Shadowing – Guichot de Fortis

In this brief text I shall endeavour both to describe the technique and provide some hints as to its use.

The technique and practice of shadowing is an indispensable tool for both the budding and the experienced simultaneous interpreter, but it is a controversial technique and is often misunderstood or discounted. In my opinion, however, all interpreting professionals would gain greatly from spending time both considering and practising the art of shadowing.

Chris Guichot de Fortis Senior Interpreter, NATO Interpreter Training Resources

Shadowing is useful into all the interpreter’s active languages, ‘A’ and ‘B’, and can be employed to correct and refine a multitude of interpretation weaknesses – accent, delivery, voice quality, vocal range, emphasis, ‘cleanliness’ of rendition, confidence etc. etc.

However, it is important that shadowing:

  • –  be carried out in a graduated, thorough and reasoned way
  • –  be regularly supervised and/or assessed by both the practitioner and his or her teachers, supervisors or colleagues
  • –  be carried out over many hours and in each of the linguistic combinations that it is desired to enhance
  • –  be coupled with more conventional training techniques

The technique consists of spending many hours in a real or virtual booth shadowing an able and fluent speaker of the target language. As the goal is to replicate the neurological and intellectual demands of simultaneous interpretation, a simple laptop/ipod/headphone combination will suffice, in the absence of a true booth. Using MP3/MP4 or flash files, DVDs, CDs or audio cassettes, choose speakers who are expressing themselves in their mother tongue and who have an excellent mastery thereof, without strong regional accents, and with a gift of oratory which allows full expression of the native cadences of the language. It cannot be over- emphasised that your chosen speaker must be carefully selected, as a function of accent, elocution, delivery, register etc. This is an excellent technique at many levels, as (this being a marked trend among recent neuro-linguistic and neurological expert studies) shadowing involves some 80% of the neuro-linguistic operations involved in simultaneous interpretation, the only factor missing being that of language transfer. Shadowing initially involves repeating the words of the speaker without modification. This allows the interpreter’s brain, ears and mouth, working as they do in concert, to begin to reproduce the sounds and rhythms of the target language, without conscious mental effort, and begins to create the ‘linguistic muscle memory’ naturally acquired by children learning their own tongue. This will require many tens of hours of actual speech production – it is essential that the language actually be voiced, or the exercise is useless.

It is also recommended, in the case of an actual or potential ‘B’ language, to shadow with a text, as it is true to say that we cannot hear or apprehend what we do not know, and if we do not hear all the articles, prepositions, and smaller sounds that make a native speaker sound native, we will not reproduce those sounds in our shadowing, and will lose much of the potential benefit. Here again, it is useful to record your shadowing, and then replay it, comparing it to the text.

The prime goal of the exercise is to accustom brain, ears and mouth to the flawless and (eventually) effortless production of the sounds and cadences of what may be (in the case of a ‘B’) a foreign language. The goal here is to establish a new network of synapses and neuronal pathways, this being an essential stage in the interpreter’s acquisition of each new language combination. It should not be thought that all lessons learned in the successful mastery of one combination can simply and instantaneously be transposed to another – many hours of actual practice are required for each language pair, and there are no shortcuts!

Let’s now begin to look in a more concrete way at the actual practice of the technique.

While shadowing, it is important to experiment with differing levels of time lag or ‘recul’ (say from 0.5 to 5 seconds), introducing a certain elasticity to reflect the fluctuating demands imposed by the speaker and to train the brain to cope with larger or smaller linguistic buffer spaces in the language combination being employed.

At the same time, gradually introduce expressions of your own, allowing for varying semantic (but of course not substantive) distance from the speaker. At one extreme you may wish to decide in advance to modify one or two words per sentence, and at the other to leave only one or two words unchanged.

In order to approach, in the ‘B’ language, the facility which characterizes an experienced interpreter’s work into his/her mother tongue, it is also important to train both voice and brain to ensure acceptable linguistic production while mental processing efforts are required elsewhere.
To this end, it is useful while shadowing to practice (for example) writing numerical sequences involving fixed gradations (1, 3, 5, 7… or 1, 6, 11, 16, 21 etc.), which can then be self-checked after the exercise, along with the recorded interpretation.
Another variant might involve writing down poems or song lyrics, which the interpreter knows by heart, while interpreting. Using increasingly complex sequences is doubly fruitful, and the goal, evidently, is to guarantee an acceptable level of linguistic production even while mental processing efforts are devoted to other, more noble, tasks such as actually understanding and transposing concepts and ideas! Such exercises are useless, of course, unless both spoken and written productions are assessed for accuracy and acceptability.

