Báo cáo khoa học: "Mechanical Translation of French"

Chia sẻ: Nghetay_1 Nghetay_1 | Ngày: | Loại File: PDF | Số trang:10

Thêm vào BST

Báo xấu

47
lượt xem 2
download

Download Vui lòng tải xuống để xem tài liệu đầy đủ

IN THE NEAR FUTURE an attempt to translate from a foreign language by machine will be made at the computational laboratory of Birkbeck College. It will differ from previous experiments in that the sentences or passages of French to be translated will not have been specially chosen or "doctored" in any way beforehand: on the contrary, they will be constructed by French scholars, who will be invited to do their best to fault the machine. What follows is an account of the method, or program, which makes such an experiment possible. ...

Chủ đề:

Bình luận(0) Đăng nhập để gửi bình luận!

Lưu

Nội dung Text: Báo cáo khoa học: "Mechanical Translation of French"

[Mechanical Translation, vol.3, no.2, November 1956; pp. 52-61] Mechanical Translation of French † L. Brandwood, Birkbeck College Research Laboratory, University of London IN THE NEAR FUTURE an attempt to translate in its dictionary. The comparison takes the from a foreign language by machine will be form of a simple subtraction. Starting with a made at the computational laboratory of Birk- number at the middle of the dictionary it sub- beck College. It will differ from previous ex- tracts from this the number of the word being periments in that the sentences or passages of translated: if the result is zero, the word has French to be translated will not have been spe- been identified and the English equivalent, cially chosen or "doctored" in any way before- which is marked alongside the French can be hand: on the contrary, they will be constructed output; if the result is not zero, then the num- by French scholars, who will be invited to do ber sought lies in one or other of the two halves their best to fault the machine. What follows of the thus divided dictionary according to is an account of the method, or program, which whether the result is positive or negative. So, makes such an experiment possible. taking the number at the middle of the relevant First of all, however, I think a brief explana- half, the machine carries out the same compa- tion of the general procedure in handling words rison as before, and if -- as is more than by electronic computer will not be out of place. likely -- this too proves unsuccessful, it con- To a layman like myself the information that tinues halving the numbers that remain until such a computer works in terms of electronic the one required is located. impulses conveys very little, but as I suspect This ingenious process of elimination, which t hat the majority of readers are in the same was devised by Dr. A.D. Booth, ensures the boat, I shall no doubt be forgiven for using speedy identification of words even in a very terms which though strictly not correct have large dictionary. In fact the size of the latter at least the advantage of being more readily can be greatly increased without appreciably intelligible. lengthening the look-up process. To be pre- We may say then that the computer works, cise, the number of comparisons required is like any other calculating machine, with digits. doubled as the number of words in the diction- It cannot recognize letters; at any rate not as ary is squared. This means that since on an they stand. In order that it may recognize average seven comparisons are required to them the alphabet has first to be coded, that is identify any word in a dictionary of a hundred, t o say, each letter represented by a particular only fourteen will be necessary to cope with a number. Since a word is merely a collection dictionary of ten thousand or twenty-eight to of letters and we have a number for each letter, cope with one of a hundred million. we can now put any words we choose into the The capabilities of the look-up method might computer to be stored on a magnetic drum in seem rather wasted when it is revealed that the terms of impulses. These constitute the ma- maximum storage capacity of the Birkbeck chine's "dictionary", which may be looked computer is at present only 250 dictionary items upon as exactly like an ordinary French- (i.e. 250 foreign words each with one English English dictionary, but with the numbers repre- equivalent), though this will shortly be increased senting the French words arranged in order of to 2000. It must be borne in mind, however, either ascending or descending magnitude. that this is not the limit of all computers, even This is important for the rapid identification of of those existing, and storage capacities will the words in the text being translated, which continue to grow larger. But that lies in the takes place as follows. The machine identifies future, and for the time being all our attempts an incoming word by comparing the number at translation have to be scaled down and done, which represents it with the numbers contained so to speak, in miniature. The machine's capacity allowed us 250 dic- tionary items. We could not use all of these † An earlier version of this paper appeared for the actual dictionary, however, because we in Babel, Vol.11, No. 3, 1956, a special issue wished to include a few elementary syntactical devoted to mechanical translation. instructions to enable the machine to produce
