III. SOME PRE-HISTORIC CHANGES AND THEIR FRUITS
The Sounds of Pro-ethnic Indo-European
§ 54. The chief sounds which can be shown to have existed in words that descended from Indo-European into Latin are these:
| Vowels: a, e, i, o, u, ə. | |||
| Liquids and Nasals: r, l; m, n. | |||
| Breathed | Voiced | ||
| Fricatives: | Dental | s | z |
| Plosives: | Velar | q | ǥ |
| Palatal | k̑ | g̑ | |
| Dental | t | d | |
| Labial | p | b | |
| Voiced Aspirates: ǥh, g̑h, dh, bh. | |||
Besides these there were the Non-vowel Sonants (r̥, l̥, m̥, n̥); the Semi-vowels, Consonantal i and u (i̯ and u̯); the Velar nasal ŋ, and the Palatal nasal n̑, all of which have been just described. Of the difference between the so-called “Pure Velars” and the other Velars something will be said in § 173.
Most of these sounds except the Aspirates (§§ 174 ff.) and the Non-vowel Sonants (§§ 197 and 207), remained the same on passing into early Latin; their changes in Latin we shall see later on.
The Neutral Vowel
§ 55. The I.Eu. neutral vowel ə (§ 29) became i in Sanskrit, but ă in Greek and Latin, e.g. in Lat. status ‘fixed,’ Gr. στατός, compared with Sansk. sthitás ‘placed, standing.’
The Sign :
§ 56. A direct connection between two words is often conveniently expressed by the symbol :, which is used to mean ‘in relation to.’ Thus we may state the facts just mentioned by writing:
I.Eu. ə appears in I.Eu. *stətós : Sans. sthitás : Gr. στατός, : Lat. status.
These three words are identical, that is, they all go back to the same original. But the sign : denotes any direct kind of relation between two forms; for example we may write “gero : gestus,” “cado : concido,” “Eng. man : Eng. men,” reading the sign as a short way of writing ‘considered in relation to’ or ‘when compared with’ or ‘beside.’
Some early Phonetic Changes
§ 57. Before studying the history of these sounds in Latin, it will help us to consider briefly two sets of changes which happened outside the separate history of Latin, but which greatly affected the relations of the sounds which we find existing in Latin to those which we find in parallel words in the kindred languages. For example we find (1) Lat. māter : Eng. mother
(2) Lat. tentus : Gr. τατός, Sans. tatás ‘stretched’ all clearly related also to Gr. τείνω ‘I stretch,’ and Lat. tenor ‘stretch, continuance.’
Similar correspondences occur in a host of other words.
§ 58. The first correspondence becomes clear as soon as we know something of the phonetic changes which happened in early, that is, pro-ethnic[1] Germanic which are called Grimm and Verner’s Law. The second group of words has been shaped by changes caused by Accent in the Indo-European parent language. Both these sets of changes are easy to understand and throw a great light on the history of all the descendant languages. Grimm and Verner’s Law also depends partly upon the Indo-European Accent; and as it affected the Germanic languages only, it will be best to state it first.
Grimm and Verner’s Law in pro-ethnic Germanic
§ 59. In the Germanic branch before any of the Germanic languages that we know (§ 8) had split off from the rest, all the Plosives and Aspirates which had been inherited from the parent Indo-European language suffered phonetic changes which are often called ‘the Sound-shift.’ These were first formulated in 1822 by the German scholar Jacob Grimm who made the famous collection of fairy-tales. His statement was completed by the Danish scholar Verner in 1875, who explained a mass of exceptions by discovering the second of the four rules which follow.
§ 60. In early Germanic, probably at some time between 500 and 200 B.C., the following changes took place:
(1) All Breathed Plosives (unless preceded by s) became the corresponding Breathed Fricatives, p became f, t became þ (Eng. th in thin) and so on. Thus
- thf in Eng. thornfull corresponds to p in Lat. plēnus,
- th in Eng. thorn corresponds„ to„ t in Lat. terō,
and, since the Germanic Palatal Fricative became later merely h in English, h in Eng. hund-red corresponds to c in Lat. centum. But after s the Plosives were unchanged, as in Eng. sta-nd : Lat. stā-re.
(2) (a) These Breathed Fricatives remained when they came at the beginning of a word, as in the examples just given, or when they occurred immediately after a syllable bearing the Accent, as in the suffix -th which in English in such words as youth, width, corresponds to -t- in Latin in such words as iuven-ta, senec-ta. We know from Greek and Sanskrit that the syllable preceding this suffix was accented in the parent Indo-European language. So Eng. mother, brother[2] beside Gr. μήτηρ (older μᾱ́τηρ), φρᾱ́τηρ ‘member of a group of families.’