Many interpreters experience difficulties, in the booth, in adopting a register or ‘persona’ which differs from their own, and shadowing can be very helpful in acquiring these more thespian-related skills which can so often make the difference between a good and an excellent interpretation. Thus, shadowing speakers who are expressing joy, grief, anger, sorrow or enthusiasm, will begin to instill the required ‘muscle memory’ that will allow the interpreter (when the chips are down and lack of the appropriate vocabulary or register would severely damage the credibility of the interpretation) appropriately and confidently to transmit the entire message and sentiments of the speaker. To this end, it is useful to shadow speakers who are expressing strong or even excessive emotion, without fear of drifting into caricature, given that there will always be a filter or some loss of intensity between ‘shadower’ and ‘shadowee’.

The above exercise is of particular utility in the interpreter’s ‘B’ language, as its extended practice helps to instill native accent and provide a more nearly instinctive feeling for register and vocabulary, in sensitive contexts where any such failures would have serious consequences. For accent correction purposes, it is preferable initially to shadow language-learning tapes/CDs, etc., because the texts are spoken slowly, thus all sounds can be easily discerned. In addition, the texts employed are simpler, but grammar and syntax are correct. An added advantage is that the text will be available to read during shadowing.

It is also useful to spend time shadowing fast speakers, as it is true to say that many (usually inexperienced) interpreters have difficulty in simply delivering even their native language rapidly, clearly and without stumbling, especially when obliged to adopt a cadence which is not their own. It goes without saying that this difficulty is exacerbated into the ‘B’ language.

It is my hope that the above hints and descriptions will help you in your interpreting life, and endow you with increased facility and confidence in all your active languages, and in all registers. I should again stress the importance of shadowing, and of spending considerable amounts of time on this exercise, to enable the brain to integrate it in a reflexive, automatic way, clearing the way for more complex intellectual operations while actually interpreting.

Christopher Guichot de Fortis (AIIC)
Senior Staff Interpreter, NATO Headquarters, Bruxelles September 2011

Perfectionnement linguistique (ESIT)

Quelques conseils librement inspirés de la brochure de l’ESIT : Comment perfectionner ses connaissances linguistiques (ESIT 1984-1995-1998) Compiled and contributed by Jean-Jacques Pedussaud.


Langue A 


1. lire des textes littéraires (roman, poésie, théâtre) 

2. lire la presse (dont obligatoirement, chaque jour, UN article sur un sujet qui a priori vous rebute !) 

3. rester critique vis-a-vis de la langue employée dans les médias ; etre un lecteur vigilant, a l’affut de l’inattendu, mais aussi des erreurs ; de meme, etre un locuteur et auditeur vigilant : faire la chasse a ses propres erreurs ou a ses tics de langage (nous en avons tous…). 


4. écrire pour soi-meme (journal intime) et a autrui (courrier, méls). 

5. pratiquer des exercices langagiers écrits (exercices de style a la Raymond Queneau ; imitation d’un modele) 


6. dresser des comptes rendus a l’oral de choses entendues ou lues (se faire corriger si possible par un autre locuteur natif : a défaut, s’enregistrer et se réécouter d’une oreille critique, en recommençant au besoin). 

7. pratiquer des exercices langagiers oraux (exercices de style, exercices d’imitation : cf. activité écrite 5 ; memes conseils que pour activité 6). 


Par définition, la COMPRÉHENSION ORALE ne doit pas poser de probleme en langue A ! 

Langue B (français) 


8. pratiquer l’écoute attentive de la radio, de la télévision, de films en VO, d’enregistrements audio ou vidéo, de discours et d’interviews sur internet. 


9. lire attentivement textes littéraires et presse (cf. activités 20 & 21), rechercher (internet, dictionnaires unilingues, encyclopédies…) ET NOTER dans un carnet le sens des expressions inconnues. 


