, perception and production follow each other; in the case of the translators or interpreters, perception and produc
tion, despite the transcoding stage, can overlap and can happen simultaneously or with a negligible time difference, as in simultaneous interpreting, or with delay, as in consecutive interpretation, and in the latter case the ability to store informa
tion, i.e. memory gains a considerable role.
(3) The translator differs from the monolingual speaker not only in the fact that he/she has to speak two languages, but also in that he/she has to follow two completely different strategies in giving a linguistic structure to the ideas. Trans
lators move from a given linguistic structure to the underlying thought and from the thought to another linguistic structure following two different strategies, and in many cases, for instance in simultaneous interpretation, they do so almost exact
ly at the same time. This explains why not everybody can translate or interpret well who speaks a foreign language.
(4) The road from the thought to the linguistic form is made even more diffi
cult for the translator by the fact that he/she has to express someone else’s ideas.
In A. A. Leontev’s terminology, translators work on the basis of an “externally defined programme” (Leontev 1969 in Shveitser 1988:28).
(5) As a result of the above mentioned causes, translators are also motivated differently. They do not satisfy their own communicative needs but somebody else’s, while they themselves are also people, and thus cannot disregard their own interests in the process of perception and their own opinions during the process of production. Compared to monolingual speech activity (direct motivation), trans
lation is characterised by indirect motivation (Zimnyaya in Zlateva 1993).
3.2. Translation and bilingualism
In contrasting translators and interpreters with monolingual speakers, we deliber
ately do not regard them as bilingual speakers. Researchers of bilingualism differ greatly with regard to who can be called a bilingual: only the ones who speak both languages as their mother tongues (Bloomfield 1935), or those who have different competence in both languages (dominance of one language [Haugen 1953]).
Since our approach is closer to the Bloomfieldian one, translators and inter
preters are not referred to as bilingual speakers here. The most dominant feature of translators and interpreters is not that they have native speaker proficiency in two languages, but that they are professional linguistic mediators, i.e. can medi
ate between two languages. Many excellent translators cannot communicate proficiently in a foreign language, but can brilliantly perform the task of linguistic mediation between a foreign language and their mother tongue. The notion of bilingual speaker is a partly narrower and a partly broader concept than that of the linguistic mediator. It is narrower, since not all bilinguals can do translation or interpretation, and it is broader, since not all translators or interpreters can be called bilingual, either. Translators and interpreters might be claimed to represent a specifically limited functional version of bilingualism.
3.3. Simultaneous interpreting as a psycholinguistic experiment
Translation theory can apply the results of psycholinguistics research in monolin
gual speech activity, that is, it can adapt the findings of psycholinguistics to the analysis of the process of translation. Interestingly, this has been acknowledged by the researchers of oral translation (interpretation) earlier than by the researchers of written translation. The analysis of recorded speech production of interpreters reveals more about the mental processes taking place during the translation than the analysis of written translations.
Audio recordings of the discourse produced by simultaneous interpreters are more appropriate for the investigation of the mental activity of translators because the target language performance of simultaneous interpreters is not so far from inter
nal speech as the corrected, proof-read, post-edited written texts of translators.
3. Translation theory and psycholinguistics
Simultaneous interpreters, working under time pressure, urged by the time con
straint, and lacking better solutions, often say things into the microphone that they would never come up with in written translation. This often rough “semi- transcoded” discourse, containing seemingly unjustified insertions and omissions, vague chunks alternating with well-formed ones, the seemingly unjustified shifts between rapid speech, slow speech and pauses, reveals abundant information about the characteristic features of speech activity conducted in two languages. The activ
ity of simultaneous interpreters may be considered as a large-scale spontaneous psycholinguistic experiment, in which the recording of the material can also be done easily. It provides tremendous possibilities for research, which have not yet been exploited.
3.4. The simultaneity of listening and speaking
Researchers of simultaneous interpreting, naturally, do not only use the data pro
vided by such spontaneous experiments, but they also conduct planned experiments.
The first experiments were aimed at finding out whether or not it is possible to listen and speak simultaneously. Some researchers claimed that simultaneous interpreters do not listen and speak at the same time, but they alternate between the two activities, using the pauses of the speaker as well as their own pauses, i.e.
the “micro-pauses” as Shiryaev (1973) called them.
3.5. The active nature of perception
This question has been investigated by psychologists and psycholinguists inde
pendently of interpretation. Interestingly enough, the rejection of the idea of simul
taneous listening and speaking is based on a correct observation. More precisely, it started out from the assumption that listening is never a passive activity, as in order to understand what has been said one has to perform a number of opera
tions.
Miller lists the following operations: (1) hearing the utterance (2) matching it as a phonemic pattern, (3) accepting it as a sentence i.e. grammatical acceptance, (4) attributing meaning to grammatical structures, i.e. semantic interpretation, (5) understanding, i.e. attaching contextual information to what has been said (Miller
1965: 295-295).