M echanical Translation of French 53 taking into account the syntactical context. something a little better than a word-for-word This, however, is out of the question with the translation. We did not know beforehand in present storage capacity. Nevertheless, dis- what proportion the dictionary space should be regarding the compound tenses, which are ac- divided between vocabulary and grammar, but counted for by the separate consideration of reckoned that with a vocabulary of about a avoir or être and the perfect participle of each hundred words there would be enough space verb, there still remain about 30 different forms left for the instructions in mind and a little over to each regular verb, the actual figures for the for subsequent modifications. What was wanted four conjugations being: donner 33, finir 29, therefore, was a vocabulary which, though vendre 32, recevoir 31. For an irregular verb, very small, would yet be capable of producing unless it is defective, the number is usually a good French sentences in a reasonable variety; little higher. In short, the 25 verbs (13 regu- in short, the proportion of the various parts of lar, 12 irregular) produce well over 800 differ- speech which it contained had to be correctly ent forms, all of which the machine must be balanced. This was achieved in a very simple able to translate. Obviously, such a large fashion by selecting a passage from a book and number will not fit into our machine's dictionary extracting the first hundred different words: to as they stand. How then are they to be incor- these were added any missing forms of the porated? The solution thought of as long ago personal, possessive and relative pronouns en- as 1947 by Booth and Richens, is that a word countered. The result was a vocabulary con- should be divided into a stem and an ending. sisting of 25 verbs, 2 adverbs, 23 nouns, 10 Here we must pause in order to define our adjectives, 3 numerals, 8 prepositions, 6 con- terms. A stem is not necessarily a stem in the junctions, all forms of the unstressed personal sense that grammarians know it. Generally pronoun (except en and y), all forms of the speaking, it is the longest part of a word simple possessive pronoun, the relatives qui, common to all forms (inflections) of that word. que, dont, the definite and indefinite articles The stem of donner, for example, is donn-; and the negative ne. . . .pas. Volunteers will be what remains of each inflection of donner after presented with this vocabulary and asked to con- this has been subtracted constitutes an ending. struct sentences or short passages for the ma- This, however, is not a universal rule, since chine to translate. They will be permitted to there are instances where a single stem would use the singular and plural forms of all nouns account for all forms of the verb, but more and adjectives, the masculine and feminine of than one is chosen for the sake of utility (e. g. the adjectives, and all parts of the verb except the stem acqu- would cover all forms of the 1st person imperative and the subjunctive acquérir, but in addition acquier- has to be mood. The former is excluded because it re- used in order to comply with the general system quires a rather periphrastic translation, which covering all verbs). Utility might in fact be combined with its comparative rarity makes said to be the only principle of grammar in me- it hardly worth bothering about, the latter chanical translation. because many of its forms are identical with Verbs are divided, then, into stems and end- those of the present and imperfect indicative, ings, and the following results cover each of and could only be distinguished from these by the Four Regular Conjugations: I. Donner (stem: donn-) Person Singular Plural Tense 1st 2nd 3rd 1st 2nd 3rd Present e es e ons ez ent Imperative e e ez ent Future r ai ras ra rons rez ront Conditional r ais r ais rait rions riez raient Imperfect ais ais ait ions iez aient Past Hist. ai as a âmes âtes erent Infinitive -er Present Participle ant, ante, ants, antes Perfect Participle é, ée, és, ées
54 L. Brandwood II. Finir (stem: fin-) Person Singular Plural Tense 1st 2nd 3rd 1st 2nd 3rd Present is is it issons issez issent Imperative is i sse i ssez issent Future i rai i ras i ra irons i rez iront Conditional i rais i rais irait irions iriez iraient Imperfect i ssais i ssais issait issions issiez issaient Past Hist. is is it îmes îtes irent Infinitive -ir P resent Participle issant, issante, issants, issantes P erfect Participle i, ie, is, ies III. Recevoir (stem: rec-) † Person Singular Plural Tense 1st 2nd 3rd 1st 2nd 3rd Present ois ois oit evons evez oivent Imperative ois oive evez oivent Future evrai evras evra evrons evrez evront Conditional evrais evrais evrait evrions evriez evraient Imperfect evais evais evait evions eviez evaient P ast. Hist. us us ut ûmes ûtes urent Infinitive -evoir Present Participle evant, evante, evants, evantes P erfect Participle u, ue, us, ues IV. Vendre (stem: vend-) Person Singular Plural Tense 1st 2nd 3rd 1st 2nd 3rd Present s s - ons ez ent Imperative s e ez ent Future rai ras ra rons rez ront Conditional rais rais rait rions riez raient Imperfect ais ais ait ions iez aient Past. Hist. is is it îmes îtes irent Infinitive -re Present Participle ant, ante, ants, antes P erfect Participle u, ue, us, ues † ç is treated as c.