10. discussion sur un sujet donné avec un interlocuteur natif qui vous écoutera en prenant des notes pour vous corriger a la fin : il relevera les erreurs, mais SURTOUT proposera des formulations correctes, avec beaucoup de synonymes => donc trouver un locuteur A fiable. 

11. SHADOWING de la radio ou de la télévision (cad répétition du message a l’identique, avec une ou deux secondes de décalage). Attention : cet exercice ne prépare pas a l’interprétation simultanée. Il vise seulement a améliorer le débit et la fluidité ainsi que la fidélité phonologique et intonative. 

12. Compte-rendu oral (d’un texte ou d’un message oral) avec REPRISE A L’IDENTIQUE d’expressions issues de l’original (peut tout a fait s’associer a l’exercice 8 ou 9). 

13. PARAPHRASER oralement un message initial écrit ou oral (peut s’associer aux activités 8 ou 9), cad reformuler en s’obligeant cette fois-ci a NE PAS utiliser les memes expressions que l’original. NB : c’est une excellente préparation a l’interprétation (méthode). Attention toutefois a ne pas mélanger perfectionnement linguistique et interprétation…

14. Lecture a haute voix (on peut s’enregistrer, se réécouter, et recommencer en corrigeant les erreurs phonologiques que l’on aura relevées. Attention toutefois au perfectionnisme mortifere !). 

15. Apprendre par coeur un paragraphe argumentatif a la fois (tiré d’un essai ou article de presse), voire un poeme, un passage de roman ou de piece de théâtre. Le réciter ou le noter par écrit de mémoire. Vérifier la fidélité. Recommencer jusqu’a mémorisation parfaite, avant d’apprendre un nouveau passage. 


16. Compte-rendu écrit avec REPRISE d’expressions issues de l’original (cf. activité 11) 

17. Compte-rendu écrit avec PARAPHRASE SYSTÉMATIQUE (cad expressions systématiquement différentes de celles employées dans l’original — cf. activité 12) 

18. Écrire sur un sujet imposé (un paragraphe sur tel ou tel theme). 

19. Écriture personnelle (journal intime) ou courrier. 


20. Noter tous les faits de langue qui vous frappent au fil de la journée (lectures, conversations, informations a la radio ou a la télévision…), notamment les expressions que vous n’emploieriez pas spontanément, dans un petit carnet que vous aurez sur vous EN PERMANENCE (cf. activité 9). Les réemployer ensuite dans la mesure du possible dans les activités d’expression (10 a 19). 

21. Chaque jour, lire attentivement UN article et relever tous les mots et expressions que vous n’auriez pas utilisés spontanément (y consacrer environ 1/4 d’heure par jour). Les réemployer ensuite dans les activités d’expression écrite ou orale (10 a 19).

Language acquisition (Daniel Gile)

This extract is taken from pages 213-225 of Daniel Gile‘s excellent “Basic Concepts and Models for Interpreter and Translator training”, 1995, and is reproduced here without the kind permission of Benjamins of Amsterdam.

Basic Concepts and Models for Interpreter and Translator training 1995. Benjamins

3. The Gravitational Model of linguistic availability 

The exercises described above deal mainly with the active component of language mastery. Moreover, because of time constraints, they only involve a small part of the relevant vocabulary. When students are told to spend some time in a foreign country or to read for linguistic enhancement, they are generally not gi’ven detailed instructions on how to proceed. They are therefore often inefficient in their endeavors because their progress is left to chance. The Gravitational Model provides a conceptual framework for comprehension of phenomena and assessment of learning strategies. 

3.1 The baseline 

Student interpreters who are admitted into interpretation schools do have a “good knowledge” of all their working languages, although this knowledge is more often than not inferior to theoretical requirements as defined in section 1 of this chapter. Their initial level of language proficiency’ includes good command of basic rules of grammar, which are few in number (not more than a few hundred, depending on the language), although students may make occasional mistakes, especially due to interference from other working lan­guages and to high processing capacity requirements during interpretation. Lexical command, however, is highly variable. This is not surprising in view of the fact that an educated person’s vocabulary consists of several tens of thousands to more than a hundred thousand different words (see Aitchison 1987: 5-8). A small fraction of this vocabulary is encountered daily and a larger proportion rarely, depending on the individual’s living environment and professional and personal activities (see section 3.7). 

