Lengyel Zsolt-Navracsics Judit
2. The experiment 1. Objectives
Experiments with the technique of miniature artificial language are generally used for measuring the knowledge of the mother tongue as well as that of a foreign language. However, we expect to find some data on the ontogenesis of translation;
Lengyel Zsolt-Navracsics Judit
more precisely, to find an answer to the above question: Is translation a part of speech activity? Does the ability to translate have some kind of natural prece
dents?
2.2. Participants
Four age groups participated in the experiment, each of them consisting of 100 children.
(1) 8-9 year olds (studying a foreign language for one year);
(2) 9-10 year olds (studying a foreign language for two years);
(3) 10-11 year olds (studying a foreign language for three years);
(4) 12-13 year olds (studying a foreign language for five years).
The children attended an average primary school, they had two lessons of for
eign language a week.
All the children studied a foreign language (English, German or Russian), they attended the same school, and even the teachers were the same for some of the classes.
2.3. The circumstances
The children were given two sheets of paper. The first gave both instructions and text as follows (instructions were given orally as well. What follows is an English translation of the original Hungarian text):
“The following message was in the ‘Interstar' newspaper. Try to decipher the message with the help of the explanations and words given here!
<S gédüv nivf s u Diikv s go Pmakv. I pögukf fo Sümiv. Sümi nepv szimf ikem.
Gédüv s gat tipiszf s nepv, zéta gat map zov Sümiv mes gédüvv s sápil, isk gat sápil"
The second sheet contained the vocabulary and grammar explanations as follows:
Dük = Earth (planet)
— f = past tense 3rd person singular (stands right next to the verb, writ
ten together with it).
fo = to (stands before nouns, the noun in accusative) gat = 1 .no (word of negation, stands before verb)
2. before noun it is the opposite meaning of the noun (e.g. light
ness - gat lightness = darkness) gédüv = visitor, guest
go = about (stands before noun, noun in accusative)
i = helshelit (3rd person singular, pronoun) [in Hungarian there is only one form]
ikem = to him,forhim/her/it [in Hungarian there is only one form]
isk = only map = is
mes = and nep = milk
niv = come
Pmak = Mars (planet) pöguk = knock in
s = \ . a, an (indefinite article, e.g. a horse, stands always before a noun).
2. the (definite article, e.g. the horse, stands before inanimate and after animate nouns),
sápil = friendship Siimi = John
szim = give (transitive, intransitive) tipisz = love (transitive, intransitive)
u = -on/onto (stands before noun, noun in accusative)
v = suffix of accusative (stands after noun, written in one word) zéta = since then
zov = between, among (stands before noun or nouns, the noun is in ac
cusative).
The children’s task was to encode the text (we deliberately avoided the word
“translation”).
2.4. General outcome
The study for various reasons will not compare the age-related correctness of translations. Just a few remarks:
(1) A general improvement is observed from grade to grade. This tendency is remarkable because children in these classes do not learn translation in an explicit way (though they may have had similar kinds of homework). This makes us deem translation a latent component of language competence which, to a certain extent, develops itself.
(2) Each age group had some children who were not able to translate the text, or, who did a very poor job. This means there are only few children who have no sense of translation at all.
(3) An important question which was not investigated by us is: What is the cor
relation between the level of language command and the quality of transla
tion?
2.5. Text as an object of translation
Text, as is well known, is a complex product of language; it has its own gram
mar, semantics, pragmatics, function, metalanguage, etc. We must suppose that from the point of view of the translation of a text, children recognise the compo
nents step by step (gradually); consequently, in the beginning, not all the textual components are the objects of translation. The elements making up a text from the child’s point of view are set up according to a kind of order of importance or hier
archy. In this study we have sought to acquire some information on this hierarchy as it is mirrored in certain mistakes made by the subjects in deciphering the artifi
cial language.
Lengyel Zsolt-Navracsics Judit
2.2. Error analysis
The analysis of errors will be done according to a general model of translation:
Memory Analysis
Semantic Representation
Encyclopedic .Knowledge
Synthesis
Target Language Source
Language Text
3.1. Analysis
First step: Total underextension (observed only in one case): a global treat
ment of the textual information and metalinguistic information, minimal analysis, on the basis of elements understood the child uses age appropriate fantasy to cre
ate a new text.
Group II.
A visitor comes to the horse after them a boat comes. The witch tells what happened to her.
Second step: Gradual analysis: textual information is first separated from metalinguistic information.
Group I.
1. In knocked he straight number John. Metalinguistic information is used (—f = past tense, singular, third person) but read in the wrong way (the sub
ject confused the Hungarian egyes szám = singular, with egyenes szám = straight number).
2. ... since then darkness is there with them John and the visitor’s friendship ...
3. The visitors liked light milk since this time between lightness,John ...
On one hand, lightness and darkness,which found their way into the transla
tion, are examples of explanation (metalanguage). On the other, gat requires nega
tive transformation with verbs (gat map — *not is — there is no) and nouns (gatsápil— *no friendship — anger). These operations above are more diffi
cult than simple lexical substitutions.
4. 1. visitor comes ...
5. 1. to the Earth, from Mars
Errors are connected with the article s (s = 1. a, an (indefinite article) 2. the (def
inite article)). In (4) the homonym (egy = a, an, one indefinite article and cardinal number) may have caused trouble. In (5) the number stands for the definite article (the), which means the symbol used in metalinguistic description is reflected.
Group II.
1. Visiting guest 1 doesn’t like ...
2. Since this time 1 is John and visitor between 1 friendship only 1 friendship 3. Visitor the lightness[acc.] doesn’t like the milk [acc.], since this time light
ness is Johnfdat.f and the visitor[dat.] between a friendship ...
