1. Many have raised the question whether in the inflections of words the art of speaking ought to follow the principle of unlikeness or that of likeness. This is important, since from these develop the two systems of relationship: that which develops from likeness is called Regularity, and its counterpart is called Anomaly. Of this, in the first book, I gave the arguments which are advanced in favour of considering unlikeness as the proper guide; in the second, those advanced to show that it is proper rather to prefer likeness. Therefore, as their foundations have not been laid by anyone, as should have been done, nor have their order and nature been set forth as the matter demands, I shall myself sketch an outline of the subject.
2. I shall speak of four factors which limit the inflections of words: what likeness and unlikeness are; what the relationship is which they call logos; what “by comparative likeness” is, which they call “according to logos”; what usage is. The explanation of these matters will make clear the problems connected with Regularity and Anomaly: whence they come, what they are, of what sort they are.
3. The first topic to be discussed must be likeness and unlikeness, because this matter is the foundation of all inflections and set limits to the relationship of words. That is like which is seen to have several features identical with those of that which is like it, in each case: that is unlike, which is seen to be the opposite of what has just been said. Every like or unlike consists of two units at least, because nothing can be like without being like something else, and nothing can be unlike without association with something to which it is unlike.
4. Thus a human being is said to be like a human being, and a horse to be like a horse, and a human being to be unlike a horse; for a human being is like a human being because they have limbs of the same shape, which separate human beings from the category of the other animals. Among human beings themselves, for a like reason a man is more like a man than a man is like a woman, because men have more physical parts the same; and so an elderly man is more like an old man than he is like a boy. Further, they are more like who are of almost the same features, the same bearing of person, the same shape of body; therefore those who have more points of identity, are said to be more like; and those who come nearest to having them all alike, are called most like, as it were, twins.
5. There are those who think that things have three natures, like, unlike, and neutral, which last they sometimes call the not like, and sometimes the not unlike; but although there are the three, like, unlike, neutral, there can also be a division into two parts only, in such a way that whatever you compare with something else either is like or is not. They think that a thing is like and is unlike if it is seen to be of such a kind as I have described, and neutral, if it does not have greater weight on one side than on the other; as if the two things which are being compared have twenty parts each, and among these should have ten to be noted as identical and ten likewise to be noted as different, in respect to likeness and unlikeness. This nature most scholars include under the name of unlikeness.
6. Therefore since it happens that the question in dispute seems rather to be about the name than about the thing, attention must rather be directed, when something is said to be like, to the problem to what part it is said to be like; for it is in this that any mistake ordinarily rests. This must be noted, I say, because it can happen that a man may not be like another man even though he has many parts like the other’s, and can be said therefore to have like eyes, hands, feet, and other physical features in considerable number, separately and taken together, like the other man’s.
7. Therefore because careful watch must be kept in words to see what parts those words which are said to show likeness ought to have alike, and in what ways, the inquirer is on this topic especially likely to slip into error, as will appear below. For to the careless person what can seem more alike than the two words suis and suis? But they are not alike, because one is from suere ‘to sew’ and means ‘thou sewest,’ and the other is from sus and means ‘of a swine.’ Therefore we admit that they are alike as spoken words and in their separate syllables, but we see that they are unlike in their parts of speech, because one has tenses and the other has cases; and tenses and cases are the two features which in the highest degree serve to distinguish the different systems of Regularity.
8. Likewise, words that are even nearer alike in kind often cause a similar mistake, as in the fact that nemus ‘grove’ and lepus ‘hare’ seem to be alike since both have the same nominative; but it is not an instance of likeness, because they stand in need of certain factors of likeness, among which is that they should be in the same noun-gender. But these two words are not, for lepus is masculine and nemus is neuter; for we say hic ‘this’ with lepus and hoc with nemus. If they were of the same gender, the same form would be set before both, and we should say either hic lepus and hic nemus, or hoc nemus and hoc lepus.
9. Therefore he who asks whether the inflections of words stand in a regular relation, must examine to see what kinds of likenesses there are and of what sort they are, which pertain to this matter. And just because this topic is difficult, those who have written of these subjects either have avoided it or have begun it without being able to complete their treatment of it.
10. Therefore in this there is seen a lack of agreement, and not merely of one kind. For some have fixed the number of all the distinctions as a whole, as did Dionysius of Sidon, who wrote that there were seventy-one of them; and others set the number of those distinctions which apply to the words which have cases: the same writer says that of these there are forty-seven, Aristocles reduced them to fourteen headings, Parmeniscus to eight, and others made the number smaller or larger.
11. If the origin of these likenesses had been correctly grasped and their logical explanation had proceeded from that as a beginning, there would be less error in regard to the inflections of words. Of these likenesses there are, I think, first principles of two kinds only, by which the likenesses ought to be tested; of which one lies in the substance of the words, the other lies, so to speak, in the form of that substance, which comes from inflection.