3.1.1 The lexical component 

In international conferences, speakers select their own words and language structures on the basis of choices determined by the grammar of the language concerned. Because of the relatively small number of rules involved, at least in the types of discourse heard in international conferences, speakers tend to use the same structures, except for some rare poetic or literary speeches. These structures are also familiar to interpreters. Lexical choices are much wider, however, and word preferences tend to vary. Hence significant lexical prob­lems, in terms of both comprehension and retrieval from long-term memory for speech production. It is important to note that in view of the extremely large and ever-increasing number of technical terms used in conferences, the lexical issue (in its terminological subset) remains important throughout a conference interpreter’s career.

More specifically, two types of lexical comprehension problems can arise in interpretation: 

Incoming source-language words are unknown to the interpreter, who therefore needs to perform a contextual and phonological and/or morpho­logical and/or etymological analysis in order to understand them. The analysis may result in full, partial, or no understanding. 

Incoming source-language words are known to the, interpreter, but are not familiar enough to be understood immediately and spontaneously, that is, fast enough and without conscious effort (see chapter 4). and their mean­ing may be misinterpreted. 

Similarly, two types of lexical production problems can occur in interpreta­tion: 

The target-language word required to express a concept is unknown to the interpreter, who therefore has to resort to another term or to a paraphrase. 

The required word is known to the interpreter, but is not available enough, that is, it does not surface fast enough or easily enough.

3.1.2 The syntactic component 

On the syntactic side, similar problems arise:

In comprehension, there should theoretically be no syntactic rules un­known to the interpreter in a working language. However, the speed of comprehension may vary, depending on knowledge of the language’s transitional probabilities (the probability that a certain type of word or group of words will be followed by another given type of word or group of words). 

In production, interpreters also know all the syntactic rules necessary to express themselves. but there may be significant differences in the avail­abilin, of these: it may take the interpreter more or less time and effort to make the appropriate syntactic decisions to start, continue., or finish sentences. 

3.2 The interpreter’s position as a speech producer and listener

In both comprehension and production, these problems are associated with an increase in protessing capacity and time requirements, which, as shown in the Effort Models M chapter 7, may lead to serious impairment of interpretation quality; hence t,~e importance of availability. This point is critically important, because a number of factors make speech production and speech comprehen­sion more difficult under conference interpretation conditions than in everyday life. 

3.2.1 Listeners and interpreters As explained in chapter 7, interpreters are generally in a position less favorable to comprehension than are listeners in most usual situations. 

In particular, during interpretation, attention-sharing reduces the capacity available for speech comprehension, and interference between between source language and target language also makes comprehension more difficult. 

The interpreter’s effort is therefore more intense than the delegates’ while it lasts; it is also longer lasting, since interpreters cannot rest as long as they are interpreting, while delegates may spend much of their listening time at a low level of attention. 

3.2.2 Speech producers and interpreters In simultaneous interpreting, the interference issue is more problematic in production than in comprehension. It requires an additional prevention and control effort (see inter alia Dejean Le Feal 1978 and Lederer 1981 a), which is another difficulty faced by interpreters and not by ordinary speakers. 

One point made earlier about interpreters’ position as listeners also ap­plies to their position as speakers: speakers in conferences talk about subjects they are familiar with, use terms that are part of their daily professional life, and often give a presentation they have prepared with ideas they have thought about carefully for hours, days, or even months or years; the interpreter knows less about the situation and the subject, and is also less familiar with the specialized terms and the speaker’s ideas. 

On the other hand, interpreters are professional speakers, and public speaking is part of their professional everyday life. Many conference partici­pants only rarely speak in public and are not used to the exercise. Moreover, interpreters use by definition working languages they know well, which is not always the case of speakers. Except for some stage fright, which interpreters learn to overcome with experience, they are generally not subjected to the same pressures as many speakers, who may have much at stake when they take the floor. There are cases where stage fright is strong for interpreters as well­important summit conferences, live interpretation on television, consecutive interpretation before a large audience-but these are generally not regular working conditions; when they are, interpreters develop more tolerance for the stress involved. 

