<. . .> AS A RESULT of well-known phonetic processes, the forms of declension and conjugation do not remain the same, but undergo changes. The changes consist chiefly in a shortening of the forms.
The question arises: do only the endings become shortened, or do the stems as well? Are only the endings subject to change, or the stems too?
In the view of most investigators, one can speak only of a history of endings, but not of stems. This view is based on the assumption that once stems had been shaped and formed in a certain way, they exist somehow, like fossils, outside of man, and remain independent of our thought processes.
Thus, if we assume that proto-Indo-European had stems ending in the vowels -a, -i, -u, then we must posit them in all languages continuing the original proto-Indo-European language despite the various changes which these underwent in the course of time, despite the fact that some of them have not preserved the slightest trace of -a, -i, -u nor, for that matter, of any ending. Thus, although the proto-Indo-European varka-s 1 is represented by the modern Russian volk, the primary root varka-, or at least volka-, is still assumed to be present in volk . It is as if we were faced with a grey-haired man whom, despite appearances, we still think of as dark-haired, because we know that his hair used to be dark. And what would we perceive if we met this man for the first time and had no idea that he used to be dark-haired? We would still insist that he is dark or blond or a redhead but never that he is grey. Similarly, we can see no clear connection between one or another isolated Slavic noun and a vocalic stem in -a, -u etc., but we continue to treat it as if it were a vocalic stem and not knowing the details, assume that the given noun contains an -a, -i, -u, -ia or other kind of stem.
We see a bald man before us, but we do not regard him as bald because in his youth he had a splendid head of hair, and we still believe him to be thick-haired and handsome.
At a point in mans embryological development, he had a small tail which later shortened and disappeared, leaving only an unnoticeable rudiment. In line with the view that the Slavic and other Indo-European languages still contain -a, -i, -u stems, every man should then be described as having a tail, a small one to be sure, but still a tail.
American farmers have bred a new, hornless breed of cow from the usual horned breed; but to the followers of the -a, -i, -u stemtheory, the heads of these cows are still graced with long horns.
Somewhere on earth a mountain has appeared where there had been none, and somewhere else a mountain has disappeared; in a third place, the sea has given way to dry land, and in a fourth, dry land has yielded to the sea. To the eye of a follower of the theory of the immutable -a, -i, -u stems, there stand as before, a plateau, a mountain, the sea, and dry land.
What would we say about biologists and geologists if they proceeded in a similar fashion? And yet this is precisely the sort of logic pursued by the champions of the incontrovertible - a, -i, -u stem-theory.
What is the raison detre of this stubborn stem-theory? What prevents its champions from reexamining its worn-out and illusory truth and from embracing an evolutionary point of view, which has prevailed as the only valid one in the biological and social sciences?
To answer this question we shall turn for a moment from lin guistics to psychology.
The founders of comparative grammar took Sanskrit as the basis for their study of all Indo-European languages. Even at the birth of this discipline, which was not so long ago, its practitioners did not attempt to explain the structure of the Indo-European languages, but were intent only to compare them with Sanskrit, looking at their development through Sanskrit glasses and imposing upon them Sanskrit categories. They found the -a, -i, -u stems in Sanskrit itself (although even here they are not retained intact), and what is more important, in Indie grammatical literature. And since these stems existed in Sanskrit, it was concluded they must exist in all other Indo-European languages. The followers of the early comparativists made this theory their own, and it is well known how strongly people feel about what they learn at the beginning of their study. Were is not for teachers who cannot distinguish sounds from letters, there would not be that profusion of grammarians (among whom some enjoy fame and respect) who confuse sounds and letters in the most simplistic manner, for whom the Russian sc is a simple consonant, ja is a soft vowel, 6 and 6 semivowels, and so on, to give but a few examples.
If it were not for the inspiration of the Indie grammarians, if the comparative grammar of the Indo-European languages had instead begun with Latin, Greek, or the Slavic languages, and if Greeks, Romans, or Slavs, and not the ancient Indians, had been the authors of illustrious treatises on the morphology of language, then all those ardent champions of the -a,—i, -u stems in all Indo- European languages would have probably failed to see them not only in their own languages, but even in Sanskrit and proto-Indo- European. Such is the power of suggestion and habit. As someone once said, the strongest drive behind human activity is laziness. And laziness is, no doubt, the main reason why scholars cling to the theory of the -a, -i, -u stems in all Indo-European languages<. . .>.