12. For there must be one, that the word be like the word from which it is inflected, and two, that in comparison from word to word the inflectional form with which the comparison is made should be of the same kind. For sometimes there are like forms reached by inflection from like words, such as datives ero and fero from erus ‘master’ and ferus ‘wild,’ and sometimes unlike forms, such as genitive eri and accusative forum, from erus and ferus. When both principles are fulfilled and word is like word and inflectional form like inflectional form, then and not before will I pronounce that the word is like, and has a twofold and perfect likeness to the other — which is what Regularity demands.
13. But I wish to avoid the appearance of trickiness in having declared that there are only two kinds of likenesses when both have a number of sub-forms — if I say nothing about these, you may think that I am intentionally leaving myself a place of refuge; I shall therefore go back and start from the origin of the likenesses which must be followed or avoided in the comparison of words and in their inflections.
14. The first division in speech is that some words are not changed into any other form whatsoever, like vix ‘hardly’ and mox ‘soon,’ and others are inflected, like genitive limae from lima ‘file,’ imperfect ferebam from fero ‘I bear’; and since Regularity cannot be present except in words which are inflected, he who says that mox and nox ‘night’ are alike, is mistaken, because the two words are not of the same kind, since nox must come under the system of case-forms, but mox must not and cannot.
15. The second division is that, of the words which can be changed by derivation and inflection, some are changed in accordance with will, and others in accordance with nature. I call it will, when from a name a person sets a name on something else, as Romulus gave a name to Roma; I call it nature, when we all accept a name but do not ask of the one who set it how he wishes it to be inflected, but ourselves inflect it, as genitive Romae, accusative Romam, ablative Roma. Of these two parts, voluntary derivation goes back to usage, and natural goes back to logical system.
16. For this reason we ought not to compare Romanus ‘Roman’ and Capuanus ‘Capuan’ as alike, and to say that Capuanus ought to be said from Capua just as Romanus is from Roma; for in such there is in actual usage an extreme fluctuation, since those who derive the words set the names on the things with utter lack of skill, and when usage has accepted the words from them, it must of necessity speak confused names variously derived. Therefore neither the followers of Aristarchus nor any others have undertaken to defend the cause of voluntary derivation as among the Regularities; but, as I have said, this kind of derivation of words in common usage is an ill thing, because it springs from the people, which is without uniformity and without skill. Therefore, in speaking, there is in this kind of derivation rather Anomaly than Regularity, 17. There is a third division, the words which are by their nature inflected. These are divided into four subdivisions: one which has cases but not tenses, like docilis ‘docile’ and facilis ‘easy’; a second, which has tenses but not cases, like docet ‘teaches,’ facit ‘makes’; a third which has both, like docens ‘teaching,’ faciens ‘making’; a fourth which has neither, like docte ‘learnedly’ and facete ‘wittily.’ The individual parts of this division are each unlike the three remaining parts. Therefore, unless the words are compared with one another in their own subdivision, even if they do agree the one word will not be so like the other that it ought to make the same inflectional scheme.
18. Since there are several species in each part, I shall speak of them one by one. The first subdivision, characterized by the possession of cases, is divided into two parts, namely into nouns and articles, which latter class is both definite and indefinite, as for example hic ‘this’ and quis ‘who.’ Whichever of these two kinds you have taken, it must not be compared with the other, because they belong to schemes of Regularity which are different from each other.
19. In the articles, Regularity is hardly even a shadow, and more a Regularity of things than of spoken words; in nouns, it comes out better, and consummates itself rather in the spoken words and the likeness of the syllables than in the things named. There is also the additional fact that it is difficult to show that Regularities reside in the articles, because they are single words; but in nouns it is easy, because there is a great abundance of like name-words. Therefore it is not so much a matter of dividing this part from that other part, as of seeing to it that the investigator should be too much ashamed even to call that other part into the same arena to do battle.
20. As there are two groups in the articles, the definite and the indefinite, so there are in the nouns, the common nouns and the proper names; for oppidum ‘town’ and Roma ‘Rome’ are not the same, since oppidum is a common noun, and Roma is a proper name. In their account of the systems, some make this distinction, and others do not; but we shall enter in our account, at the proper place, what this difference is and why it has come to be.
21. That noun may be like noun, it ought to have the qualities of being of the same gender, of the same kind, also in the same case and with the same ending: kind, that if it is a proper name which you are comparing, it be a proper name with which you compare it; gender, that not merely one, but both words be masculine; case, that if one is in the dative, the other likewise be in the dative; ending, that whatever last letters the one has, the other also have the same.
22. To this fourfold spring two sets of lines are drawn up, the ones crosswise and the others vertical, as is the regular arrangement on a board on which they play with movable pieces. Those are crosswise which are the oblique cases formed from a nominative, like albus ‘white,’ genitive albi, dative albo; those are vertical which are inflected from one nominative to other nominatives, as masculine albus, feminine alba, neuter album. Both sets of lines are of six members. Each member of the crosswise lines is called a case; each member of the vertical lines is a gender; that which belongs to both in their crossed arrangement, is a form.