Last but not least, technically sp eaking interpreters do not have to per­form the same speech-planning operations as speaker, since they follow the 

speaker’s speech. Not only do they not have to think about what ideas to express and in what order, or what linguistic style and register to use, but even syntactically and lexically they can often follow the speaker’s lead (which has both positive and negative implications, as discussed in section 2.2 of chapter 7).

It follows that on the whole, contrary to comprehension, which may be considered easier for the delegate most of the time, production may be either easier or more difficult for the interpreter, depending on the particular circum­stances. If for example the speaker is experienced, using his or her native tongue, speaking on a subject he or she knows well, and expressing well­rehearsed ideas, and if he or she takes the floor without much to gain or to lose, it may be assumed that the interpreter’s production task is more difficult. If, however, the speaker is taking part in an international conference for the first time, using a language he or she does not know very well, speaks on a subject he or she is not very familiar with, has not had time to prepare the presentation thoroughly, and knows his or her future may depend on the speech, his or her task can be more difficult than the interpreter’s. 

3.3 The Gravitational Model of linguistic availability In view of the importance of the lexical component of language skills in interpretation, an essentially lexicon-oriented model of language proficiency was developed. However, the Model is also applicable to syntactic and other linguistic rules. The Model represents the status of an individual’s oral or written command of a language at a particular point in time and in particular circumstances, by describing the relative availabiliy of lexical units and linguistic rules. 

The Model consists of a variable part and an invariable part. The latter refers to language elements the availability of which is assumed to be constant or to vary very slowly. This applies to the most basic rules of grammar (basic conjugations, the formation of plurals, etc.) and to a small number of the most frequently used words in the language. The variable part is larger by several orders of magnitude, as it includes at least dozens of rules and many thousands of words and idioms. 

In the diagram (Figure 1), the invariable part is represented by a Nucleus. The variable part is made of Words and Rules revolving on Orbits around the Nucleus. The term Words (capitalized) refers to lexical units such as individual words and idioms, as well as to frequently used phrases (“Thank you, Mister Chairman” and “Mister Chairman, Ladies and Gentlemen” are typical exam­ples). Rules are all the linguistic rules that apply when words are selected, modified, and combined when linguistic utterances are constructed. Because of the general nature and intended didactic use of the Model in the classroom, I do not consider it useful to define Words and Rules further: the Model deals with tens of thousands of words and the general dynamics of the System rather than with precise, quantitative rules, and is used to character ize the interpret­ers’ language skills and requirements rather than measure them. If the Model is to be used for lexicometric purposes, Words and Rules { hereinafter Words ) will have to be defined more precisely, depending on the particular study it is used for.

Orbits represent various degrees of availability of the Words: the more distant an Orbit is from the Nucleus, the more processing capacity and the more time are required to access Words on that Orbit. 

Orbits belong to one of two concentric areas: the Active Zone, directly around the Nucleus, and the Passive Zone, which surrounds the Active Zone. The Active Zone is composed of Words which are available to the speaker (or writer) for Text production. The closer the Orbit is to the Nucleus, the easier and quicker the Words it carries are available to the speaker. The Passive Zone is composed of Words which the speaker understands, but which are not available for speech production. The closer an Orbit is to the Active Zone (and to the Nucleus), the faster and easier the comprehension of the Words on it. 

The distinction between the Active Zone and the Passive Zone may seem artificial, if only because active Words can generally also be understood and because any Word that has been understood can also be repeated, and can therefore be considered “active.” However, the dual System is helpful in representing dynamic trends of Words as outlined below, and can be quite accurate in representing situations in which there are Words which can be understood if used by another speaker, but are not available for expression. This may for instance be the case of Words that the subject has not learned, but can understand when they are first encountered because of their morphology or their similarity to Words he or she does know in another language (see the discussion of the Escort Effect in section 3.4). Another ease is that of Words which the speaker knows, but is reluctant to use for one reason or another. He or she: may not be sure that the Word is appropriate for the particular situation as regards style, level of politeness, etc.; may not be sure of the precise meaning of the Word; may fear that the Word that comes to mind is a faux ami (`false friend’, i.e. an unwanted intruder from another language which looks similar to a Word in the language intended to be used).

