IDIOM refers to a way of speaking that is outside ordinary usage and the general rules of language and adapted to the unique character of a particular tongue. R. ίδιος, peculiaris, propre, particulier . This is a general term that can be used with regard to all languages; a Greek, Latin, French idiom , etc. In many cases it is the only term that can be used; we can speak only of a Spanish, Portuguese, Turkish, etc. idiom . But with regard to several languages we have specific terms subordinate to that of idiom , and we say Anglicism, Arabism, Celticism, Gallicism, Germanism, Hebraism, Hellenism, Latinism , etc.

When I say that an idiom is a way of speaking adapted to the specific character of a particular language, it is in order to make clear that it is rather a distinctive effect of the characteristic genius of this language than a locution that cannot be translated into any other language, as it is usually defined. The riches of one language can pass easily into another with which it has some affinity; and all languages have more or fewer of them [affinities], depending on the degrees of connection there are between the peoples who speak them or with whom they have spoken them. Even if Italian, Spanish, and French are grafted from the same language stem, these three languages obviously each have their own idioms , because they are different languages; but it would be difficult for all three of them not to have adopted some idioms from the language that is their common source, and it would not be surprising to find some Celticisms in all three. It would be no more amazing to find some idioms of one of these three in another of them, because of the connections of proximity, political interests, commerce, religion, which have subsisted for a long time among the peoples who speak them; as one is not surprised to find Arabisms in Spanish, when one knows the history of the long domination of Spain by the Arabs. Everybody knows that the best Latin authors are full of Hellenisms : and if all writers agree that it is easier to translate from Greek into French than from Latin, it is because the character of our language is closer to that of the Greek language than to that of the Latin language, and because our language is practically one continuous Hellenism .

But a remarkable proof of the communicability of languages that seem to have less affinity among them, is that even in French we Hebraicize. A well-known Hebraism is the repetition of an adjective or an adverb that one wants to elevate to what is commonly called a superlative . See Amen and Superlative. And the most energetic superlative is indicated in Hebrew by the triple repetition of the word: from this comes the kirie eleison that we sing in our churches to give more force to our invocation; and the triple sanctus to express better the profound adoration of the heavenly spirits. Now, it is plausible that our très [very], formed from the Latin tres [three], was only introduced into our language as the symbol of this triple repetition, très-saint [very holy], ter sanctus , or sanctus , sanctus , sanctus : and our usage of linking très to the positive form of a word with a hyphen is no doubt based on the intention of making it clear that this addition is purely material, that it does not interfere with the unity of the word, but that it must be repeated three times, or at least that one should attach to it the meaning that it would have if it were repeated three times; and, in fact, the adverbs bien [well] and fort [strongly], which express in themselves the superlative sense that is at issue here, are never connected to the positive form of a word to which they are joined to communicate it. We encounter Hebraisms of another sort in common speech: a man of God, God’s wine, God’s harvest , to mean a very upstanding man, very fine wine, an abundant harvest ; or, in giving the same meaning throughout in the same way, a perfect man, a perfect wine, a perfect harvest — the Hebrews indicating perfection with the name of God, who is the model and the source of all perfection. It is this kind of Hebraism which is found in Psalm 35, verse 7, justitia tua sicut montes Dei , and in Psalm 64, verse 10, flumen Dei , for flumen maximum . [1]

Despite the Hellenisms that are recognizable in Latin, people have believed quite easily that idioms were unique and untranslatable locutions, and as a consequence false or suspect ideas about them have been held and given, and many people continue to believe that by this general term, or by any of the analogous specific terms, is meant only vicious locutions clumsily imitated from some other language. See Gallicism. This is an error that I believe is sufficiently destroyed by the observations that I have laid out for the reader. I will move on to another one that is even more universal and that is no less contrary to the true idea of idioms .

It is commonly said that these are ways of speaking contrary to the general laws of Grammar. There are in fact idioms that are like this, and as they are for that very reason the most striking and the easiest to recognize, it is has been easy to believe that this opposition to the immutable laws of Grammar was the common feature of all of them. But there is another sort of idioms that are ways of speaking that are simply different from ordinary usage, but which conform in every way that might be wished to the fundamental principles of the general Grammar. We can call these regular idioms because they follow the immutable rules of speech, and they do not violate anything but arbitrary and common institutions; the others, by contrast, can be called irregular idioms , because they violate the immutable rules of speech. These two types fall under the definition that I gave at the top, and I will shortly make them clear through examples, but by applying to them the principles that it is appropriate to follow to get at their meaning and to discover, if possible, the unique character of the language from which they originated.

I. Regular idioms require only to be explained literally in order then to be brought back to the turn of phrase in the natural language that one speaks.