(b) But when the Breathed Fricative stood before a vowel which itself bore the accent, the Breathed Fricative became Voiced in early Germanic, and then generally became a Voiced Plosive; so in this case we have regularly Eng. b, d, g, corresponding to I.Eu. and Lat. p, t, k.
Thus the Eng. -d of participles and adjectives (as in loved, or moneyed) comes from the original -t- which is preserved in the Lat. ending -tus (§ 282), Greek ‐τός, Sans. -tás; Greek and Sanskrit keep the original accent on the ending. So the -d- in Eng. gard-en comes from the -t- which is preserved in Gr. χορτóς, ‘cleared space,’ Lat. hortus. So the -d- in Cumberland-English fader represents correctly the -r- before the accented syllable in Gr. πατήρ, Sans. pitā́; the -th- in southern English is due to an analogical assimilation to the -th- in mother.
(3) The Indo-European Voiced Aspirates became in Germanic Voiced Fricatives and these, as we have just seen, became Voiced Plosives.
Thus Eng. b in to bear corresponds to Sans. bh in bhar-āmi, Gr. φέρω, Lat. fero (§ 175), all meaning ‘I bear.’
Eng. d in dare, durst answers to Sans. dh in dhársati ‘he dares,’ Gr. θ in θάρσος ‘boldness.’
Eng. g in garden answers to I.Eu. g̑h which appears as χ in Gr. χορτός and h in Lat. hortus.
(4) The Voiced Plosives became Breathed.
| Thus | Eng. | p | in | thorpe | corresponds | to | Lat. | b | in | trabs, ‘beam for building.’ |
| „ | „ | t | „ | (to) tow | „ | „ | „ | d | „ | dūco, ‘I draw, lead’. |
| „ | „ | c | „ | (bishop)-ric | „ | „ | „ | g | „ | rēgnum, ‘kingdom’. |
Accent
§ 62. By an Accent we mean the degree of prominence in speaking given to one particular syllable above the other syllables of the same word or phrase. Thus in Eng. merrily the first syllable, in abandon the second syllable, in ascertain the third syllable has the Accent of the word; in hear me, or down with it the first syllable has the Accent of the phrase; in be quiet! the second syllable.
Enclitics and Proclitics
§ 63. In all such phrases the unaccented words that come after the accented syllable are said to be Enclitic (me and with it in the first two phrases): those that come before it are called Proclitic (be in be quíet).
Stress Accent
§ 64. But there are two ways in which this prominence of a syllable can be made audible in speech.
The first is by pronouncing the accented syllable with greater force than the rest, as we do in English. This is called a Stress Accent, or an Exspiratory Accent, because it depends on the force of our exspiration.
Musical Accent
§ 65. But in some languages the Accented syllable is pronounced not necessarily with more force, but upon a higher note than the others, the Vocal Edges (§§ 24–27) being more tightly stretched while the Sonant part of that syllable is being made. This is called a Tone Accent, or Musical Accent.
In Ancient Greek, in Sanskrit, and in the latest period of pro-ethnic Indo-European this Musical accent prevailed. It still survives in Lithuanian, which in this and some other respects is the most primitive of the spoken languages of modern Europe. But in some period earlier than the latest of the undivided IndoEuropean language, the Accent must have carried a strong Stress (see § 69).
Change of Character in Accent
§ 66. The character of the Accent may change in a language. Thus in Greek in the 5th Century B.C. it was certainly Musical; but by the 2nd Century B.C., if not earlier, it had certainly become Exspiratory as it is in Modern Greek; hence came the lengthening of the vowel in the accented syllable, frequent in even Attic inscriptions of the 2nd Century A.D. (e.g. Σώλωνος instead of Σόλωνος C.I.A. iv. 11, written between 174 and 178 A.D.), and occurring much earlier in Greek papyri from Egypt. In Modern Greek accented vowels are regularly long.
Character of Latin Accent
§ 67. In Latin the Accented syllable was pronounced with greater Stress, but it is possible that there was some raising of the Tone also in the language of polite society. But in the everyday speech of the camps from which all the Romance languages outside Italy arose, the Stress accent was very strong. This was the cause of the remarkable contractions and mutilations that appear in some of these languages, especially in French; for example in such a word as cité from Lat. cīvitā́tem, where the stress accent on the syllable -tā- (see § 85 (4)) has helped
- (1) to suppress the syllable -vi-,
- (2) to shorten the i of cī-,
- (3) to suppress the final syllable altogether.
Written Signs of Accent
§ 68. Here it is well to note that the ‘acute’ sign ´ is properly only used to mark an accented syllable; and the ‘circumflex’ ( ̂ or ̑ or ˜) is properly used only in languages with a Musical accent, and then to mark a syllable in which the tone rises on the first half of the syllable and falls on the second half, as in Gr. τοῦτο.