23. I shall speak first of the crosswise lines. Scholars have given various sets of names to the cases; we shall call that case which is spoken for the purpose of naming, the case of naming or nominative...
here three leaves are lacking in the model copy 24.... To indicate one ‘broom’ the plural scopae is used, not the singular scopa. For they are different by nature, because the names first mentioned are set upon simple objects, and those mentioned later apply to compounded objects; thus bigae ‘two-horse team’ and quadrigae ‘four-horse team’ are employed in the plural because they denote a union of objects. Therefore we do not say one biga, like one lata ‘broad’ and alba ‘white,’ but one bigae, with the numeral also in the plural; nor do we say duae ‘two’ with reference to bigae and quadrigae, as we say duae ‘two’ with application to the plural forms latae and albae, but we say binae ‘two sets’ of bigae and quadrigae.
25. Likewise the character of the form of a word is important, because in the form of the spoken word a change is sometimes made in the first part of the word, as in sŭit ‘sews’ and sūit ‘sewed’; sometimes in the middle, as in curso ‘I run to and fro,’ and cursito, of the same meaning; sometimes at the end, as in doceo ‘I teach’ and docui ‘I have taught’; sometimes the change is common to two parts, as in lĕgo ‘I read,’ lēgi ‘I have read.’ It is important therefore to observe of what letters each word consists; and the last letter is especially important, because it is changed in the greatest number of instances.
26. Because of this, since the likenesses in these parts also are wont to be used in the comparison of case-forms, and this is done ill by some and well by others, we must see whether this has been done rightly or wrongly. Yet wherever the letters are altered, not only the altered letters must be noted, but also those which are next to them and are not affected; for this proximity has considerable influence in the inflections of words.
27. Among these forms we shall not call those words like which denote like things, but those which are of such a stamp that such forms are in most instances wont by custom to denote like things, as by a man’s tunic or a woman’s tunic we mean not a tunic that a man or a woman is wearing, but one which by custom a man or a woman ought to wear. For a man can wear a woman’s tunic, and a woman can wear a man’s, as we see done on the stage by actors; but we say that that is a woman’s tunic, which is of the kind that women customarily use to dress themselves in. As an actor may wear a woman’s dress, so Perpenna and Caecina and Spurinna are said to have names that are feminine in form; they are not said to have women’s names.
28. The likeness of the inflection also must be watched, because the way which some words take is clear from the very words from which their inflection starts, as how it is proper to use praetor and consul, dative praetori and consuli. Others are properly appreciated only as a result of the change seen in the inflections, as in socer ‘father-in-law’ and macer ‘lean,’ because the one becomes socerum in the accusative, and the other macrum; after making this change, each of them follows its own way in the remaining forms, both in the inflections of the singular and in those of the plural. This method is employed because in the inflections there are two kinds of natures which can be compared with each other, one which can be seen in the word itself, such as homo ‘man’ and equus ‘horse,’ but the second cannot be seen through without bringing in something from outside, as in eques ‘horseman’ and equiso ‘stable-boy’ — for both are derived from equus ‘horse.’
29. By this method, you will, on making a comparison, know that of men observed in person one is or is not like the other; but you could not say that the two are in like fashion taller than their brothers, if you should not know how tall those shorter brothers are with whom they are compared. In this way the likenesses of things broader and higher, and others of the same kind, cannot be examined without bringing in some help from outside. So therefore, inasmuch as certain case-forms are of this kind, it is not easy to say that they are like, if you observe the spoken words in one case only; to make a correct judgement, you will have to bring in another case-form to which the spoken word passes as it is inflected.
30. These considerations are what I have thought enough to touch upon, for observing the likenesses of nouns. It remains to speak of the articles, of which some are like nouns and others are different. For of the five classes the first two have the same properties, because they have forms for masculine, feminine, and neuter, they have some forms to denote the singular and others to denote the plural, and they have five cases; the vocative is not indicated by a separate spoken form. They have this of their own, that some are definite, like hic ‘this,’ feminine haec, and others are indefinite, like quis ‘which,’ feminine quae. But since their system of Regularity is shadowy and thin, it is not necessary to speak further of it in this book.
31. The second subdivision consists of those words which have tenses but not cases, and have persons. The categories of their inflections are six: one which is that of the tenses, as legebam ‘I was reading,’ gemebam ‘I was groaning,’ lego ‘I read,’ gemo ‘I groan’; the second is that of the persons, as sero ‘I sow,’ meto ‘I reap,’ seris ‘thou sowest,’ metis ‘thou reapest’; the third is the interrogative, as scribone ‘do I write?’, legone ‘do I read?’, scribisne, legisne; the fourth is that of the answer, as fingo ‘I form,’ pingo ‘I paint,’ fingis, pingis; the fifth that of the wish, as dicerem ‘would I were saying,’ facerem ‘would I were making,’ dicam ‘may I say,’ faciam ‘may I make’; the sixth that of the command, as cape ‘take,’ rape ‘seize,’ capito, rapito.