I find, for example, that Germans say diese gelehrten manner , as in Latin, hi docti viri , or in French, ces hommes savans [these learned men], and the adjective gelehrten agrees in every way with the noun manner , as the Latin adjective docti does with the noun viri , or the French adjective savans [learned] with the noun hommes [men]. Thus in this case the Germans follow both the general laws and common usage. But they say diese manner sind gelehrt and to render this literally in Latin one must say hi viri sunt doctѐ and in French, ces hommes sont savamment [these men are wisely], which means unquestionably these men are learned . Gelehrt is thus an adverb and one must recognize here that the Germans stray from common usage, which prefers the adjective in such a case. It is thus clear what a Germanism consists of when it is a question of expressing an attribute, but what might be the cause of this idiom ? The verb expresses the existence of the subject with an attribute. See Verb. The attribute is simply a particular way of being, and it is up to adverbs to express simply the ways of being and consequently the attributes: voilà, the character of German. But how can we reconcile this reasoning with the almost universal usage of expressing the attribute with an adjective that agrees with the subject of the verb? I respond that perhaps the only difference between the common way and the German way is what there would be between two paintings in which two different moments of the same action have been captured: the germanism captures the instant that immediately precedes the act of judgment, when the mind is still considering the attribute in a vague way and without applying it to the subject; the common phrase presents the subject as it appears to the mind after judgment and when there is no longer any abstraction. The German must thus express the attribute with the appearance of independence, and this is what is done with the adverb, which has no ending and thus cannot designate agreement with any determinate subject. The other languages must express the attribute with the characteristics of the application, which is fulfilled by the agreement of the attributive adjective with the subject. But perhaps then we should say that the noun is implied before the adjective and that hi viri sunt docti is the same as hi viri sunt viri docti and ego sum miser is the same as ego sum homo miser . In fact, the agreement of the adjective with the noun and the identity of the subject expressed by the types are not understood clearly or in a satisfying manner except in the case of apposition, and the apposition cannot occur here except by means of the ellipsis. From all this I will draw a surprising conclusion: The German phrase is thus a regular idiom and the common phrase is an irregular idiom .

Here is a regular Latinism whose development can again lead to useful insights: neminem reperire est id qui velit . There are four words here that are not at all problematic: qui velit id (who would wish for that) is an incidental determinative proposition of the antecedent neminem ; neminem (no one) is the complement or the objective grammatical regime of the verb reperire ; neminem qui velit id (to find no one who would wish for that) is an exact and regular construction. But what to do with the word est ? It is in the third person singular; what is its subject? How can the infinitive reperire agree with the words dependent on it? Let us consult other phrases that are clearer whose solution might guide us.

In Horace (III. Odes. 2 ) we find dulce et decorum est pro patriâ mori ; and again (IV. Odes. 12) dulce et desipere in loco . [2] Now, the construction is simple: mori pro patriâ est dulce et decorum; desipere in loco est dulce : the infinitives mori [to die] and desipere [to play] are treated as nouns, and one can consider them as such. I find another even stronger proof in Persius, Satires. I. scire tuum nihil est . [3] The adjective tuum [your] put in agreement with scire [to know], indicates well that scire is considered to be a noun. Thus the difficulty raised by our first phrase is removed: the verb reperire is what one commonly calls the nominative of the verb est ; or, more accurately, it is the grammatical subject of it, which would be the nominative if it were declinable. Reperire neminen qui velit id is thus its logical subject. The proper construction is thus reperire neminem qui velit id, est , which signifies literally, to find no one who wishes it, is or exists ; or, transposing the negation, to find someone who wishes it, is not or does not exist ; or, finally, in returning the same idea to our way of speaking, there is no one who wishes it.

It is the same syntax and the same construction throughout where there is an infinitive used as the subject of the verb sum , when the verb has the meaning of an adjective, that is, when it is not simply a substantive verb but incorporates the idea of concrete existence as an attribute, and consequently it is equivalent to existo. Only in this case is there a Latinism , since there is nothing as common in the majority of languages as seeing the infinitive as a subject of the substantive verb when one then expresses a determinate attribute. Thus, one says in Latin turpe est mentiri , and in French mentir est une chose honteuse [to lie is a shameful thing]. But we cannot say voir est [to see is] for on voit [one sees], voir étoit [to see was] for on voyoit [one saw], voir sera [to see will be], for on verra [one will see], as the Latins say videre est, videre erat, videre erit . The infinitive considered as a noun also serves to explain another type of Latinism that it seems to me has not yet been understood as it should be, and for which the ridiculous and insoluble difficulties of the redoubtable entrenched que have been substituted for the rudiments of an explanation . See Infinitive.

II. In regards to irregular idioms , in order to penetrate the meaning, it is necessary to discern with care the type of deviation which determines them, and to return, if it is possible, to the cause that gave rise to them: this is also the only possible means of recognizing the precise nature of the particular character of a language, since this character consists simply in bringing together the views that it lays out and the means that it has authorized.