The Model or System as a whole thus represents the full set of Words and Rules available to a subject for comprehension and/or active use at a given point in time. When the focus of interest is a subject’s command of words in a given field, say technical, scientific, or literary, a Sector can be defined by way of two straight segments going from the Nucleus to the most distant Orbit in the Passive Zone (see Figure 1). A Sector thus accounts for the full set of Words available to the subject for comprehension and/or active use in a particular field or in a thematic subset of the total System. 

Alternatively, one may be interested in a subset of the Words available to the subject within a certain range of processing capacity values. For example, those Words which are highly available, that is, in the innermost Orbits in the Active and/or Passive Zones, are the most relevant ones in interpreting, as explained in section 3.5. 

3.4 Dynamics of the Gravitational Model

As mentioned above, any given Model is a snapshot of the situation for a given individual at a given point in time under given circumstances, although aver­age values may theoretically be computed for a whole population or for an individual over a period of time. A very important feature of lexical and syntactic availability is its dynamic nature: Words are learned, become more or less available, are forgotten. These dynamics can be described by a small set of rules. 

These rules were derived intuitively, from observation. Some are sup­ported by psycholinguistic studies, which are quoted in the following sections, but many have not been tested empirically. Moreover, no quantitative assess­ment of their actual manifestation in linguistic performance seems to be available. They are therefore only presented here as trends, to be used for explanatory purposes and to support the selection of linguistic enhancement strategies as explained below. 

Rule 1: The Centrifugal Principle 


What is meant by stimulation here is either active use in speech or writing, or passive exposure, when words and rules are heard or read (and identified) by the subject. 

Rule 1 refers to the phenomenon that when Words are not used, they tend to become less active (if in the Active Zone), then become passive only, then less available as passive entities, and then disappear from the subject’s System. Though no precise quantitative assessments can be made, the process is generally rather slow. Under ordinary conditions, it may take months or years for a Word to be forgotten. In a speaker with a good command of the language, the process is much slower for Rules than for Words (one- reason being that they are much smaller in number, and most of them are stimulated much more frequently). 

A corollary of this rule is that, all other things being equal, the more recently a Word or Rule has been acquired, the closer in it is (see Matthei and Roeper 1985: 184). The reason is that the Centrifugal Effect has had less time to act on recently acquired entities. 

Rule 2: The Centripetal Effect of stimulation 


When a Word or Rule is heard or read (passive stimulation), or used by the subject in oral discourse or in a written text (active stimulation), it becomes more available for passive and/or active use. This migration is very rapid as compared to the centrifugal effect: a newly learned Word can become very active within minutes. On the other hand, it is quite possible that the outward migration of Words which migrated inward rapidly is faster than that of Words that became more available in a long, slow process. This sequence, comprising a rapid centripetal progression followed by a slower centrifugal migration, is often found in terminological preparation for conferences: interpreters achieve within hours or even minutes high active and passive availability for technical terms they had never encountered before; this lasts for the duration of the conference, after which they may forget these terms in a few days to a few weeks. 

Rule 3: Stimulation frequency and the Centripetal Effect 


Words used very frequently tend to become more available than Words or Rules used less frequently. 

I have found no direct description of this process in the literature, but the dependency between frequency of stimulation and ease of perception is well documented, with statements to the effect that: 

the frequency of occurrence of a word in a language affects the time it takes to gain access to that word in the mental lexicon (Matthei and Roeper 1985: 182); 

frequently used words are perceived more easily and read more rapidly (Miller 1956: 272-273); 

word frequency plays an important role in coding and decoding (Leeson 1975: 116); 

rare words are “more difficult to process” (Clark and Clark 1977: 56); 

the more frequent a linguistic element, the more “deeply it is rooted in the psyche of the individual and the community” (Mahmoudian 1982: 189). 

However, frequency of stimulation should not be regarded as the only impor­tant factor affecting the Centripetal Effect. In particular, automatic repetition without a context and without cognitive operations does not appear to be very efficient. It seems that some processing has to be involved for the Centripetal Effect to occur, as is the case in actual comprehension or production circum­stances. It is also possible that there are particular repetition intervals that produce the strongest Centripetal Effect, rather than an ever-increasing repeti­tion frequency. Nevertheless, for practical purposes in the context of linguistic skill enhancement for interpretation purposes, the frequency-in-context rule seems to be the best approximation. (See, for example, a study reported by Biderman and Ravazzi 1984, in which the frequency of word repetition was found to be a very strong predictor of memorization in students.) 