32. Likewise there are four categories of inflections which have tenses without persons: in the interrogative, as foditurne ’is digging going on? seriturne ’is sowing going on?’ and fodieturne ‘will digging be done?’, sereturne ‘will sowing be done?’; of the category for the answer the same forms are used, but without the last syllable ne; the category for the wish, as vivatur ‘may there be living,’ ametur ‘may there be loving,’ viveretur ‘would there were living,’ amaretur ‘would there were loving.’ Whether the inflections for the impersonal command exist, is somewhat doubtful; there is also doubt about the scheme of the forms, which is given as paretur ‘let there be preparation,’ pugnetur ‘let there be righting,’ or parator, pugnator.
33. There are added to these categories those which proceed from the four sets of pairs consisting of the divisions: from that of the incomplete and the completed, as emo ‘I buy’ and edo ‘I eat,’ emi ‘I have bought’ and edi ‘I have eaten’; from that of the act done once and the act done more often, as scribo ‘I write’ and lego ‘I read,’ scriptito ‘I am busy with writing,’ and lectito ‘I read and reread’; from that of active and passive, as uro ‘I burn’ and ungo ‘I anoint,’ uror ‘I am burned’ and ungor ‘I am anointed’; from that of singular and plural, as laudo ‘I praise’ and culpo ‘I blame,’ laudamus ‘we praise’ and culpamus ‘we blame.’ With regard to the words of this class whose categories I have described, the matter of how full an equipment of forms each has, and what sort of forms it makes, will be set forth with more attention to detail in the books which are to be on the paradigms of verbs.
34. The words of the third subdivision, which are inflected with tenses and cases and are by many therefore called participials, are of this kind...
here three leaves are lacking in the model copy 35.... When we meet a new word, we ask about its case-forms, as to how we shall inflect them; and yet if some poet has made up some word and has himself formed from it some case-form in an incorrect way, we blame him rather than follow his example. Therefore Ratio or Relation, of which I am speaking, is present in both: in the words which are imposed upon things, and in those which are formed by inflection; and then also there is that third kind of Relation, which combines the characteristics of the two.
36. Among these, each and every relation, when compared with another, is either like or unlike; and often the words are different but the relation is the same, and sometimes the relation is different but the words are the same. The same relation which is in amor ‘love’ and dative amori is in dolor ‘pain’ and dative dolori, but not in dolor and accusative dolorem. The same relation which is in amor and genitive amoris is in plural amores and genitive amorum; and yet, because the subject-matter in it is not compared as it should be, this relation cannot of itself effect Regularities, on account of the differences in the forms of the spoken word, because a singular word has been associated with a plural. So, when it is by a proportionate likeness that the word has the same relation, then and not until then does this relation achieve what is demanded by Analogia or Regularity; of which I shall speak next.
37. There follows the third topic: What is Ratio or Relation that is pro portione ‘by proportionate likeness’? This is in Greek called ‘according to logos’; and from analogue the term Analogia or Regularity is derived. If there are two things of the same class which belong to some relation though in some respect unlike each other, and if alongside these two things two other things which have the same relation are placed, then because the two sets of words belong to the same logos each one is said separately to be an analogue and the comparison of the four constitutes an Analogia.
38. For it is as in a matter of twins: when we say that the one Menaechmus is like the other Menaechmus, we are speaking of one only; but when we say that a likeness is present in them, we are speaking of both. So, when we say that a copper as has the same relation to a half-as as a silver libella has to a half-libella, then we are showing what an analogue is; when we say that both in copper and in silver there is the same relation, then we are speaking of Analogia ‘Regular Relation.’
39. As sodalis ‘fellow’ and sodalitas ‘fellowship,’ civis ‘citizen’ and civitas ‘citizenship’ are not the same, but both come from the same origin and are connected, so analogue and Analogia are not the same, but are likewise congenitally connected. Therefore, if you take away men, you have taken away the sodales; if you take away the sodales, you have taken away the sodalitas: just so, if you take away the logos or Relation, you have taken away the analogue, and if you have taken this away, you have taken away the Analogia.
40. Since these are of such close kinship to each other, you must listen with keen understanding rather than wait to be told, that is, when I have said something about either, it will be also of general application to both; you should not wait for me to repeat it in writing in a later part of my work, but you should rather continue to follow up the line of thought.