To discern exactly the type of deviation that determines an irregular idiom recall what we said in the article Grammar: that all the fundamental rules of this science can be reduced to two leading principles, Lexicology and Syntax. Lexicology concerns everything that has to do with the knowledge of words considered in themselves and outside of speech. Thus, in each language the vocabulary is like the inventory of subjects in its domain, and its main job is to fix well the proper meaning of each word authorized in this idiom. Syntax concerns everything having to do with the organization of words brought together in speech, and in all languages its decisions are related to three general points: concordance, system, and construction.

If the particular usage of a language authorizes the alteration of the true meaning of some words and the substitution of a foreign meaning, it is then a figure of speech which is called a trope. See this article.

If usage authorizes a locution contrary to the general laws of Syntax, this is then a figure that is commonly called a figure of construction , but which I would rather call by the more general term, figure of Syntax , reserving the term figure of construction only for those locutions that deviate from the rules of construction properly speaking. See Figure and Construction. Here are two types of deviation that can be observed in irregular idioms .

When a trope is so particular to one language that it cannot be rendered literally in another one, or when it is rendered literally it has a completely different meaning, it is an idiom of the original language that adopted it, and this idiom is irregular because the proper meaning of the words has been abandoned, which is contrary to the initial introduction of words. Thus the superstitious euphemism which, in the Latin language has given the meaning of sacrifice to the verb mactare , whereas etymologically this word means to increase more (magis auctare). This euphemism, I say, is so particular to the character of this language that the literal translation one would make into another one could never give rise to the idea of sacrifice . See Euphemism.

Similarly, it is a trope that introduced into our language those idioms already discussed in the article Gallicism, in which we employ the two verbs venir [come] and aller [go] to express, in the first case imminent preterits and, in the second, imminent futures ( see Time); as when one says, je viens de lire, je venois de lire [I came to read] for I have or I had just read a little while ago ; je vais lire, j’allois lire [I am going to read, I will read], for I must , or I must read in a little while . The two auxiliary verbs come and go then lose their original meaning and no longer indicate the shift from one place to another; they serve only to indicate the proximity of anteriority or posteriority, and our phrases rendered literally in any other language would either be meaningless or have a meaning different from ours. This is a catachresis introduced by necessity ( see Catachresis), and based nevertheless on an analogy between the proper meaning and the figurative meaning. The verb to come , for example, supposes a future existence in the place to which one comes, and at the moment when one comes there, that one was there not long before: this is precisely the reason for the choice of this verb for the expression of imminent preterits. Similarly, the verb to go indicates later existence in the place to which one goes, and at the time that one goes there one has the intention of being there soon. Here again is the justification of the preference given to this verb to designate imminent futures. But it remains no less true that these verbs, having become auxiliary, really lose their primitive and fundamental meaning, and that they retain from them only ancillary and distant ideas.

What we have just said about tropes is equally true of figures of Syntax: such a figure is an irregular idiom because it cannot be rendered literally in another language or because the literal version that would be made would have another meaning. Thus our usage, in the French language, of employing the masculine possessive adjective, mon, ton, son [my, your, your], when it precedes a feminine noun that begins with a vowel or with an h , is an irregular idiom in our language, a Gallicism , because the literal imitation of this figure in another language would be nothing but a solecism. We say mon ame [my soul], and one would not say meus anima ; ton opinion [your opinion], and one cannot say tuus opinio . This is because to avoid the hiatus caused by the merging of the vowels the Latins have other mechanisms that we are prohibited from using due to the constitution of our language and which it would be more reasonable to use than to violate a law as essential as the [gender] agreement that we transgress; they could say anima mea, opinio tua , and we cannot imitate this construction and say ame ma, opinion ta [opinion my, opinion your]. Our language thus sacrifices here a reasonable principle to the pleasures of euphony ( see Euphony), in conformity with Cicero’s sensible remark in Orations. N. 47: impetratum est à consuetudine ut peccare, suavitatis causâ, liceret . [4]

Here is an ellipsis that has become a locution specific to our language, a Gallicism , because its usage has prevailed to the point that it is no longer permitted to follow the full Syntax in such a case: il ne laisse pas d’agir [he does not bother to act] , notre langue ne laisse pas de se prêter à tous les genres d’écrire [our language does not allow for the possibility of being lent to all types of writing], on ne laisse pas d’abandonner la vertu en la louant [virtue cannot be abandoned in praising it]; that is, he does not make the effort to act , our language does not have the ability to lend itself to all types of writing, we do not have the weakness to abandon virtue in praising it . We prefer in these phrases the merit of brevity to the full locution, which without being any clearer, would have the undeniable unpleasantness of superfluous length.