4. A visitor a horse from planet a mars from 5. 1 visiting guest
6. 1 earth 1 Mars Group III.
1. visitor comes to 1 earth 1. Mars ... John to. Visitor no 1 likes. Since this time no is John and guest between 1 friendship
2. 1 guest came ...
Group IV.
1. 1 guest came
2. ... since this time there is darkness between John and visitor’s friendship 3. Horse guest came onto horseearth[acc.] fromhorseMan [acc.] ... brings
John [acc.].
4. Guest horse...Jo/m [acc.] and guest horse between friendship ...
The above examples infer that separation of metalinguistic information from textual information is not easy at all. This kind of error occurs in all age groups, so we must suppose it is a longer, prolonged process to decide what is to be trans
lated and what is just to be related to the operation of translation - as a kind of speech activity - itself.
Lengyel Zsolt-Navracsics Judit
Third step: Gradual separation of textual information from metalin
guistic information and making sense out of the metalinguistic informa
tion which was not understood Group I.
1. 2 visitors came ...
Group II.
1. ... an Earth[acc.] from that mars [acc.] is bringing Jo hn[acc.]
Group III.
1. A visitor is coming for 1 hour for one to the earth 2. The guest lightness liked since this time.
Group IV.
1. ... only fade friendship 2. ... only obscure friendship
Since there are a couple of examples from each age group of subjects “making sense", this process may not be age-dependent, but rather person-dependent.
3.2. Encyclopaedic knowledge
First step: The lack of encyclopaedic knowledge results in nonsense texts.
Insufficient knowledge of the lexical morpheme’s semantics makes impossible the correct use of grammatical morphemes and rules of topicalisation.
Group I.
1. ... came to a planet a Mars ...
Group II.
1. ... to a planet. From a Mars 2. ... comes the planet Mars froma 3. on a mars from ... John to.
4. an earth a from planet from ... brings John[ acc.] guest a Group III.
1. to the Earth from the Mars from planet 2. comes anto on our Earth afrom from Mars Group IV.
1. Guest comes from the planet from a Mars 2. ... from an earth from the mars
3. ... came a to the earth from a Mars
4. ... came an fromlonto Earth from a Mars 5. ... comes to an Earth from a planet
6. A visitor comes a to from an earth from a Mars.
This mistake occurred in all groups (although less frequently among the younger ones, since fewer of these children reached this point of “translation”).
Second step: Trying to avoid the lack of encyclopaedic knowledge (mak
ing “sense”) by creating a pseudo-meaningful text:
Group IV.
1. A visitor came to the earth a mars. The earth knocked in to him.
2. ... came to the planet Mars.
3. ... comes from a planet, it is the Mars
4. ... he called earth came from a Mars. He knocked in to a John.
5. A guest is visiting the Earth from the Mars.
6. ... from the earth to the Mars.
This tendency occurred only among the older students. Thus “making a text sensible” is a propensity which becomes stronger and stronger during language learning.
Third step: The “overgeneralization” of encyclopaedic knowledge Group III.
1. From a heavenly body a guest came from a heavenly body from the Mars.
Since this was a rare phenomenon it may be somewhat person-dependent.
3.2. Grammatical analysis
This type of analysis can only be the subject of inquiry if a subject has the suffi
cient prerequisites, namely the ability to separate metalinguistic from textual in
formation and then mobilise encyclopaedic knowledge.
Putting the segments which were hard to analyse into the target language text (mixed sentences, semilingual texts):
Group I.
1. I knocked in ... the insert of / (i = ő ..., there was a small “i”).The person
al pronoun ő (correctly!) is cleared from the text, but as an untranslated element it is placed in the target language text.
2. ... no is zov John and guest a friendship 3. 5guest... s mars
Group III.
1. 5 visitor ... knocks in is bringing John.
These solutions remind us of the usage of source language words which have no lexical equivalents in the target language.
The child is able to analyse the elements of the source language, he is able to carry out the lexical equivalent operation, but he is unable to perform the needed morphological, syntactical operations in the target language.
Group II.
1. ... Earth to the Mars from
2. ...John to ... Jo/m [acc.] and guest between ...
3. He knocks in John to.
4. ... on Earth to an Earth to afrom Mars from
Lengyel Zsolt-Navracsics Judit 5. ... Guest a friendship
6. Visitor a doesn’t like one comes ...
7. He earth an to what mars the from. He knocks in he John to He. John milk what gives to him. Guest a doesn’t like he a milk what.
Group III.
1. ... a friendship He does not recognise that the articles belong to the pre
ceding (guest) word.
2. Visitor a...
3. ... John and visitor [2lcc.] between a friendship ...
4. ... no is John[ acc.] and his guest[ acc.] between a friendship 5. ... John[ acc.] and visitor[ acc.] between us a friendship ...
Group IV.
1. A friendship between 2. Visitor a ...
3. John and the guest[acc.] between 4. John to
4. Summary
Our findings seems to support the thesis that translation is a natural ingre
dient of the language faculties.
(1) The task was performed (admittedly at different levels) by most of children without their being conscious of “translating”.
(2) Translation has certain components which develop and improve without in
structed learning (not differing in this respect from some of the processes of first language acquisition).
Translation has its instructed elements, too:
(1) Metalinguistic skills play two roles. On the one hand, they are necessary pre
requisites for translation. On the other hand, translation and metalinguistic awareness are complementary processes. This last point relates translation to the ongoing processes of learning to read and write.
(2) Encyclopaedic knowledge is also a necessary prerequisite. The relation be
tween the encyclopaedic knowledge and translation reminds us of the inter
twining development of speech and thought.