I propose that each subject be first of all treated in terms of its own properties without imposing upon it alien categories. A feeling for language, and especially for its psychological aspect, must serve as an objective criterion for the scientific investigation of the facts of language. I submit that feeling for language is not just a clever formula or a will o the wisp, but a real and objective phenomenon of language.
Even a foreigner, or rather, a speaker of a foreign language, can discover and acquire a feeling for a given language if he studies it thoroughly, if he masters its facts with insight and perspicacity and, above all, if he is free of all the prejudices which we acquire by tradition.
If we approach language in this way, it appears that the stems or themes are not merely fictions in the hazy atmosphere of the protolanguage, not merely echoes of a golden age of language frozen in their immobility, but living parts of declined and conjugated words. They are a necessary component of inflection so long as the latter is not replaced by a different word-structure consisting of morphologically indivisible wholes, which happens when declined and conjugated words cede their place to an analytical form of language development.
Since it is generally agreed that stems, like other morphological parts of the word, become petrified in the secondary or analytical languages, why not admit that they could similarly develop and change in the primary languages, in which we recognize the development and regeneration of roots, suffixes, and endings?
Furthermore, if the theory of fixed and petrified stems concedes that certain consonantal noun stems were replaced by vocalic stems, that and -i and -u stems were replaced by -a stems, and moreover that the stems admitted phonetic diversity (e.g., the distinction of strong, middle, and weak stems in Sanskrit), then why rule out the possibility that the end of a stem could have been reduced and its final vowels -a, -i, and -u dropped?
In the light of the forgoing remarks we would have to conclude that stems or themes, like everything else in language, are subject to constant change. These changes are either (1) purely phonetic or (2) due to analogy. Because of these changes, the once extant stems in -a, -i, -u and the like (which were not fully retained even in Sanskrit) lost their final vowel in Slavic and other Indo-European languages and changed into shorter, consonantal stems. The phonetic reflexes of the old endings now serve in Slavic, as in the other primary Indo-European languages, as simple endings expressing different case relations.
This reduction of stems in favor of endings must have begun in proto-Indo-European in those instances where the final vowel of the stem coalesced with the initial vowel of an ending into one long vowel. Thus, for example, one can hardly detect a vocalic -a stem in abl. sg. masc.-neut. in -ād or -āt (Sanskrit -āt, Lat. -ōd, Lith. -o, Slavic -a), or in other cases that were similarly formed. The proto-language itself showed in such instances a reduction of the stem in favor of the endings yielding a shorter stem form (which included only the last consonant).
The historical development of declined nominal stems from proto-Indo-European to the present as represented, for example, by the Slavic languages, can be reconstructed approximately as follows:
The posited oldest proto-Indo-European period contained two groups of stems:
I. Stems ending in vowels and diphthongs.
II. Stems ending in consonants.
The first group was more numerical both in terms of the words belonging to it and in the variety of thematic types.
The principal types included:
1. Stems in final -ā (long a) (fem. and in part masc.).
2. Stems in final -ă (short a) or -e | -o (masc. and neut.).
3. Stems in final -u or -u | -eu │-ou (of all three genders, but preponderantly masc.).
4. Stems in final -i or -i | -ei | -oi (also of all three genders, but preponderantly fem.).
Types 3 and 4 were transitional to the second principal group of stems ending in a consonant.
The final vowels of the first and second type were preceded either by a hard, nonpalatalized consonant, or by j (alone or clustering with a preceding consonant). This distinction is of great importance for the development of proto-Slavic from proto-Indo-European, since the final vowel split into two vowels, depending on the quality of the preceding consonant.
The second principal group of proto-Indo-European declensional stems was represented by only one subgroup, comprising substantives of all three genders, but only a few isolated forms belonging to this type were retained in the transition to the individual Indo-European languages, especially proto-Slavic, namely: stems ending in -ū | ŭv (fem.), -er | -r (-ter | -ter) (fem.), -men | mn (masc. and neut.), and -as (-es | -os) (neut.).