If it is easy to place the deviations that determine the different idioms under a fixed number of leading principles, the same cannot be said for the particular views that can influence them: the variety of causes is too great, the influence on them is too delicate, the complications are sometimes too numerous to be able to establish in this matter anything very certain. But it is no less constant that all of them, more or less, inhere to the character of the different languages, that they emanate from them, and that they can become indexes of them. “This is true of entire peoples as it is of a single individual,” says du Tremblay, Traité des langues, chapter 22 . [5] “Their language is the living expression of their customs, their character, and their inclinations; and one need only examine this language closely to penetrate all the thoughts of their soul and all the movements of their heart. Each language must thus necessarily hold the perfections and the faults of the people who speak it.” “Each one of them,” he had said a bit earlier, “will have some perfection that will not be found among the others, because each one of them holds all the customs and the character of the peoples who speak it: each will have terms and ways of speaking that are specific to them and will constitute as it were its unique character.” In fact, we recognize the oriental phlegmatic character in the repetition of the adjective or the adverb ( amen, amen; sanctus, sanctus, sanctus ); French vivacity cannot accommodate it, and très-saint [very holy] is much more to its taste than saint, saint, saint [holy, holy, holy].

But if we wanted to disentangle from the regular and irregular idioms what the specific character of the language may have contributed to them, the first essential thing to do would be to establish a good literal interpretation. This assumes two things: the rigorous translation of each word in its true meaning, and the reduction of the full phrase to the fullness of the analytic construction, which alone can fill gaps in an ellipsis, correct the redundancies of a pleonasm [tautology], counter the deviations of an inversion, and bring the whole back into the immutable system of the general Grammar.

“I am well aware,” says M. du Marsais, Methode pour apprendre la langue latine, page 14 , “that this literal translation causes a lot of trouble for those who do not understand the point of it; they do not see that the goal that is suggested for this type of translation is simply to show how Latin is spoken, which can only be done by explaining each Latin word by means of the French word that corresponds to it.

“In the first years of our childhood, we connect certain ideas to certain impressions; habit confirms this connection. Animal instincts take a predetermined path for each discrete idea, such that when later one wishes to stimulate the same idea in a different way, a contrary movement is activated in the brain to the one to which it is accustomed, and this movement stimulates either surprise or laughter, and sometimes even pain. This is why every people finds the costume or the language of other peoples extraordinary. In Florence they laugh at the way that a Frenchman pronounces Latin or Italian, in Paris they mock the pronunciation of the Florentine. By the same token the majority of those who understand the translation of pater ejus to be le pere de lui [the father of him], rather than son pere [his father], are at first inclined to make fun of the translation.

“However, as the most efficient way of making the way foreigners dress understood is to show their clothes as they are and not to dress them in the French style; similarly, the best method for understanding foreign languages is to learn their unique turns of phrase, which can only be done by means of a literal translation.

“To conclude, there is no reason to fear that this type of explanation would teach someone to speak French badly.

“1. The clearer and more precise one’s mind, the better one writes and the better one speaks. Now, there is nothing more appropriate for giving young people clarity and precision of mind than practicing literal translation because it demands precision, accuracy of terms, and a certain meticulousness that prevents the mind from straying into strange ideas.

“2. Literal translation makes one feel the difference between the two languages. The more the Latin turn of phrase is removed from the French one, the less one has to fear that it will be imitated in discourse. It makes the character of the Latin language known; after that, usage, better than the teacher, teaches the French way of speaking.”  [6]

1. “Thy justice stands firm as the hills of God,” “God’s river,” “Greatest river.” Current versions of the Bible number these psalms as 36 and 65. A recent translation into English provides the following note for the first quotation, which it translates as “Thy justice stands firm as the everlasting hills:” “Literally ‘the hills of God,’ a Hebrew form of speech for what is magnificent in nature.” The Knox Bible.

2. The first phrase, from Book 3, stanza 2, is perhaps the most famous line from Horace’s Odes: “It’s sweet and fitting to die for one’s country.” Less familiar is the second phrase, from Book 4, stanza 12: “it’s sweet sometimes to play the fool.” Translations by A. S. Kline from Poetry in Translation.

3. “Is [your] knowledge nothing?” Translation by A. S. Kline from Poetry in Translation. The author’s full name is Aulus Persius Flaccus (34-62).

4. Cicero, Orator 157. "It has been established by habit that one can err in order to achieve pleasantness." Cicero, Orator 157. See Beauzée’s article Grammar, translated by Julia Wallhager, and her note 1.

5. Jean Frain du Tremblay, Traité des langues ou l’on donne es principes et des règles pour juger du mérite et de l’éxcellence de chaque langue, et en particulier de la langue françoise (Paris, 1703).

6. César Chesnau Dumarsais, Méthode raisonnée pour apprendre la langue latine (1722), in Œuvres complettes (Paris, 1797) 1 :1-41. The quotation can be found on pp. 11-14 of this edition.