M ec ha nica l Tra n sla tio n o f F re nc h 55 the 1st conjugation). Nor should they be includ- If now the French verb stems and their English ed as they appear above for two reasons: translations are entered into the machine's dic- Firstly the use of different inflections to indi- tionary together with an instruction that each cate difference in person and number is on the word fed in for translation is to be identified whole a linguistic redundancy, so that where with the stem in the dictionary which it most two identical endings occur we need in fact closely resembles, we will be supplied with the make only one entry in the dictionary: secondly, basic meaning of every verb in the passage to inasmuch as the endings are used merely to dis- be translated. The tense, however, is still tinguish tense, there is no necessity to enter lacking and for this we have to refer to the end- the whole ending but only sufficient to enable ings. These too need to be included in the dic- t his distinction to be made (e.g. the -rai p art tionary, though at the same time they must be of the ending -evrai in recevoir is enough to kept apart and distinct from the stems; other- indicate that the tense is future). These dis- wise confusions will arise (e.g. between the tinctive endings are 46 in number, i.e. singular present indicative of avoir and the endings denoting the singular Past Historic of Present e s z t ois oit =stem (is) (it) Future r ai ras ra rons rez ront Conditional r ais rait rions riez raient Imperfect ais ait ions iez aient Past Historic ai as a mes tes rent is it us ut P erfect Part. é ée és ées i ie (is) ies ue ues Infinitive r re u (us) P resent Part. ant scheme. If the stem had been fini- the follow- One or two explanatory remarks are necessary ing difficulties would have arisen: here. The "=stem" means that there is no ending (e.g. il vend). The feminine and plural 1. Endings -s and -t forms of the present participle are omitted 1st, 2nd, 3rd pers. sing. pres. indic. because they belong strictly speaking to its ad- = jectival function. The forms in parentheses, 1st, 2nd, 3rd pers. sing. past historic though characteristic of the tense under which 2. Ending -s they are marked, are also found in another 1st, 2nd, pers. sing. pres. indic. and past tense of which it is preferable to regard them historic as primarily characteristic. This is because = the limited capacity of the machine prevents the p erf. particip. masc. plural inclusion of instructions (assuming that they 3. No ending: bare stem could be formulated) which would enable it to Sing. perf. particip. masc. distinguish these formally identical tenses; con- = sequently, for the time being one translation 3rd pers. sing. pres. indic. of vendre type must be made to serve for both, and it is na- verbs turally desirable that this translation should be determined by the tense in which this disputed 4. Ending -e and -es form has the wider occurrence. The ending Perf. particip. fem. sing. and plural -is, for example, represents the present indi- = cative in finir type verbs, the past historic in 1st and 2nd pers. sing. respectively pres. finir, vendre and sentir type verbs. Therefore indic. donner type verbs we take it as characteristic primarily of the latter tense; it is just a question again of utility. With the stem fin- on the other hand, only the Clearly, however, the fewer there are of first two of these are encountered. Fin-, there- these identical tense forms, the lesser the con- fore, is the best stem, but only, it must be re- fusion will be. Hence the stem fin- instead of peated, in the present scheme. As the diction- fini- for the finir type verb in the present
56 L. Brandwood indic. ending in -x, and one or two such as ary is enlarged and takes in more irregular p rendre a nd asseoir a 3 rd pers. sing. in -d . verbs we may find that there are ambiguities This is solved by adding -x and -d to the list of which can be solved by making the stem fini- endings denoting the present tense. instead of fin - (e.g. the verb saisir will have a stem identical in form to the 1st and 2nd pers. 2. There are some verbs with a past historic s ing. pres. indic. of savoir , unless the 'i ' is form which resembles none of the other tenses added). The important thing to remember is (except in one or two instances the perfect par- that there can be no one optimum system for ticiple), and consequently requires a stem