This series of operations, although it looks like “decomposition” at first sight, is also composition. Since it is unlikely that every sentence has a separate internal representation in the mind of the listener, it has to be assumed that during listen
ing listeners actively produce the internal representation of what is being said.
This idea was formulated by Halle and Stevens in 1962, but the active nature of listening had been pointed out much earlier than that.
Blonskiy wrote the following in 1935: “The reason why one cannot speak about something else while listening carefully to somebody is that while we listen to some
one’s speech we simultaneously reproduce it. Listening to speech is not simply just listening, but to a certain extent we also speak together with the speaker” (1935:
155).
As mentioned before, it was the correct observation that speech perception is not a passive process that led to the misbelief that speech perception and speech production are not simultaneous processes.
However, it is important to note that we are still not speaking about interpreta
tion; we are still investigating the relationship between perception and production within monolingual speech activity. This is precisely why psycholinguistics is impor
tant from the point of view of translation and interpretation theory: it had com
pleted a large number of experiments well in advance, from which research on simultaneous interpretation could start out. Before the practice of simultaneous interpretation could have proved that it is possible to listen and speak at the same time psychologists had conducted several experiments to justify the legitimacy of this claim (Yermolovich 1978).
3.6. Experiments independent of interpretation
In 1952, Broadbent conducted an experiment to see whether participants can answer a question while listening to the next one in the meantime. Poulton con
tinued Broadbent’s experiments in 1955, overcoming the problem in previous experiments that the voice of the experimenter and that of the respondent were disturbing each other. In these experiments the questions were raised via head
phones, and he also tried to make sure that respondents are not disturbed in under
standing the questions by giving too loud answers. The results of Poulton’s experi
ment confirmed those obtained by Broadbent, according to which it is possible to listen and answer at the same time. By improving the conditions of simultaneous asking and answering (headphones, volume control) he also managed to reduce the rate of error.
Sokolov investigated the relationship between perception and production in the following experiment: in the first phase of the experiment, during listening, participants read out a poem, in the second phase they counted, and in the third phase they recited a poem by heart. The experiment allowed for three important conclusions. The first is that the more automatic production is (e.g., reciting a poem by heart), the more successful perception will be. The second is that perception takes place by capturing certain conceptual nodes. And finally, the third is that familiarity with topic considerably affects the efficiency of perception (Sokolov
1968).
The role of topic familiarity in understanding was confirmed by an interesting experiment conducted much earlier by David Bruce. He recorded a set of ordi
nary sentences and played them, in the presence of noise so intense that the voice was just audible, but not intelligible. He played it several times for the same audi
ence. Before each playing he defined the topic of the text differently, and listeners always picked out sentences from the text which matched the given topic content- wise (Bruce 1956).
The active nature of listening and understanding does not therefore consist merely of constructing the sentence while listening, but it also indicates that some preliminary ideas are formed about what is going to be heard already before
start-3. Translation theory and psycholinguistics
ing to listen on the basis of the topic specified or the communicative situation.
Thus we listen “accordingly.”
Summarising the findings of his experiment, Poulton also referred to the fact that his participants worked on the basis of particular hypotheses. Miller draws the same conclusion when evaluating Bruce’s experiment: “With an advance hypoth
esis about what the message will be we can tune our perceptual system to favour certain interpretations and reject others” (Miller 1965: 297).
3.7. Probability prediction
The sporadic references to hypothesis-forming and preliminary adjustment in psychology and psycholinguistics were applied with a great deal of originality by two Russian authors, Chernov and Zimnyaya, creating the theory of probability prediction. They offered an apt answer to the question whether production may happen simultaneously with perception or not, or whether it can only follow it:
in simultaneous interpreting production neither follows perception nor appears simultaneously with it; in fact it precedes it (Chernov and Zimnyaya 1970, 1973, Chernov 1978, 1994).
The gist of the theory is this: Chernov and Zimnyaya do not share the view that interpreters work entirely on the basis of an externally motivated programme.
In their opinion, translators, taking into consideration all of the linguistic and extra- linguistic aspects of the situation of interpretation, form an internal programme, which consists of two types of hypotheses. On the one hand, knowing the broad context of the event (time, place, speakers, topic, aims, text-types, etc.) the inter
preter has a general view of the contents of the text to be interpreted (long-range hypothesis). On the other hand, based on the actual sentence started by the speak
er, he/she has a specific idea about the contents of the sentence being uttered (short-range hypothesis).
These two hypotheses make it possible for the interpreter to stop the process of listening actively to the speaker and from time to time work on the basis of a hypothesis. This hypothesis enables the interpreter to predict the structure and contents of the sentence, which is especially important in interpreting from lan
guages such as German, in which the predicate is uttered at the end of the sen
tence. Periodically the interpreter switches back to active listening to check the correctness of his/her hypothesis. If it is confirmed, the interpreter continues to work on the basis of the hypothesis. If it happens to be incorrect, he/she stops the process of interpretation for a while to actively listen to the speaker until a correct hypothesis is reached again. After this he/she continues the work following this hypothesis, i.e. the separate internal programme, otherwise he/she would not be able to catch up.