Rule 4: The Centripetal Effect of active vs. passive stimulation 


Using a Word or Rule when speaking or writing pushes it more strongly toward the Nucleus than reading or hearing and decoding it. This rule is well known in foreign language teaching and provides justification for the numer­ous active drill exercises in language classes. Combined with Rule 5, it has strong implications in professional practice, as explained in section 3.7. 

Rule S: The Escort Effect and Interference Effect 


When a word becomes more available, other words that sound or look similar, or that have been associated with it psychologically (through a learning situation, an emotional situation, etc.) also tend to become more available. The phenomenon is also noted in the linguistic literature (see Costermans 1980: 20). This rule is very important with respect to both lexical acquisition and lexical maintenance: it suggests that although the initial acquisition of one particular Word may take some time and repeated active or passive stimula­tion, the initial acquisition of other Words closely related to it (for instance grammatical variations thereof, or other Words having the same etymological root) will be much faster; it also suggests that the use of one Word will not only push it toward the Nucleus or slow down its centrifugal drift, but will also have a similar effect on other Words associated with it. 

Another important fact is that the Escort Effect crosses interlinguistic boundaries. For instance, in the group that escorts the inward migration of the French Word “controleur”, one can expect to find Words such as the English “control” and “controller”, the German “Kontrolle ” and “kontrollieren “, etc. In particular, the Escort Effect accounts for the comparative speed at which adults can learn foreign languages, especially as regards scientific and techni­cal Words, which often have common roots in each family of languages. Conversely, it also explains why persons who have achieved a high level of proficiency in a language related to their own in a very short time may struggle for years with an unrelated language and show relatively little progress (see Gile 198$b for the case of Japanese learned by a Westerner). 

The negative side of the Escort Effect is linguistic interference, already mentioned several times in this volume: linguistic interference may induce the interpreter or translator to use a Word incorrectly as regards its meaning or connotation; to distort its meaning, sound, or spelling; or even to use a Word from the wrong language in the middle of a speech or text. For instance, in the ongoing experiment mentioned in section 1 of chapter 7, one interpreter translated “They think you’re stupid or you’re foolish” (third example given in section 1) into “ils peuvent penser que vous etes stupide ou fou” (stupid or insane). The error is most probably due to the phonetic similarity between “foolish” and ` fou”. 

Interference phenomena in interpretation and translation are not the same as in foreign language learning. Conference interpreters and professional high­level translators, are well-educated individuals with a very good command of their working languages. Furthermore, by training, they are very much aware of the dangers of linguistic interference and constantly endeavor to avoid it. It follows that although some gross interference may occasionally be found in the booth or in translations, most of it is subtler and less salient. In particular, what might be called silent interference may be frequent and is very difficult to detect. Silent interference, as defined here, is interference not manifest through a visible, significant alteration of the lexical or syntactic output in the target language. It can involve some slowing down of speech production due to increased processing associated with the filtering out of possible interfering language Words and Rules. It can also narrow the range of Words and Rules used by speakers, as they are eliminating those which they suspect may be due to interference. Silent interference is difficult to detect not only because its symptoms are not very salient, but also because other factors may induce similar phenomena. For instance, according to Meier 1964 (quoted by Hormann 1972), lexical restriction occurs under stress. Finally, anti-interfer­ence control as it is practiced by professionals during interpreting undoubtedly increases the processing capacity requirements of the production component, and may induce more fatigue and related effects (see chapter 7) that are difficult to ascribe directly to linguistic interference. 

3.5 The Gravitational Model and interpretation An obvious requirement of interpreters is that they have enough Words, in both the Active and the Passive Zones, to comprehend and produce speech in a conference situation. With respect to standard general language, this condi­tion is, at least theoretically, always met in competent conference interpreters. Problems may arise in nonstandard general language (regional dialects, slang, etc.), literary language, or specialized language. The latter category is particu­larly important, as it makes up much of the vocabulary of conferences and cannot be learned once and for all, both because of its extent (hundreds of thousands to more than a million lexical units in each language in the various scientific and technical fields), and because it is constantly changing (see chapter 6). 