pe- all translation programs: their general prin- culiar to itself. One would think, since this ciples may very well be the same, but the de- s tem is peculiar to a past tense, that its trans- tails must always vary according to the size of lation in the dictionary could be in the past the dictionary and the nature of its contents. tense without reference to the ending. For This applies particularly to stems of verbs, eus, the past historic of avoir, for example, which will continually have to be modified in the stem would be eu- and the translation "had" order to avoid ambiguities (e. g. 1st, 2nd, 3rd This, however, will not work. The singular pers. sing. pres. indic. of vivre = 1 st, 2nd, with its "present" endings -s and -t comes 3 rd pers. sing. past historic of voir, w hich in out all right, but the plural endings -mes, its 1st and 2nd persons may again be confused -tes , and -rent are marked in our system as with the stem of the verb viser). Since corre- characteristic of the past historic, so that the spondences of this type are far from being un- translation of the plural will be the past tense common, it may prove advisable in a larger of "had", which is not at all right. No matter dictionary to revert to the conjugational classi- whether we give the stem the equivalent "have" fication each with its separate procedure, in or "had", there will be an anomaly between the order to reduce the confusion. translation of singular and plural forms. The Returning to our explanation of the table of solution is to make the stem not eu-, but e-. endings, the reason for using these particular The endings will now be found all in the past forms to represent the present indicative is to historic list instead of half in this and half in be found in the rule previously mentioned, ac- the present indicative list, and a completely cording to which the machine identifies any in- correct translation will be ensured. Besides coming word or part of a word with the nearest avoir this solution is applicable to only a few equivalent (by which is meant the nearest o ther irregular verbs (e. g. croire, c roître , shorter equivalent) in its dictionary. There- plaire) and not to all those to which it would be f ore, if we enter, for example, -s i n the dic- d esirable to apply it (e.g. être , voir , savoir, tionary as a present indicative ending, the m ouvoir , pouvoir , boire , faire , lire , prendre, machine will identify with it not merely the mettre). This is because any stem devised for bare -s of vendre and sentir type verbs, but the past historic of these verbs will be ignored also the -es of the 1st and the -ons of all con- by reason of the rule that incoming words must jugations, simply because there is no other be identified with the longest similar stem in entry in its dictionary with which it can identify the dictionary. Let me make this clear with them. It is for this reason that the 1st/2nd an illustration: for the past historic of voir, a nd 3rd pers. sing. pres. indic. of recevoir i .e. vis , vis , vit , vîmes , vîtes, v irent , we type verbs (-ois, -oit) has to be entered in full, can only set up a stem to cover vit (i. e. vi-) since otherwise they would be identified not vîmes and vîtes (i.e. vî-); whatever we do with the present forms -s and -t, but the past about the other three they are bound to be con- h istoric -is a nd -it . fused with the stem of either the verb vis-er The system described above covers not only or the verb vir-er. In cases like this a possible the regular but also the irregular verbs in the solution is to enter the ambiguous forms in the majority of their forms. The only difference dictionary as they stand together with an in- is that many of the latter require several stems struction that if they are found without an end- instead of one; aller, for example, the follow- ing they belong to a different verb and have a ing: vais, va-, vont, all-, i-. Of course, different translation than if they are found with difficulties do arise, just as they do in the re- an ending. So when vis occurred alone, it gular verbs, but their number is relatively would be recognized as belonging to voir, when small. There are five main types. it occurred with an ending, that is in a longer form, it would be ascribed to viser. The 3rd 1 . O ne or two verbs such as valoir a nd pers. virent would no longer be ambiguous, if vouloir have a 1st and 2nd pers. sing. pres.
