the other restricts it to a part of speech such as lego and scribo. To avoid this ambiguity, some authorities prefer the terms voces, locutiones, dictiones.  Individual words will either be native or imported, simple or compound, literal or metaphorical, in current use or newly-coined. A single word is more likely to be faulty than to possess any intrinsic merit. For though we may speak of a word as appropriate, distinguished or sublime, it can possess none of these properties save in relation to connected and consecutive speech; since when we praise words, we do so because they suit the matter.  There is only one excellence that [p. 81] can be isolated for consideration, namely euphony, the Greek term for our uocalitas: that is to say that, when we are confronted with making a choice between two exact synonyms, we must select that which sounds best.  In the first place barbarisms and solecisms must not be allowed to intrude their offensive presence. These blemishes are however pardoned at times, because we have become accustomed to them or because they have age or authority in their favour or are near akin to positive excellences, since it is often difficult to distinguish such blemishes from figures of speech.1 The teacher therefore, that such slippery customers may not elude detection, must seek to acquire a delicate discrimination; but of this I will speak later when I come to discuss figures of speech.1  For the present I will define barbarism as an offence occurring in connexion with single words. Some of my readers may object that such a topic is beneath the dignity of so ambitious a work. But who does not know that some barbarisms occur in writing, others in speaking? For although what is incorrect in writing will also be incorrect in speech, the converse is not necessarily true, inasmuch as mistakes in writing are caused by addition or omission, substitution or transposition, while mistakes in speaking are due to separation or combination of syllables, to aspiration or other errors of sound.  Trivial as these points may seem, our boys are still at school and I am reminding their instructors of their duty. And if one of our teachers is lacking in education and has done no more than set foot in the outer courts of his art, he will have to confine himself to the rules published in the elementary text-books: the [p. 83] more learned teacher on the other hand will be in a position to go much further: first of all, for example, he will point out that there are many different kinds of barbarism.  One kind is due to race, such as the insertion of a Spanish or African term; for instance the iron tire of a wheel is called cantus,2 though Persius uses it as established in the Latin language; Catullus picked up ploxenum3 (a box) in the valley of the Po, while the author of the in Pollionem, be he Labienus or Cornelius Gallus, imported casamo from Gaul in the sense of “follower.” As for mastruca,4 which is Sardinian for a “rough coat,” it is introduced by Cicero merely as an object of derision.  Another kind of barbarism proceeds from the speaker's temper: for instance, we regard it as barbarous if a speaker use cruel or brutal language.  A third and very common kind, of which anyone may fashion examples for himself, consists in the addition or omission of a letter or syllable, or in the substitution of one for another or in placing one where it has no right to be.  Some teachers however, to display their learning, are in the habit of picking out examples of barbarism from the poets and attacking the authors whom they are expounding for using such words. A boy should however realize that in poets such peculiarities are pardonable or even praiseworthy, and should therefore be taught less common instances.  For Tinga of Placentia, if we may believe Hortensius who takes him to task for it, committed two barbarisms in one word by saying precula for pergula: that is to say he substituted c for g, and transposed r and e. On the other hand [p. 85] when Ennius writes Mettocoque Fufetioeo,5 where the barbarism is twice repeated, he is defended on the plea of poetic licence.  Substitution is however sometimes admitted even in prose, as for instance when Cicero speaks of the army of Canopus which is locally styled Canobus, while the number of authors who have been guilty of transposition in writing Trasumennus for Tarsumennus has succeeded in standardising the error. Similar instances may be quoted. If adsentior be regarded as the correct form, we must remember that Sisenna said adsentio, and that many have followed him on the ground of analogy: on the other hand, if adsentio is the correct form, we must remember that adsentior has the support of current usage.  And yet our fat fool, the fashionable schoolmaster, will regard one of these forms as an example of omission or the other as an instance of addition. Again there are words which when used separately are undoubtedly incorrect, but when used in conjunction excite no unfavourable comment.  For instance dua and tre are barbarisms and differ in gender, but the words duapondo and trepondo6 have persisted in common parlance down to our own day, and Messala shows that the practice is correct.  It may perhaps seem absurd to say that a barbarism, which is an error in a single word, may be made, like a solecism, by errors in connexion with number or gender. But take on the one hand scala (stairs) and scopa (which literally means a twig, but is used in the sense of broom) and on the other hand hordea (barley) and mulsa (mead): here we have substitution, omission and addition of letters, but the blemish consists in the former case merely in the use of singular for plural, [p. 87] in the latter of plural for singular. Those on the other hand who have used the word gladia are guilty of a mistake in gender.  I merely mention these as instances: I do not wish anyone to think that I have added a fresh problem to a subject into which the obstinacy of pedants has already introduced confusion. The faults which arise in the course of actual speaking require greater penetration on the part of the critic, since it is impossible to cite examples from writing, except in cases where they occur in poetry, as when the diphthong is divided into two syllables in Europai and Asiai7; or when the opposite fault occurs, called synaeresis or synaloephe by the Greeks and complexio by ourselves: as an example I may quote the line of Publius Varro:
Once supply the thought,
And words will follow swift as soon as soughtArs Poetica, 311.