41. These phenomena are produced in unlike things, as in numbers, if you compare two with one and so also twenty with ten; for twenty has to ten the same relation which two has to one. It is found also in like things; in coins, for example, one denarius is to one victoriate as a second denarius is to a second victoriate. So likewise in all other things those are said to be in a status of comparative likeness, wherein there is a fourfold nature of such a kind as among children the daughter is to the mother as the son is to the father, and in matters of time the midnight is to the night as the midday is to the day.
42. The poets make a great use of this kind of relationship in their similes, and the geometricians use it with greatest keenness; in reference to speech, Aristarchus and the grammarians of his school use it with more care than others do, as when acc. amorem and dat. amori, acc. dolorem and dat. dolori are said to be like by comparative likeness, although they see that amorem is unlike amori because it is in another case, and likewise dolorem is unlike dolori; but they say that the four are like, because they come from like words.
43. Sometimes it has two crossed relationships, in such a way that one is vertical and the other crosswise. What I mean will become clearer by this: Suppose that some numerals are so set down that in the first line there are 1 2 4 in the second line there are 10 20 40 in the third line there are 100 200 400
In this scheme of numerals there will be two examples of what I have called logos, which make different systems of Regularity: one is the twofold which is in the crosswise lines, because two is to four as one is to two; the other is the tenfold relation in the vertical lines, because ten is to one hundred as one is to ten.
44. Likewise the inflections of words may go in two directions, because from the nominative case they are inflected into the oblique cases, and from the nominative to the nominative, so that they make a similar scheme; which is in line 1: masc. nom. albus, dat. albo, gen. albi; in line 2: fem. nom. alba, dat. albae, gen. albae; in line 3: neut. nom. album, dat. albo, gen. albi. Therefore by the crosswise inflections there are made from these words systems of Regularity like Albius and Atrius, Albio and Atrio, which to be sure is only a small part of that binary scheme; and by the vertical inflections are made Albius and Atrius, Albia and Atria, which is a part of the tenfold scheme of Regularities of which I have spoken above.
45. Of that which is called Regularity, there are two kinds. One is disjoined, thus: as one is to two, so ten is to twenty. The other is conjoined, thus: as one is to two, so two is to four. Because in it two is said twice, both when we compare it with one, and then when we compare it with four, 46. this kind also is said to be fourfold by nature. So the cithers, though with seven strings, are none the less said to have two sets of four strings, because just as the sound of the first string stands in a certain relation to that of the fourth, so the fourth stands in the same relation to the seventh; the middle string is the first of the one set and the last of the other. The doctors who watch the seven days when a man is ill, note the symptoms of the illness with greater care on the fourth day, for the reason that the relation which the first day had to the fourth, foretells that the day which will be fourth from it, that is, seventh from the first, will bear the same relation to the fourth.
47. The Regularities are disjoined and fourfold in the cases of nouns, such as rex ‘king,’ dative regi, and lex ‘law,’ dative legi; they are conjoined and threefold in the three tenses of verbs, such as legebam ‘I was reading,’ present lego, future legam, because the relation which legebam has to lego, this same relation lego has to legam. In this, almost all men make a mistake, because they cite these verbs wrongly in the three tenses, when they wish to express them in a proportion.
48. For since some verbs denote incomplete action, like lego ‘I read’ and legis ‘thou readest,’ and others denote completed action, like legi ‘I have read’ and legisti ‘thou hast read,’ and since in the conjoined form they ought to be connected with others of their own kind and by this principle lego is rightly related to legebam — lego is not rightly related to legi, because legi denotes something completed; so that they are wrong in finding fault with tutudi ‘I have pounded’ and pupugi ‘I have pricked,’ tundo and pungo, tundam and pungam, as well as necatus sum ‘I have been killed’ and verberatus sum ‘I have been beaten,’ necor and verberor, necabor and verberabor, because the tenses of incomplete action are like one another, and those of completed action are like one another. Thus we should say tundebam tundo tundam, and tutuderam tutudi tutudero, and in the same way amabar amor amabor, and amatus eram, amatus sum, amatus ero. Therefore those who speak against the Regularities are unfair in finding fault on the ground that whereas Regularity is fourfold by nature certain words are cited in a different way, in three tense-forms merely.
49. This seems sometimes to have fewer parts, as I have said; similarly it seems, at other times, to have more parts, as when it is thus: as one and two are to three, so two and four are to six. Yet this form is included in the fourfold type, because sets of two are compared with sets of one. In speech also, this will sometimes be found, thus: As nominative Diomedes is compared with genitive Diomedi and Diomedis, so from nominative Hercules are said the genitive forms Herculi and Herculis.
50. And as these move away from one starting-point and nominative into two oblique case-forms, so on the other hand from two starting-points of the nominative many words unite in a single oblique case-form. For as from the nominatives Baebiei (masc.) and Baebiae (fem.) comes the dative Baebieis, so from the nominatives Caelii and Caeliae comes Caeliis. From two like words forms are developed in unlike fashion, as happens in nemus ‘grove’ and holus ‘vegetable,’ plural nemora and holera. Others from unlike words are developed in like fashion, as in the articles the accusatives hunc and istunc come from hic ‘this’ and iste ‘that.’