Mechanical Translation of French 57 entered in entirety. ric, whilst the fem. forms of the same have Other verbs with past historic forms which endings which are marked in the above scheme cannot be treated in the manner of avoir are as primarily characteristic of the present indi- tenir and venir (tins and vins) and those which, cative. like finir in the regular verbs, have forms b) All perfect participles ending in -u or identical in the singular to those of the present -i in the masc. sing. will end in -us and -is t ense (e.g. dire , rire , suffire , confire, c ir - in the masc. plur. -forms characteristic of the concire, conclure). The answer for tenir and past historic. The solution is again that venir is again to enter the offending forms into mentioned in no. 3. In our present program, t he dictionary in their entirety; for dire , rire, however, no difficulty is encountered over this etc., as for finir, only an examination by the similarity of perfect participle and past histo- machine of the context in which they occur will ric, since we are confining ourselves to indi- succeed. This, however, lies beyond the scope cating in translation only whether a verb is of our present modest program, which must p ast, present, or future. -- As regards mean- perforce allow the ambiguity to remain unre- ing no distinction is made between the various solved. forms of the past tense. These, then, are the main difficulties which 3. S everal verbs (e.g. dire, f aire , écrire, have been encountered in drawing up a system conduire, craindre) have a perfect participle for translating verbs. Some may have been i dentical in form to the 3rd pers. sing. pres. overlooked, important ones at that -- I cannot indic. This causes no difficulty in the compound be certain. Of one thing, however, I am quite tenses, where the accompanying auxiliary certain; that is that there is not a single diffi- serves to distinguish the participle as such, but culty anywhere which is incapable of being sur- when it is used adjectivally, then not only the mounted. m asc. sing. but also the masc. plur. and the In the method described above, first the feminine forms will all be interpreted by the meaning of a verb is obtained from the stem, machine as present indicatives. A solution by then the tense is discovered from the ending. consideration of form is possible only for the This is only one way. There is another, as the unique ending of the masc. plur. -ts (e.g. reader may already have observed for himself condui -ts), those of the others being common in perusing the table of endings, namely by a t o other tenses, i.e. the masc. sing.(-t ), as particular letter in the ending instead of by the we have said, to the 3rd pers.sing. pres. indic., whole ending. The most obvious example is t he fem. sing. ( -te) to the 1st and 3rd pers. the letter 'r ', which is sufficient to indicate sing. pres. indic. of 1st conjugation verbs like that the tense is either conditional or future jeter which double the 't' (i.e. jet-te), and the indicative. In place of an ending dictionary, f em. plur. ( -tes ) to the 2nd pers. sing. pres. therefore, we can have a series of instructions indic. of these verbs (i.e. jet-tes) or the 2nd like this: pers. plur. past historic of all verbs. An extra s tem (i.e. jett -) for the jeter t ype verbs helps 1. I f the ending ends in -r o r -re , then the a little, but a complete solution is possible only verb is an infinitive, and "to" should be in- if the machine can take account of the syntacti- serted before the translation of the stem. The cal context of these adjectival perfect participles. parentheses denote that the "to" will not always be needed to make good English in the transla- 4. Two of these verbs, faire and dire, present tion, e . g . , in puis-je sortir? (may I get out), a slightly more complicated problem, since not whereas il va partir (he is going to leave) does only is the fem. plur. of their perfect participle require it. When larger storage capacities identical in form with the 2nd pers. plur. of the are available, it will be possible to incorporate p res. indic. (faites a nd dites), but this form instructions enabling the machine itself to itself has a past historic ending ( -tes). This make this distinction. means that faites and dites have to be entered into the dictionary as they stand, which gets 2. If the ending ends in -ant, we are dealing round the second part of the difficulty, leaving with the present participle and "-ing" should the first to be dealt with in the way suggested be added to the translation of the stem. Again in no. 3 above. it would be quite easy with a larger storage at one's disposal to include an instruction re- 5. a) The masc. sing. perfect participle of quiring the machine to deduct the mute -e from asseoir (assis) and acquérir (acquis) are identi- words ending so, before adding the "-ing". cal with the 1st and 2nd pers. sing. past histo-