turn te flagranti deiectum fulmine Plaethon.8 If this were prose, it would be possible to give the letters their true syllabic value. I may mention as further anomalies peculiar to poetry the lengthening of a short syllable as in Italiam fato profugus,9 or the shortening of a long such as unĭius ob noxam et furias;10 but in poetry we cannot label these as actual faults.  Errors in sound on the other hand can be detected by the ear alone; although in Latin, as regards the addition or omission of the aspirate, the question may be raised whether this is an error when it occurs in writing; for there is some doubt whether h is a letter or merely a breathing, practice having frequently varied in different ages.  Older authors used it but rarely even before vowels, saying aedus or ircus, while its conjunction with consonants was for a long time avoided, as in words such as [p. 89] Graccus or triumpus. Then for a short time it broke out into excessive use, witness such spelling as chorona, chenturia or praecho, which may still be read in certain inscriptions: the well-known epigram of Catullus11 will be remembered in this connexion.  The spellings vehementer, comprehendere and mihi have lasted to our own day: and among early writers, especially of tragedy, we actually find mehe for me in the older MSS.  It is still more difficult to detect errors of tenor or tone (I note that old writers spell the word tonor, as derived from the Greek τόνος), or of accent, styled prosody by the Greeks, such as the substitution of the acute accent for the grave or the grave for the acute: such an example would be the placing of the acute accent on the first syllable of Camillus,  or the substitution of the grave for the circumflex in Cethegus, an error which results in the alteration of the quantity of the middle syllable, since it means making the first syllable acute; or again the substitution of the circumflex for the grave on the second syllable of Appi, where the contraction of two syllables into one circumflexed syllable involves a double error.  This, however, occurs far more frequently in Greek words such as Atrei, which in our young days was pronounced by the most learned of our elders with an acute accent on the first syllable, necessitating a grave accent on the second; the same remark applies to Nerei and Terei. Such has been the tradition as regards accents.12 [p. 91]  Still I am well aware that certain learned men and some professed teachers of literature, to ensure that certain words may be kept distinct, sometimes place an acute accent on the last syllable, both when they are teaching and in ordinary speech: as, for instance, in the following passage:
 where they make the last syllable of circum acute on the ground that, if that syllable were given the grave accent, it might be thought that they meant circus not circuitus.13 Similarly when quale is interrogative, they give the final syllable a grave accent, but when using it in a comparison, make it acute. This practice, however, they restrict almost entirely to adverbs and pronouns; in other cases they follow the old usage.  Personally I think that in such phrases as these the circumstances are almost entirely altered by the fact that we join two words together. For when I say circum litora I pronounce the phrase as one word, concealing the fact that it is composed of two, consequently it contains but one acute accent, as though it were a single word. The same thing occurs in the phrase Troiae qui primus ab oris.14  It sometimes happens that the accent is altered by the metre as in pecudes pictaeque volucres15; for I shall read volucres with the acute on the middle syllable, because, although that syllable is short by nature, it is long by position: else the last two syllables would form an iambus, which its position in the hexameter does not allow.  But these same words, if separated, will form no exception to the rule: or if the custom under discussion prevails, the old law [p. 93] of the language will disappear. (This law is more difficult for the Greeks to observe, because they have several dialects, as they call them, and what is wrong in one may be right in another.) But with us the rule is simplicity itself.  For in every word the acute accent is restricted to three syllables, whether these be the only syllables in the word or the three last, and will fall either on the penultimate or the antepenultimate. The middle of the three syllables of which I speak will be acute or circumflexed, if long, while if it be short, it will have a grave accent and the acute will be thrown back to the preceding syllable, that is to say the antepenultimate.  Every word has an acute accent, but never more than one. Further the acute never falls on the last syllable and therefore in dissyllabic words marks the first syllable. Moreover the acute accent and the circumflex are never found in one and the same word, since the circumflex itself contains an acute accent. Neither the circumflex nor the acute, therefore, will ever be found in the last syllable of a Latin word, with this exception, that monosyllables must either be acute or circumflexed; otherwise we should find words without an acute accent at all.  