Unfortunately these convenient signs have come to be used in writing many languages to denote not Accent at all, but merely a particular quality or quantity of vowel; as in Fr. passé, cité to mark a particular kind of e; or as in Fr. hâter to mark a long vowel. And still more unluckily these actual signs are called ‘accents’ although in fact they do not in such cases denote Accent at all.
Ablaut
§ 69. In some period of the Indo-European parent language the Stress-accent upon one syllable robbed a preceding or following syllable of much of its quantity; words like Lat. dūco (older *déuco) ‘I lead,’ fīdo, older feido ‘I trust,’ show in their first syllable what is called the Normal form of the root; and this was preserved because in these verbs that syllable was accented in Indo-European. But there are words from the same roots, like dux, ducis ‘leader,’ fidēs ‘trust,’ which show a short form of the root, contaming only ŭ or ĭ instead of eu and ei; and this short or ‘Weak’ form arose not in Latin but in Indo-European in words which had originally contained the Normal form of root but in which the Accent in that period fell on the following syllable, and had thus reduced the root-syllable to a Weak form.
We find similar variation due ultimately to the same cause in suffixes (§ 226–243), as in -ter- -tr-, -en- and -n-, -meno- and -mno-.
§ 70. The commonest case of this weakening of syllables through their coming to stand in an unaccented position is the Participle in -tós, where, as we have seen in § 60 (2) (b), the suffix always bore the accent. Hence e.g. the partc. of I.Eu. *ten-iō ‘I stretch’ (Gr. τείνω, Lat. tendo) became not *ten-tós but *tntós which appears in Gr. τατός, Sans. tatás, for in Greek and Sanskrit n and m both became a. In Latin however n always became en, so the Latin form is tentus. The Verbal Nouns ending with the suffix -tí-, which was always accented, also show the Weak form of the root, as Gr. φάτις, from the root of Lat. fama, Gr. φᾱμί, φημί, or Gr. πίστις beside πείθω, or Lat. ratis beside rē-ri ‘to think’ (§ 216).
§ 71. From roots whose Normal form contained ā, ē, or ō, the Weak form in Indo-European had the neutral vowel ə which became short a in Latin (§ 55). This explains such groups of forms as the following:
Root dhē- (sometimes in a longer form dhēk̑-) ‘to make, put,’ Gr. τί‐θη‐μι, ἔ‐θηκ‐α.
: Sans. da-dhā-mi.
: Lat. fēcī beside facio.
Root stā- ‘to stand, make to stand,’ Gr. (Doric) ἵ-στᾱ-μι.
: Sans. ti-ṣṭhā́-mi.
: Lat. stā-re, stā-men beside status.
Root dō- ‘to give,’ Gr. δί‐δω‐μι.
: Sans. dā́-da-mi.
: Lat. dō-num beside datus, and also damus, datis, dabo.
§ 72. There seems to have been under some special conditions a still further weakened stage in which even the -a- completely vanished: Sans. devattás ‘god- given’ stands for *deiu̯o-d-tós; and Lat. vici-ss-im ‘by giving in turn’ seems to show the same grade of the same root with -ss- from -d-t- (§ 166); on -im see § 269.
§ 73. The changes just described happened in a period of the I.Eu. parent speech long before any of the descendant languages had split off. But at a later period, not long before the earliest separation, perhaps as recently as 8000 or even 2000 B.C., the character of the Accent had become more musical (§ 65). This seems to have produced quite a different kind of effect on the vowels of unaccented syllables. These were not shortened or crushed out as unaccented vowels had been in the Stress-accent period, but changed in quality. Thus -e- (and probably -a-) in a syllable preceding the accented syllable became -o-. This affected a great number of words which had been created in the language after the period of Stress-accent was over. Thus from I.Eu. *lég̑o ‘I tell, count’ were formed abstract nouns *log̑ā́ and *log̑ós ‘telling, counting’; from *tég̑o ‘I cover,’ an abstract noun *tog̑ā́; ‘covering’; from *bhéidhō ‘I trust’ (or transitively ‘I make to trust’) which became Lat. fīdō, the reduplicated perfect *bhébhoidha ‘I have trusted’ which became Gr. πέποιθα. These -o- forms are said to contain the Deflected form of the root.