58 L. Brandwood 3. If the ending contains the letter r and this cative, which would consequently be translated is not followed by -ent (in which case the tense as a past historic. As it is, only the singular is the past historic), then the verb in question forms of the past historic are confused with is either future or conditional. If now the letter those of the present indicative. ) i occurs after (but not necessarily immediately 5. Everything not accounted for by the previ- after) this r and is not the last letter of the ous instructions is in the past tense, and "-(e)d" ending, the verb is conditional, and the word should be added to the translation of the stem. "would" must be inserted in front of the stem (Note: this will result in some odd looking meaning. Otherwise it is the future tense, and forms in the case of the English strong verbs, the word to be inserted is "will". (The English but as we have said so often before, it is only idiom of "shall" with the 1st person and the the restricted storage capacity of the present interchange of this and "will" with a change of machine which forbids the incorporation of these emphasis could likewise be included in a ma- strong forms (e.g. "sang, sung") alongside the chine with a larger dictionary space). regular form ("sing") together with instructions 4. If the ending begins with e (not followed by for their correct application.) a), s (not followed by a or i), t (not followed These are two ways of ascertaining the tense by anything), x, o, n, 1, d, or if there is no of a verb, which we may call the stem-ending ending (i. e. the stem is bare) and avoir or and the eliminating method respectively. The être does not precede, then the verb is in the question now arises of which is the better, but present indicative, and the stem meaning is an this is not easily answered, depending as it adequate translation. (The same can be said of does to a great extent on the machine, the size the suffixed -s in the 3rd pers. sing. of the of the dictionary employed, the nature of the English present as of the other refinements text to be translated and the quality of transla- mentioned above. ) If there is no ending and tion required. The eliminating method, for in- a voir o r être d oes precede, the verb is in the stance, must assume that the person receiving perfect tense (see no. 5). the translation will be content to have all past The reason why e, s, x, o, d, t indicate the tenses translated simply as such and prepared present tense will be clear. The reasons why to use his own common sense and judgement to n and 1 do, and why e should not be followed by distinguish where the sense requires a perfect a, s, by a or i, or t by any letter are as ( i.e. has done), imperfect (i.e. was doing), or follows : p ast historic (i.e. did). If the more precise n - enables certain irregular verbs which translation is desired, then there is no doubt double the final consonant of their stem in the that the stem-ending method is to be preferred. 3 rd pers. plur. (e. g. venir - v iennent) to be As regards the nature of the text, if the tense comprehended under one stem (i. e. vien-) in which it is written is the past, the stem- instead of two. ending method will probably be quicker than the 1- does the same for 1st conjugation verbs in eliminating method, which arrives at this tense - eler ( e.g. atteler - a ttelle) . only at its last stage; if, however, the tense is t is not followed by another letter since -tes the present and future, then it may well be that is an ending characteristic of the past historic. the eliminating method will prove quicker. (Note this means that 1st conjugation verbs in All this smacks too much of "perhaps" and -eter [e. g. jeter - jette] need two stems). "maybe", however, as would any remarks e is not followed by a or â because 1st conju- which we might hazard on the amount of space gation verbs in -ger (e. g. manger) retain the which each method would occupy in the ma- e before a and o, so that, but for this precau- chine's dictionary. A definite answer to all tion, the imperfect and past historic of these such questions will, after all, be forthcoming verbs would be translated as the present tense. when the machine starts on its program, since s is not followed by a or i, as happens in the this will be tried with both methods. In the imperfect of finir type verbs, if the stem fini- meantime two facts are quite certain: is chosen instead of fin - (i. e. -ssais , -ssait, a) the eliminating method is superior to the -ssions, -ssiez, -ssaient. ) stem-ending method when it comes to translat- ing perfect participles, because it needs to con- (Note: with the present method the stem fini- sider only the first letter of the ending, which is preferable to fin- for this reason: if the i is the one distinctive of the perfect (i.e. é, i, u) were not included in the stem, it would come whereas the stem-ending method has to have all at the beginning of all forms of the present indi- the forms (fem. and plur. ) written down for it
M echanical Translation of French 59 to choose from. The plural form of nouns and adjectives, both masculine and feminine, is easily recognized-- b) the stem-ending method is more economical as long as there is in fact a distinct plural f or treating the irregular verbs, because the form to be recognized--inasmuch as the ending fact that it can ignore the middle part of some will have either -x or -s added to it. Apart such verbs enables it to account for them with from irregulars the only class of nouns and ad- one stem less in each case than the eliminating jectives which does not comply with this "rule" method. The verb asseoir, for example, re- is that ending in -al, which drops the 1 before q uires three stems in the latter method, i.e. adding -ux (e.g. cheval, chevaux). This is re- a ssied- ( for the singular of the pres. indic., solved in the same way as the Type 2 feminine past historic and conditional), assey- (for the form above, that is, by putting the letters imperf. and plural of the pres. indic.), and common to both singular and plural (in this ass- (for future, past historic and conditional), case cheva- ) in the stem dictionary, the letter but only two in the former method