Second, since interpretation is performed under heavy time pressure, only highly available Words are useful. As a rule, Words encountered occasionally and understood or available for speech production only after a comparatively large amount of processing (which may only take a second or a fraction of a second, but still involves a significantly longer process than the retrieval of highly available Words) cannot be used in interpretation, as they may take up valuable processing capacity and time and dead to serious problems i see ) chapter 7 ,. 

This means that interpreters in the booth do not use a11 their vocabulary, but only a elevant sub of the Words they know. This in turn may lead to further polarization of their System as compared to that of an ordinary speaker: in both the Active Zone and the Passive Zone, a number of highly available Words may be maintained in close Orbits by frequent stimulation, while the number of Words of medium availability is smaller than in an ordinary speaker (because most Words are either pushed inward by frequent repetition because they are relevant, or drift away because they are not used often enough ); and the number of low-availability words may be somewhat higher m the inter­preter than in the ordinary speaker. 

This polarization may not be very significant in the language of the country the interpreter lives in, because in that language, the environment offers natural and balanced stimulation similar to that which non-interpreters are exposed to. In a working language not spoken in the interpreter’s country of residence, however, the phenomenon may be more significant. Some inter­preters who are known to have a solid B language in the booth seem to use only basic vocabulary plus technical terms in that language (for the time being, this remains an impression, as it has not been investigated empirically). Beyond the professionally relevant Sector of their vocabulary, the availability of Words, and even the knowledge of Words in such individuals, may be significantly lower than what could be expected from educated adults displaying an appar­ently good command of the language while interpreting. 

The relevance of Words to the needs of interpreting is an important question, in terms of both initial vocabulary acquisition and maintenance. Although there are countless glossaries of technical terms, no study of the basic non-technical vocabulary used in conferences has ever been published. Interpreters reading books, newspapers, and magazines in non-native working languages with which they seem to have no difficulty in the interpreting booth, do encounter words unknown to them but familiar to native speakers. This suggests that the theoretical requirements for native-like command of working languages may be highly exaggerated, on the one hand, and non-optimized on the other: an ordinary native-like System (a native-like System not specifically representing the case of conference interpreters) may contain a significant proportion of Words which are not useful in interpretation, while useful Words may not be available enough. This also has some implications for language acquisition and maintenance methods, as explained in section 3.7. 

Another important point is that linguistic availability for written language is generally not the same as availability for oral language. The two systems are obviously correlated, both because they reflect the same root language and because of the Escort Effect between them, but: 

The vocabulary and Rules of written language differ from those of spoken language. Differences are more or less marked depending on the language and on the verbal habits of the speaker, as a function of sociolinguistic factors (professional occupation, socio-economic class, etc.). This means that Words are not stimulated equally in the oral and written Systems, depending on their relative frequency in spoken vs. written language. 

Word recognition in speeches is based on sound, while Word recognition in texts is based on images (although mental sound reconstruction may be involved in visual recognition and vice versa). In languages having a phonetic writing system, this may not have a very strong effect. In languages using ideograms and pictograms, however, the situation is quite different, because the association between meaning, visual aspect, and sound is not as strong. In particular, in Japanese it is possible to understand written words without knowing how to pronounce them. Thus the Escort Effect may play no role in the stimulation of the oral System from passive exposure in the written system, which is one of the reasons for the rather slow progress of Japanese vocabulary acquisition for oral use in a foreign learner (as discussed in Gile 1988b). 

Another characteristic feature of the vocabulary of conference interpreters is the richness and high volatility of its technical component: interpreters en­counter and use many thousands of technical terms in the course of their careers at a rate of a few dozen to a few hundred at each conference. Subjects cover an extremely wide range of human activities, virtually limitless: no activity is theoretically excluded if it involves people who do not speak the same language but want to communicate, provided some money is available to pay for the service. It follows that interpreters have a much wider technical vocabulary than most individuals, but with a large volatile component, as technical terms tend to be forgotten rather rapidly (which may be due to infrequent stimulation because of the long intervals between conferences on the same subject).

(Read an update on the Effort and Gravitational Models by Daniel Gile here)