§ 74. Thus we get sets of forms such as these appearing in separate languages:
| Normal form of root. | Deflected form of root. |
| Lat. ago, Gr. ἄγω ‘I drive’ | : Gr. ὀγ‐μός ‘furrow.’ |
| Lat. tego ‘I cover’ | : Lat. toga ‘a man’s robe.’ |
| Gr. δέμω ‘I build’ | : Gr. δόμος Lat. domus ‘house.’ |
| Gr. λείπω ‘I leave’ | : Gr. λέλοιπα ‘I have left.’ |
| Old Lat. feido, Gr. πείθομαι ‘I trust, am persuaded’ | : Gr. πέποιθα ‘I have learnt to trust,’ Lat. foed-us ‘pledging of trust, covenant.’ |
(The changes of the consonants in this last set will be explained in §§ 174 ff., 182.)
§ 75. The examples given, e.g. Gr. δόμος, show that even in Greek the Indo-European accent had been shifted in many words from the place in the word which it had had in Indo-European; and we shall see (Chap. V) that the place of the Accent was completely changed in Latin. Nevertheless the forms produced by the old system remained in the separate languages, sometimes isolated like bits of wreckage left high and dry on the beach by the tide.
§ 76. But the different branches of Indo-European started on their separate development already possessing a great number of sets of words thus produced. The words in each set were linked together by their meaning but showed their different types, those with the Normal, those with the Weak, and those with the Deflected form of root. These types are called Grades of Ablaut. Ablaut is a convenient German name (literally: ‘variation of sound’) for the connexion between the Normal forms on the one hand and the Weak and Deflected forms on the other.
§ 77. Now these new forms, though they had arisen by unconscious phonetic changes, had come to be associated with special kinds of meaning, e.g.
| Normal. | Weak. | Deflected. |
| I.Eu. *ten-iō ‘I stretch’ | *tn̥tós ‘stretched’ | *tétone ‘he has stretched.’ |
| *tonós ‘a stretching.’ | ||
| Gr. τείνω | τατός | τέτονε, τόνος. |
| Lat. tendo, tenor | tentus | |
| I.Eu. *lei̯q-ō ‘I leave’ | *éliqom ‘I left’ | *leloi̯qe ‘he has left.’ |
| *liq-tós ‘left’ | *loi̯qós ‘remaining.’ | |
| Gr. λείπω | ἔ‐λιπον | λέλοιπα, λοιπός. |
| Lat. | re-lic-tus |
Thus the Weak form came to be felt as proper to the partc. in -tos and the Deflected form as proper to the Perfect Active (in the Singular) and to derived Nouns or Adjectives of what we call the First and Second Declensions; compare also the examples in §§ 71, 73 and 74.
Hence it is true to say that though the phonetic changes which had first produced the Grades of Ablaut had long ceased, yet the system of Ablaut so created had become charged with grammatical meanings, and new forms could be made upon the pattern so established, and were made, in fact, in many of the branches of Indo-European.
§ 78. In English, we have verbs conjugated like this
| Present tense | Participle | Past tense |
| sink | sunken | sank |
which represent the type; for the -in- of the English present has come regularly from an earlier -en-; the -un- of the Partc. represents a Sonant like the n̥ of I.Eu. *tn̥tós; and the -a- of the past tense has come regularly from an earlier -o-.
§ 79. Besides these three common Grades, there were others which appear in Latin only in a few forms. We have the Lengthened grade of the suffix-syllable in Gr. Nominatives like πα‐τήρ ‘father’ and ποι‐μήν ‘shepherd,’ beside the Normal form in the Acc. πα‐τέρ‐α, ποι‐μέν‐α, and the Weak form in the Gen. πα‐τρός and in derivatives like πά‐τρ‐ιος ‘belonging to one’s father,’ ποι‐μν‐ή ‘flock.’ The same variety appears also in the Deflected forms, e.g. in the compound Adj. Nom. εὐπάτωρ ‘well-sprung,’ εὔφρων ‘well-minded, cheerful’ beside the Accusatives εὐπάτορα, εὔφρονα just as in the simple nouns πατήρ, πατέρα and φρήν, φρένα ‘mind.’ Examples of the Lengthened forms appear in Old Lat. Nominatives like patēr (later pater, § 95); and the Lengthened Deflected forms appear in Nouns of the Agent in -tōr, later -tor; or of the suffix -es- : -os- in abstract Nouns like odōs, later odor (§ 243). In both these types of Nouns the long vowel was carried by Analogy all through the declension (Acc. monitōrem, etc.), though by the time of Cicero the final syllable of the Nom. had been shortened (§ 180). Contrast with these the Weakened form -tr- on which the corresponding Feminines are based, e.g. Acc. moni-tr-īc-em (§ 231). In nouns like homo from older homō (§ 95) the Strengthened Deflected form appears in the Nom. Sing. But in some words, such as sermō ‘conversation,’ tēmō ‘pole of a chariot,’ the Strengthened form spread through all the cases.