which can ig- 1 which is dropped in the ending dictionary. nore the -ied- in the sing. pres. indic. and so So we have the following three types of regular use ass- for this too. Out of about seventy ir- nouns and adjectives: regular verbs examined, this economizing ap- plied to seventeen, (i. e. acquérir, boullir, Stem Ending Plural t enir , sentir, s ortir , partir , servir , asseoir, Type 1 noun chien (m) -ne (f ) -s s avoir , pouvoir , écrire , joindre , craindre, mettre, battre, s uivre). To these must be adj. joli (m) -e (f) -s added their compounds and any similarly conju- Type 2 noun chame- ( -au (m) -x gated verbs, though some of the latter are mis- leading because a verb with a similar stem -lle (f) -s prevents the reduction of the number of stems. adj. bre- -f (m) -s For instance, the only stem that would be -ve (f) -s needed for dormir w ould be dor-, but for the existence of a verb dorer; so the stem has to Type 3 noun cheva- -1 (m) -ux be dorm-, as for the eliminating method, and adj. principa- -1 (m) -ux t he sing. forms of the pres. indic. (dors, dort) have to be entered in the dictionary as they -le (f) -s stand. In short, we can identify all plurals which are That completes our treatment of verbs, and genuine forms and not identical with the singu- we can pass on now to explain the program for lar by the three suffixes -s, -x, -ux. The only dealing with the other parts of speech. Nouns difficulty is in getting the machine to recognize and adjectives are taken together, since both where these suffixes occur after the ending and may have feminine and plural forms, and these where directly after the stem. The latter is can be satisfactorily accounted for by the same true of the masculine forms of Type 1 nouns stem-ending procedure, thus: and adjectives and masculine adjectives and all Type 1. If the feminine form is created nouns of Type 3; the former of Type 2 nouns merely by adding extra letters to the masculine and adjectives and Type 3 feminine adjectives. form without involving any change in the latter, It can be done quite easily if the machine pro- then the masculine form (so marked) is entered ceeds according to the following instructions, i n the stem dictionary, the extra letters de- which do not even require it to distinguish be- noting the feminine form ( so marked) in the tween the three suffixes: e nding dictionary, (e.g. chien -ne, j oli -e ). Type 2. If the feminine form is created by 1. If when the machine looks in its stem dictionary, first altering the masculine form, then adding a) the incoming noun or adjective can be iden- t o it, (e.g. chameau , chamelle), t hen the tified exactly with one of the stems, then it is letters which are common to both forms (i.e. a Type 1 masculine noun/adjective and can be chame-) are entered in the stem dictionary, translated immediately. those peculiar to the masculine and feminine b) the incoming noun or adjective cannot be in the ending dictionary (i.e. -au, -lle). The identified exactly with a stem, because an ex- gender of a noun which has only one form for t ra letter or letters is left over, it cannot be both masculine and feminine forms must be de- translated immediately; t ermined by other means (e.g. by reference to 2. then the machine looks in its ending diction- the definite or indefinite article or an adjective, ary. if it is accompanied by any of these).
60 L. Brandwood a) If the extra letter(s) can now be identified tinction of gender in translation, (Consequently exactly with any ending contained in the diction- all adjectives can be included in this class.), ary, then the incoming word is either a femi- II. those which have a feminine or plural form nine singular noun/adjective of Type I, a mas- that does require a different translation from culine or feminine singular noun/adjective of the masculine or singular form. Type II, or a masculine or feminine singular The first class, in which the ending is re- noun/adjective of Type III, and can be trans- quired only for identification not translation, lated accordingly (see below); can be dealt with in the manner suggested above, b) the extra letter(s) can be identified with an that is by having an ending dictionary for the ending contained in the dictionary, but if there whole class of only 14 distinct endings (or 28, i s still an extra letter left over, then the in- if we wish to include the plurals and so get coming noun/adjective is the plural of the immediate recognition of these, rather than stem + ending word and its gender is that provide the machine with instructions). This marked after the ending. The plural is denoted will retain the speed of identification while at by adding -s to the English translation. the same time improving the economy. c) the extra letter(s) cannot be identified with For the second class the unmodified system an ending contained in the dictionary, then the will be retained, and each noun will be accom- incoming noun/adjective is the plural of the panied by its feminine and plural endings, stem word and its gender is that marked after alongside each of which will be the appropriate the stem. (Note: the identification of number translation where different from that of the i n adjectives is, of course, irrelevant for masculine singular. When the feminine form their actual translation, but as was noted above, is merely a lengthened form of the masculine, it can be useful for determining the gender and the two translations will appear with the stem number of nouns with identical forms for the and ending respectively, for example, stem: masculine and