51. Regularity has its foundations either in the will of men or in the nature of the words, or in both. By will I mean the imposition of the word-names: by nature I mean the inflection of the words, through which passage is made without special instruction. He who starts from the imposition, will say that if dolus ‘guile’ and malus ‘bad’ are alike in the nominative, there will be found in an oblique case dolo and malo. He who starts from the nature of the words, will say that if Marco and Quinto are alike in the oblique cases, there will be nominatives Marcus and Quintus. He who proceeds from both, will say that if there is a likeness, then as the change is in servus ‘slave’ and vocative serve, so also there will be cervus ‘stag’ and vocative cerve. It is a common feature of all, that the four word-forms have their inflectional changes in a proportional relation.
52. The first kind starts from the likeness in the nominatives, the second from a likeness which is in oblique cases, the third from a likeness which is in the changes from case to case. In the first kind we set out from the imposed name to the nature, in the second we go in the other direction, in the third we go in both directions. Therefore in fact this third can be divided into two parts and called the third and the fourth, because in it the argument can actually go both forward and backward.
53. He who makes the imposed forms the starting-point for the Regularity, will have to develop the oblique forms from these; he who makes the nature the starting-point, will have to work in the other direction; he who starts from both, will have to make the rest of the inflections from the changes of the same kind. The imposition is in our power, but we are under the control of the nature of the words: for each one imposes the name as he wishes, but he inflects it as its nature requires.
54. But since a noun is imposed in two ways, either on a singular thing or on a plural — singular like cicer ‘chickpea,’ plural like scalae ‘stairs’ — and there is no doubt that the line of the inflections wherein things which are singular only will be declined, proceeds from some case of the singular, as cicer ciceri ciceris; and likewise that in the line of inflections which is in the plural only, it is proper to begin from some case of the plural, as scalae scalis scalas: another point must be examined, since their connected nature is twofold and two lines of inflections are made, like Mars and Martes, namely from what place the relation of Regularity ought to start, whether from the singular to the plural or vice versa.
55. For not even if nature does proceed from one to two, should the conclusion be drawn that in teaching the later thing cannot be the clearer, for the purpose of beginning from it, to show what the prior thing is. Therefore even those who deal with the nature of the universe and are on this account called physici ‘natural philosophers,’ proceed from nature as a whole and show by backward reasoning from the later things, what the beginnings of the world were. Though speech consists of letters, it is nevertheless from speech that the grammarians start in order to show the nature of the letters.
56. Therefore in the explanation, since one ought rather to set out from that which is clearer than from that which is prior, and rather from the un-corrupted than from a corrupt original, from the nature of things rather than from the fancy of men, and since these three factors which are more to be followed are less present in the singulars than in the plurals, one can more easily commence from the plural than from the singular, because in the latter as starting-points there is less of a basis for relationship in the forming of words. That the singular forms of words can be more easily interpreted from plural forms than plural forms from the singular, is shown by these words: plural trabes ‘beams,’ singular trabs; plural duces ‘leaders,’ singular dux.
57. For we see that from the plural nominatives trabes and duces the letter E of the last syllable has been eliminated and thereby in the singular have been made the nominatives trabs and dux. But on the other hand, if we start from the singulars we do not so easily see how they have become trabs, from B and S, and dux, from C and S.
58. If the nominative plural is by any chance a corrupted form, which rarely occurs, we shall correct this before we make it our starting-point; it is proper to take from the oblique cases, either singular or plural, some forms which are not ambiguous, from which can be seen the make-up which the other forms ought to have.
59. For sometimes the one is seen from the other and at other times the other is seen from the one, as Chrysippus writes, as the father’s qualities may be seen from the son, and the son’s from the father, and in arches the right-hand side stands on account of the left-hand side, no less than the left on account of the right. Therefore the oblique forms can sometimes be regained from the nominatives, and sometimes the nominatives from the oblique forms; sometimes the plural from the singular forms, and sometimes the singular forms from the plural.
60. The principle that we should most of all follow, is that in this the foundation be nature, because in nature there is the easier relationship in inflections. For it is easy to note that error can more easily make its way into those impositions which are mostly made in the nominative singular, because men, being unskilled and scattered, set names on things just as their fancy has impelled them; but nature is of itself for the most part uncorrupted, unless somebody perverts it by ignorant use.
61. Therefore, if one has founded the principle of Regularity on the natural cases rather than on the imposed case-forms, not many awkwardnesses will be his to face in usage; human fancifulness will be corrected by nature, and not nature by fancy, because those who have wished to follow imposition will in reality act in the opposite way.