There are also faults of sound, which we cannot reproduce in writing, as they spring from defects of the voice and tongue. The Greeks who are happier in inventing names than we are call them iotacisms, lambdacisms,16 ἰσχνότητες (attenuations) and πλατειασμοί (broadenings); they also use the term κοιλοστομία, when the voice seems to proceed from the depths of the mouth.  There are also certain peculiar and indescribable sounds for which we sometimes take whole nations to fault. To sum up then, if all the faults of which we have just spoken be avoided, [p. 95] we shall be in possession of the Greek ὀρθοέπεια, that is to say, an exact and pleasing articulation; for that is what we mean when we speak of correct pronunciation.  All other faults in speaking are concerned with more words than one; among this class of faults is the solecism, although there have been controversies about this as well. For even those who acknowledge that it occurs in connected speech, argue that, since it can be corrected by the alteration of one word, the fault lies in the word and not in the phrase or sentence.  For example whether amarae corticis17 or medio cortice18 contains a solecism in gender (and personally I object to neither, as Vergil is the author of both; however, for the sake of argument let us assume that one of the two is incorrect), still whichever phrase is incorrect, it can be set right by the alteration of the word in which the fault lies: that is to say we can emend either to amari corticis or media cortice. But it is obvious that these critics misrepresent the case. For neither word is faulty in itself; the error arises from its association with another word. The fault therefore lies in the phrase.  Those who raise the question as to whether a solecism can arise in a single word show greater intelligence. Is it for instance a solecism if a man when calling a single person to him says uenite, or in dismissing several persons says abi or discede? Or again if the answer does not correspond to the question: suppose, for example, when someone said to you “Whom do I see?”, you were to reply “I.” Some too think it a solecism if the spoken word is contradicted by the motion of hand or head.  I do not entirely concur with this view nor yet do I [p. 97] wholly dissent. I admit that a solecism may occur in a single word, but with this proviso: there must be something else equivalent to another word, to which the word, in which the error lies, can be referred, so that the solecism arises from the faulty connexion of those symbols by which facts are expressed and purpose indicated.  To avoid all suspicion of quibbling, I will say that a solecism may occur in one word, but never in a word in isolation. There is, however, some controversy as to the number and nature of the different kinds of solecism. Those who have dealt with the subject most fully make a fourfold division, identical with that which is made in the case of barbarisms: solecisms are brought about by addition, for instance in phrases such as nam enim, de susum, in Alexandriam;  by omission, in phrases such as ambulo viam, Aegypto venio, or ne hoc fecit: and by transposition as in quoque ego, enim hoc voluit, aulem non habuit.19 Under this last head comes the question whether igitur can be placed first in a sentence: for I note that authors of the first rank disagree on this point, some of them frequently placing it in that position, others never.  Some distinguish these three classes of error from the solecism, styling addition a pleonasm, omission an ellipse, and transposition anastrophe: and they assert that if anastrophe is a solecism, hyperbaton might also be so called.  About substitution, that is when one word is used instead of another, there is no dispute. It is an error which we may detect in connexion with all the parts of speech, but most frequently in the verb, because it has greater variety [p. 99] than any other: consequently in connexion with the verb we get solecisms of gender, tense, person and mood (or “states” or “qualities” if you prefer either of these terms), be these types of error six in number, as some assert, or eight as is insisted by others (for the number of the forms of solecism will depend on the number of subdivisions which you assign to the parts of speech of which we have just spoken). Further there are solecisms of number;  now Latin has two numbers, singular and plural, while Greek possesses a third, namely the dual. There have however been some who have given us a dual as well in words such as scripsere and legere, in which as a matter of fact the final syllable has been softened to avoid harshness, just as in old writers we find male merere for male mereris. Consequently what they assert to be a dual is concerned solely with this one class of termination, whereas in Greek it is found throughout the whole structure of the verb and in nouns as well, though even then it is but rarely used.  But we find not a trace of such a usage in any Latin author. On the contrary phrases such as devenere locos,20 conticuere omnes21 and consedere duces22 clearly prove that they have nothing to do with the dual. Moreover dixere,23 although Antonius Rufus cites it as proof to the contrary, is often used by the usher in the courts to denote more than two advocates.  