feminine or the singular and chien = dog, ending -ne = bitch (dog). When it plural. ) involves a change in the masculine form, the This scheme will provide rapid identification two translations will accompany the two endings, of all regular adjectives and nouns, but it has thus: stem: act- (no translation), endings: the fault of being somewhat uneconomical as -eur = actor, -rice = actress. regards storage space, since the ending dic- In closing the section on nouns and adjectives tionary needs to be of the same size as the the observation made previously with verbs stem dictionary, or even larger, most stems must be repeated, namely that if the stem- having one and often two endings. A possible ending method anywhere involves a confusion of modification offers itself, however, when we two words with different meanings, the stems realize that the number of different endings chosen for these words should be altered, where for regular adjectives and nouns is only 15 possible, to create a distinction. Where it is (5 masculine and 10 feminine) i.e. masculine: not possible, the offending forms will have to -eur, -au, -1, -f, -x; feminine: -euse, be regarded as irregular and entered in the -rice, -11e, -ne, -te, -sse, -e, -ve, -se. dictionary in full. It would seem feasible, therefore, to have an The only other part of speech in the program ending dictionary consisting of only these 14 which requires separate instructions for its endings which would be capable of producing a identification is the pronoun, to be more expli- correct translation of all regular French ad- cit the personal (unstressed) and possessive jectives and nouns. And, in fact, it would be, pronouns. Four difficulties arise, namely but for English often having a completely dif- those of distinguishing the accusative of the 3rd ferent feminine form of a noun where French personal pronoun (le , la, l' , les ) from the de- has only a partially different form (e.g. fils, finite article, the dative plural of the 3rd per- fille -- son, daughter), or an irregular plural sonal pronoun (leur) from the identical posses- where French has a regular one (e. g. homme, sive form, and the nominative nous and vous -s -- man, men). A reasonable solution is to from the accusative/dative, and lastly that of divide all nouns and adjectives into two classes: removing the oblique forms of the personal I. those which have a separate word for mas- pronoun from their French position before the culine and feminine, e.g. cheval (horse) - verb to their normal English position after the jument (mare), those which are without gender verb. The first two difficulties are surmounted in English, e.g. le village (village), abstract by assuming that whenever le, la, l', les are nouns -- all in fact, which do not require a dis- followed by a pronoun or verb, they are forms
Mechanical Translation of French 61 of the personal pronoun not the definite article, Type 3. the sequence is - pronoun + verb. and similarly that whenever leur is followed The procedure is exactly the same for as by a verb it is the personal pronoun. The last Type 2. two are also solved simultaneously in the This, then, is the program which will be following way: used for translating from French mechanically. The objection might well now be raised that it Type 1. the sequence is is devoted only to the grammatical side, the pronounl + pronoun2 + pronoun3 + verb parsing of words, and gives no attention to syn- if nous or vous is the first pronoun, then it is tax. Apart from the fact that an instruction nominative: translate in order - will, as it happens, be included to reverse in pronounl + verb + pronoun2 + pronoun3 translation the order of adjectives following (e.g. nous le leur donnons = we give it them) their nouns in French, this objection is quite valid. In defense we must make the plea made Type 2. the sequence is so often already that the restrictions imposed pronounl + pronoun2 + verb by the present limited storage capacity of the If computer render any comprehensible scheme a) pronounl is a nominative form (je, tu, il, for dealing with syntactical problems academic. elle, ils, elles), pronoun2 is oblique. There- When the storage capacity is enlarged, as it fore translate in order: pronounl + verb + soon will be, there will be the possibility of pronoun2 (e. g. je le donne = I give it) making use of such a scheme. b) pronounl is oblique, pronoun2 is also ob- Nevertheless, it is probably true to say that, lique. Translate in order: verb + pronounl + s ince its word order is more or less identical pronoun2 (e.g. le berger le lui donne = the with that of English, the main difficulty in shepherd gives it him) translating French lies precisely in what we c) pronounl is a form which is not recogni- have endeavored to achieve in the program zable per se as nominative or oblique (i.e. described, namely the successful identification nous or vous), then the verb must be examined of the various forms of words. This being so, to see whether it agrees with nous or vous. the program will prove capable of producing an Thus: 1) if the verb ends in -ons or -mes, adequate translation of such French prose as nous- is nominative has no literary pretensions — for example, 2) if the verb ends in -z or -tes , scientific publications, for which indeed it is vous is nominative primarily intended. Whether this belief is Then translate in order given in a). Otherwise justified or not will be decided by the actual nous and vous are oblique and the translation results of the experiment. These will be pub- order is that of b). lished.