62. But if one should prefer to start from the singular, he ought to start from the sixth case, which is a case peculiar to Latin; for by the differences in the letters of this case-form he will be more easily able to discern the variation in the remaining cases, because the ablative forms end either in A, like terra ‘earth,’ or in E, like lance ‘platter,’ or in I, like clavi ‘key,’ or in O, like caelo ‘sky,’ or in U, like versu ‘verse.’ Therefore, for the explaining of the declensions, there is this way, which may proceed from either of two starting-points.
63. But where there is Regularity, there are three factors, one which is in the things, a second which is in the spoken words, a third which is in both; the first two are simple, the third is twofold. In view of this, attention must be given to the relation which they have to one another.
64. First, of the differences which exist in the things, there are some which have no bearing on speech, others which are connected with it. Those which are not connected with it are like those which the artificers observe in making buildings and statues and other things, of which some are called harmonic, and others are called by other names; but no one of these becomes an element in speaking.
65. The differences which pertain to speech, consist of those things which are expressed by the words in a proportionate way, and yet do not have a likeness of the spoken words also to help in forming the inflections: such as nominative Iupiter and Marspiter, dative Iovi and Marti. For these are like one another in the gender of the nouns, and in the number, and in the cases; because both are nouns, and are masculine, and singular, and nominative and dative in case.
66. The second kind has to do with the sounds, in which the spoken words only are similar in a proportionate way — and not the things — as in biga and bigae, nuptia and nuptiae. For in these there is no underlying unit thing expressed by the singular when we say biga or quadriga, nor have the plural forms which are derived from these words any plural meaning. Yet all plurals which are derived from a unit singular, like merulae from merula ‘blackbird,’ do have such plural meaning; for they are of such a sort that there is subordination to a singular form: thus two merulae ‘blackbirds,’ three catulae ‘female puppies,’ four faculae ‘torches.’
67. Therefore since there cannot be the same subordinating relation because we do not say una biga, duae quadrigae, tres nuptiae, but instead unae bigae ‘one two-horse team,’ binae quadrigae ‘two teams of four horses,’ trinae nuptiae ‘three sets of nuptials,’ it is clear that bigae and quadrigae are not from biga and quadriga, but belong to another series: the usual series una, duae, tres, has una as its beginning, but in this second series unae, binae, trinae, the beginning is unae.
68. The third kind of Regularity is that which has two elements, which I mentioned, in which both the things and the spoken words are uttered with a similarity in a proportionate way, like bonus ‘good’ and malus ‘bad,’ plural boni and mali; Aristophanes and others have written about the Regularity in such words. And indeed this is a perfected Regularity in speech, but those two simple forms of Regularity are only incomplete beginnings; yet I shall speak of them separately, because we use them also in speaking.
69. But first I shall speak of the perfected Regularity, in which both the things and the spoken words are held together by a certain likeness; of this there are three kinds: one native, born here among us; the second coming from abroad; the third hybrid, born here of foreign paternity. The native type is such as sutor ‘cobbler’ and pistor ‘baker,’ dative sutori and pistori; the foreign type is such as Hectŏrĕs ‘men like Hector’ and Nestŏrĕs ‘men like Nestor,’ accusative Hectŏrăs and Nestŏrăs; that third type, the hybrid, consists of such words as Achilles and Peles.
70. Of these, many use the first type, not merely poets, but also almost all who speak in prose. At first they used to say Hectōrem and Nestōrem like quaestōrem and praetōrem; so Ennius says:
That Hector’s son be hurlèd from the Trojan wall. Accius in his tragedies began to take these words away from the early usage and rather to restore them to their Greek forms; hence Valerius says:
Accius would not use Hectōrem, but Hectŏra rather.
Because most of our foreign words are Greek, it has followed that the greatest number of the hybrid nouns which we have are also Greek in origin. Therefore, as in these types some words are Greek and others are Greek in origin, so also are the systems of Regularity.
71. Of the hybrid inflectional forms which are made from these materials in our country, some are early, like Bacchidēs and Chrysidēs, others are younger, like Chrysidĕs and Bacchidĕs, and still others are recent, like Chrysidăs and Bacchidăs; our fellow-countrymen use all three, but those who follow the middle forms in speaking give the least offence, because those of the first set seem insufficiently like the Greek forms from which they are taken, and those of the third seem insufficiently like our own forms.
72. The basis of all Regularity is a certain likeness, that, as I have said, which is wont to be in things and in spoken words and in both; we must see in which one of these sections each word should be entered, and of what sort it is. For, as I have said, neither the likeness of the things nor that of the spoken words is separately sufficient to express these double Regularities of the words, which we seek in speaking, because there must be a likeness in both respects. To introduce them into speech there must be also actual use; for the method by which you make a garment is quite different from that in which you wear it.