Again, does not Livy near the beginning of his first book write tenuere arcem Sabini24 and later in adversum Romani subiere? But I can produce still better authority. For Cicero in his Orator says, “I have no objection [p. 101] to the form scripsere, though I regard scripserunt as the more correct.”25  Similarly in vocables and nouns solecisms occur in connexion with gender, number and more especially case, by substitution of one for another. To these may be added solecisms in the use of comparatives and superlatives, or the employment of patronymics instead of possessives and vice versa.  As for solecisms connected with expressions of quantity, there are some who will regard phrases such as magnum peculiolum26 as a solecism, because the diminutive is used instead of the ordinary noun, which implies no diminution. I think I should call it a misuse of the diminutive rather than a solecism; for it is an error of sense, whereas solecisms are not errors of sense, but rather faulty combinations of words.  As regards participles, solecisms occur in case and gender as with nouns, in tense as with verbs, and in number as in both. The pronoun admits of solecisms in gender, number and case.  Solecisms also occur with great frequency in connexion with parts of speech: but a bare statement on this point is not sufficient, as it may lead a boy to think that such error consists only in the substitution of one part of speech for another, as for instance if a verb is placed where we require a noun, or an adverb takes the place of a pronoun and so on.  For there are some nouns which are cognate, that is to say of the same genus, and he who uses the wrong species27 in connexion with one of these will be guilty of the same offence as if he were to change the genus. Thus an and aut are conjunctions, but it would be bad Latin to say in a question hic and ille sit28;  ne and [p. 103] non are adverbs: but he who says non feceris in lieu of ne feceris, is guilty of a similar mistake, since one negative denies, while the other forbids. Further intro and intus are adverbs of place, but eo intus and intro sum are solecisms.  Similar errors may be committed in connexion with the various kinds of pronouns, interjections and prepositions. It is also a solecism29 if there is a disagreement between what precedes and what follows within the limits of a single clause.  Some phrases have all the appearance of a solecism and yet cannot be called faulty; take for instance phrases such as tragoedia Thyestes or ludi Floralia and Megalensia30: although these are never found in later times, they are the rule in ancient writers. We will therefore style them figures and, though their use is more frequent in poets, will not deny their employment even to orators.  Figures however will generally have some justification, as I shall show in a later portion of this work, which I promised you a little while back.31 I must however point out that a figure, if used unwittingly, will be a solecism.  In the same class, though they cannot be called figures, come errors such as the use of masculine names with a female termination and feminine names with a neuter termination. I have said enough about solecisms; for I did not set out to write a treatise on grammar, but was unwilling to slight the science by passing it by without salutation, when it met me in the course of my journey.  I therefore resume the path which I prescribed for myself and point out that words are either [p. 105] native or foreign. Foreign words, like our population and our institutions, have come to us from practically every nation upon earth.  I pass by words of Tuscan, Sabine and Praenestine origin; for though Lucilius attacks Vettius for using them, and Pollio reproves Livy for his lapses into the dialect of Padua, I may be allowed to regard all such words as of native origin. Many Gallic words have become current coin,  such as raeda (chariot) and petorritim (four-wheeled wagon) of which Cicero uses the former and Horace the latter. Mappa (napkin) again, a word familiar in connexion with the circus, is claimed by the Carthaginians, while I have heard that gurdus, which is colloquially used in the sense of “stupid,” is derived from Spain.  But this distinction between native and foreign words has reference chiefly to Greek. For Latin is largely derived from that language, and we use words which are admittedly Greek to express things for which we have no Latin equivalent. Similiarly they at tines borrow words from us. In this connexion the problem arises whether foreign words should be declined according to their language or our own.  If you come across an archaistic grammarian, he will insist on absolute conformity to Latin practice, because, since we have an ablative and the Greeks have not, it would be absurd in declining a word to use five Greek cases and one Latin.  