Zimnyaya, Chernov and their colleagues justified their theory partly by conduct
ing guided experiments and partly by analysing audio-recordings of conferences.
Guided experiments produced data about the operation of hypotheses working at various levels.
3.7.1. Types of hypotheses
Sladkovskaya, for example, attempted to experimentally justify the operation of hypotheses formed on the basis of the aim of utterance. She had an eleven-sen
tence greeting speech interpreted by practicing interpreters. The interpreters were only informed about the fact that a UN representative was to greet the new presi
dent of the general assembly entering into office. They were not told that the greet
ings would be delivered by the representative of an Arab country and that the new president was the representative of a country on friendly terms with Israel. Out of the eleven sentences of the greeting, the fourth, fifth, seventh and ninth sentences carried hidden reproaches because of the new chair/president’s earlier pro-Israeli policy.
These hints were transferred by only one interpreter, who was experienced enough in UN interpretation to recognise the speaker. The others either left out the hidden reproaches or altered the text. This implies that after the first three sentences they decided that it was considered an ordinary greeting speech (let us not forget that the first reproach came in the fourth sentence of the greeting), and subsequently worked based on their own hypothesis. Anything that contradicted this hypothesis they either did not hear or did not want to hear (Sladkovskaya 1971).
Danica Seleskovitch relates a similar case from her own experience. She was interpreting for a speaker who had an excellent command of English and Anglo- Saxon speech styles at a conference without knowing the speaker’s nationality.
During the process of interpretation she noticed with surprise that the speaker’s position did not harmonise with the official English position, what is more, it sharply contradicted it. Using all her skills, she tried to evade the embarrassing places in the speech and blunt the sharpness of the anti-England phrases. Finally, it turned out that the speaker represented Denmark, but it was already too late (Seleskovitch 1968).
These examples neatly show what would be the ideal procedure of research within translation and interpreting studies. A problem is identified in the every
day practice of translation or interpreting (Seleskovitch), motivating a search for a theoretical explanation (Chernov and Zimnyaya), which is ultimately justified or rejected by guided experiments and not examples taken from practice (Slad
kovskaya).
Seleskovitch writes down this event in 1968, which might merely seem as some
one’s own mistake, but in the light of the theory proposed by Zimnyaya and Cher
nov (1970), it may be considered as evidence for the assumption that interpreters work based on some hypothesis. Seleskovitch formed the hypothesis on the basis of the performance of the speaker. Sladkovskaya’s participants in the experiment worked on the basis of the misperceived aim of the utterance (1971).
In a series of experiments, Belyayevskaya and her colleagues (1973) tried to provide evidence for the simultaneous operation of several hypotheses at the same time. They conducted experiments to demonstrate the operation of hypotheses on the level of morphemes, grammatical structure, lexical structure, and meaning.
They also investigated the nature of the so called opornije punkti (‘basic points’ or
‘sign-post elements’). Following Zimnyaya’s and Chernov’s theory, the essence of basic points is that these are the pieces of information the interpreter has to active
3.Translation theory and psycholinguistics
ly listen to, to confirm or reject the hypothesis. Belyayevskaya and her colleagues have shown that these basic points, that is, the parts of the text with the highest information value and lowest redundancy rate, are not necessarily complete words.
In many cases, it is enough for the interpreter to hear only part of the word or merely its first sound to form the correct hypothesis.
3.7.2. The nature of hypothesis generation
Despite the large number of experimental evidence, some practicing interpreters rejected the theory of interpretation based on probability prediction, saying that the interpreter has no right to translate anything that has not been uttered yet.
Others criticise the content and structure based nature of the hypothesis claiming that simultaneous interpreters work instinctively and not consciously;
after finishing interpreting they do not even know what exactly they said and many interpreters think about completely different things while they work. In their view, interpreters generate hypotheses on the basis of the speaker’s intonation, and it is more correct to refer to it as “acoustic adjustment”.
Steier (1975) claims that interpreters cannot form content based hypotheses because in the case of simultaneous interpretation they only use their short-term memory, which stores words on an intuitive-associative basis, and they forget these as they move forward in the process of interpretation. They use long-term memory only in those cases when the normal process of simultaneous interpret
ing is interrupted for some reason, for instance if the information is produced either too slowly or too fast.
Although no definitive evidence has been found to decide this debate, it has been shown that recordings of discourse produced by simultaneous interpreters contain extremely valuable data for research aimed at looking into the minds of
Although no definitive evidence has been found to decide this debate, it has been shown that recordings of discourse produced by simultaneous interpreters contain extremely valuable data for research aimed at looking into the minds of