73. The categories of use appear to be three: one that of old usage, the second that of to-day’s usage, the third that of neither. Old words are such as cascus casci ‘old,’ surus suri ‘stake’; words of to-day’s usage, such as albus ‘white,’ caldus ‘hot,’ datives albo and caldo; words of neither usage, such as scala and acc. scalam ‘stair,’ phalera and phaleram ‘trapping.’ To these there can be added a fourth kind which does not belong exclusively to one category, like amicitia ‘friendship’ and inimicitia ‘enmity,’ accusatives amicitiam and inimicitiam. The first is that which the ancients used and we have abandoned; the second is that which we now use; the third is that which the poets use.
74. That Analogia or Regularity which is directed toward the nature of the words is not to be defined in the same way as that which is directed toward the actual use in speaking. For the former should be defined thus: Analogia is the like inflection of like words; and the latter thus: Analogia is the like inflection of like words, not inconsistent with common usage. But when to the end of these two there has been added “within a certain range,” then poetic Analogia will be defined. The first of these is that which the people ought to follow; the second is that which all the individuals in the people ought to follow; and the third is that which the poets ought to follow.
75. I think that these things have been said with more care than clarity, but not more obscurely than are the definitions of the same subject given by the grammarians, such as Aristeas, Aristodemus, Aristocles, and others, whose obscurities are the less to be found fault with, because most definitions, being on an unknown theme and being expressed with extreme brevity, are not easily understood unless they are expounded point by point.
76. Therefore the matter will be more apparent if there is a clear exposition of the parts one by one, as to what is meant by a word, what is meant by the likeness of the word, by inflection, by likeness of inflection not inconsistent with common usage, and by “within a certain range.”
77. By word I mean that part of spoken speech which is the smallest indivisible unit. If a word has natural inflection, then a word is like another word when it is similar to the other word in the thing which it denotes and in the spoken word by which it denotes the thing and in the form which it has after an inflectional change has taken place. Inflection is that which takes place when some change of the spoken word is made from word-form to word-form or to a new word-stem by derivation, in order to express a change of the thought. Likeness of inflection exists, when it passes from some form to another form in the same way in which that other word passes with which it is being compared.
78. There is the addition “not inconsistent with common usage,” because usage tolerates some words inflected contrary to the old practice, as it suffered Hortensius to say cervix ‘neck’ instead of the plural cervices, but does not tolerate certain others, as when you should say faux ‘throat’ instead of the plural fauces. When the addition “within a certain range” is made, it means that in the relevant words not all the forms are in use, as, for example, there is derived from amo ‘I love’ and vivo ‘I live’ the passive amor but not the passive vivor.
79. What Analogia or Regularity in speech is seen to be and what categories it has, and which of these seem essential to follow, I have set forth as briefly as I could. Now I shall speak of the categories in which it ought not to exist and yet it is usually looked for just as if it ought to be there; these are in general of four kinds. First, Regularity ought not to be looked for in such words as are not inflected, for example nequam ‘worthless,’ mox ‘soon,’ vix ‘hardly.’
80. Among these, a greater error is made in one word than in another. For they grant that mox and vix have no cases, but assert that nequam has, because we use it with nominative hic ‘this,’ with genitive huius, with dative huic. For when we say hic nequam and huius nequam, then we are uttering the cases of this man whom we wish to show as worthless, and before the word we then set hic to represent the name of him whose worthlessness we are considering.
81. This word is made like nolo ‘I do not wish’ from non ‘not’ and volo ‘I wish’; thus from ne ‘not’ and quicquam ‘anything,’ with loss of the middle syllable, is likewise compounded nequam. So as him whom we think to be non hili ‘worth not a whit’ we call nihili, him in whom we think that there is ne quicquam ‘not anything’ we call nequam.
82. Second, Regularity is not to be looked for if the words have only one case in their spoken form, because they are not inflected, like all names of letters. Third, it is not to be looked for if the series of forms which the noun has is unique and has nothing with which it can be compared, as they consider true of caput ‘head,’ dat. capiti, gen. capitis, abl. capite. Fourth, it is not to be sought if those four noun-forms which are compared with one another fail to have the mutual relation which they should have, as in socer ‘father-in-law’ and socrus ‘mother-in-law,’ accusative plural soceros and socrus.
83. On the other hand, in words in which Regularity ought to be looked for, in general the same number of stages should be found in conjunction: first, the things should exist; second, the things should be in use; third, these things should have names; fourth, they should have natural inflection. As for the first stage, because the nature of plural and singular is basic, we say plural nom. asses, acc. asses, singular nom. as, acc. assem; on the other hand, because in definite plural numerals the singular nature does not exist, only plural forms are used, such as nominative duo ‘two’ and tres ‘three,’ dative duobus and tribus.
84. In the second stage, if the nature exists but there is no practice of making this kind of distinction, as happens in faba ‘bean’ and in that class of words which we use for one and for all collectively, without change of form: for there was no need, as in the matter of slaves...