He will also praise the patriotism of those who aimed at strengthening the Latin language and asserted that we had no need of foreign practices. They, therefore, pronounced Castorem with the second syllable long to bring it into conformity with all those Latin nouns which have the same termination in the nominative as [p. 107] Castor. They also insisted on the forms Palaemo, Telamo, and Plato (the last being adopted by Cicero), because they could not find any Latin nouns ending in -on.  They were reluctant even to permit masculine Greek nouns to end in -as in the nominative case, and consequently in Caelius we find Pelia cincinnatus and in Messala bene fecit Euthia, and in Cicero Hermagora.32 So we need not be surprised that the majority of early writers said Aenea and Anchisa.  For, it was urged, if such words are spelt like Maecenas, Sufenas and Asprenas, the genitive should terminate in -is not in -e. On the same principle they placed an acute accent on the middle syllable of Olympus and tyrannus, because Latin does not allow an acute accent on the first syllable if it is short and is followed by two long syllables.  So too we get the Latinised genitives Ulixi and Achilli together with many other analogous forms. More recent scholars have instituted the practice of giving Greek nouns their Greek declension, although this is not always possible. Personally I prefer to follow the Latin method, so far as grace of diction will permit. For I should not like to say Calypsonem on the analogy of Iunonem, although Gaius Caesar in deference to antiquity does adopt this way of declining it. Current practice has however prevailed over his authority.  In other words which can be declined in either way without impropriety, those who prefer it can employ the Greek form: they will not be speaking Latin, but will not on the other hand deserve censure. Simple words are what they are in the nominative, that is, their essential nature.  Compound [p. 109] words are formed by the prefix of a preposition as in innocens, though care must be taken that two conflicting prepositions are not prefixed as in imperterritus:33 if this be avoided they may in certain cases have a double prefix as in incompositus or reconditus or the Ciceronian subabsurdtim. They may also be formed by what I might term the combination of two independent units, as in maleficus.  For I will not admit that the combination of three is possible at any rate in Latin, although Cicero asserts that capsis34 is compounded of cape si vis, and there are to be found scholars who contend that Lupercalia likewise is a compound of three parts of speech, namely luere per caprum.  As for Solitaurilia it is by now universally believed to stand for Suovelaurilia, a derivation which corresponds to the actual sacrifice, which has its counterpart in Homer35 as well. But these compounds are formed not so much from three words as from the fragments of three. On the other hand Pacuvius seems to have formed compounds of a preposition and two vocables (i.e. nouns) as in
quae circus litora, circum piscosos scopulos,Aen. iv. 254.
Nerei repandirostrum incurvticervicum pecs:
The flockthe effect is unpleasing.  Compounds are however formed from two complete Latin words, as for instance supefui and subterfui; though in this case there is some question as to whether the words from which they are formed are complete.36 They may also be formed of one complete and one incomplete [p. 111] word, as in the case of malevolus, or of one incomplete and one complete, such as noctivagus, or of two incomplete words as in pedisecus (footman), or from one Latin and one foreign word as in biclinium (a dining-couch for two), or in the reverse order as in epitogium (an upper garment) or Anticato, and sometimes even from two foreign words as in epiraedium (a thong attaching the horse to the raeda). For in this last case the preposition is Greek, while raeda is Gallic, while the compound is employed neither by Greek nor Gaul, but has been appropriated by Rome from the two foreign tongues.  In the case of prepositions they are frequently changed by the act of compounding: as a result we get abstulit, aufugit, amisit, though the preposition is ab, and coil, though the preposition is con. The same is true of ignauus and erepublica.37 But compounds are better suited to Greek than to Latin,  though I do not think that this is due to the nature of our language: the reason rather is that we have a preference for foreign goods, and therefore receive κυρταύχην with applause, whereas we can scarce defend incurvicervicus from derisive laughter. Words are proper when they bear their original meaning;  metaphorical, when they are used in a sense different from their natural meaning. Current words are safest to use: there is a spice of danger in coining new. For if they are adopted, our style wins but small glory from them; while if they are rejected, they become a subject for jest.  Still we must make the venture; for as Cicero38 says, use softens even these words which at first seemed harsh. On the other hand the power of onomatopoeia is denied us. Who would tolerate an attempt to imitate [p. 113] phrases like the much praised λίγξε βιός,39 “the bow twanged,” and σῖζεν ὀφθαλμός40 “the eye hissed”? We should even feel some qualms about using balare “to baa,” and hinntre, “to whinny,” if we had not the sanction of antiquity to support us.
Of Nereus snout-uplifted, neck-inarched