The transition from proto-Indo-European to proto-Slavic was marked by two phonetic processes which are reflected in the declension. One was the influence of the consonant ƒ on the following vowel. Under the influence of j, the first and second declensional types of the first principal group split into two subtypes, (A) and (B) exhibiting the following vocalic endings:
The phonetic process was the loss of all consonants at the end of syllables and the appearance of open syllables at the end of all words. The endings, too, lost their final consonants. But the loss of final consonants did not eliminate the need to express case relations by means of endings, which came now to be expressed by the vowels of the former stems, whereas the new stems terminated in a consonant. Thus, there came about a general reduction of the stems in favor of the endings. The endings at that time consisted either of a part of the former stem, or of a part of the former stem combined with the phonetic continuation of the old ending.
The two phonetic processes brought about morphological changes only in the first principal group of declensional types, that is, in the group going back to the proto-Indo-European vocalic stems (-a, -u}-i. . .). The result of the first process was a rearranged distribution of the first and second subtypes of the first group, while the second process affected the morphological character of all four subtypes of the first group.
In the second (consonantal stems) group the word-final consonants disappeared without entailing morphological consequences.
The declension of proto-Slavic and its more or less faithful continuation in Old Church Slavonic presents the following main types:
I. The first group includes six subtypes instead of the previous four.
1a. continues proto-Indo-European type 1 with a final hard consonant. Its characteristic endings are: nom. sg. -a, abl.-gen. sg. -y (Ъ); all newly formed endings begin either with back vowels (a, o, ŭ, ū, ў, q), or with front vowels (e, ï, e).
2a. continues proto-Indo-European type 2 with a final hard consonant. Its characteristic endings are: acc.-nom. sg. masc. -u (S), acc.-nom. sg. neut. -o, abl.-gen. sg. -a; as in type ia, its endings begin with either a back or a front vowel.
1b. continues proto-Indo-European type 1 with the final stem consonant j. Its endings are: nom. sg. -a, rarely, -i (И), abl.-gen. sg. -ę (Δ); the endings begin with the back vowels (a, ą, u), or the front vowels (ǐ, ǐ, e, ę). The front vowels of certain endings developed from back vowels under the influence of a preceding j.
2b. continues proto-Indo-European type 2 with the stem final consonant ƒ. Its endings are: acc.-nom. sg. masc. -i (И), acc.- nom. sg. neut. -e, abl.-gen. sg. -a; as in type lb, they begin with either a back or front vowel.
3. continues proto-Indo European type 3. Its endings are: nom. sg. -ŭ (Ъ), abl.-gen. sg. -ū ( םy), dat. sg. -ov-i (םBИ), nom. pl. - ov-e (םBE ) ; in general all newly formed endings begin with back vowels (ü, ü, o, y).
4. continues proto-Indo-European type 4. Its endings are: nom. eg. i ( 6), abl.-gen. sg. -í; they all begin with a front vowel.
Proto-Slavic type 3 contains only masculine nouns, and type 4 masculine and feminine nouns. Types ia and lb have mostly feminine nouns and a small number of masculine nouns. Types 2a and 2b remain, as before, masculine and neuter.
Even in proto-Slavic the nom. sg. forms had become associated with particular genders: a with feminine, ŭ (Ъ) with masculine, and o or e with neuter.
II. The second, consonantal group of proto-Indo-European stems relinquished many individual types. Proto-Slavic retained the following types:
All these types share the ending -e of the abl.-gen. sg.
The old morphological relationships were retained in this group with minor modifications.
If we now compare the two major proto-Slavic groups I and II with the corresponding proto-Indo-European groups, we obtain the following features characteristic of both groups.
|Group I:||Group II:|
|Nom. sg. = stem + ending||Nom. sg. < stem|
(shorter than the stem)
|The ending of the abl.-gen. sg. is phonetically the continuation of either a diphthong or a long vowel; morphologically it consists of a part of the old stem plus the old ending.||The ending of the abl.-gen. sg. is phonetically the continuation of the short vowel e (es); morphologically it is the old ending itself.|
The distinctive features between proto-Indo-European and proto-Slavic are:
|In group I||In group II|
|there occurred a morphological reduction of stems in favor of endings.||there occurred only phonetic reduction of whole case forms, while the morphological boundaries between the stems and endings remained the same.|
This period witnessed the development of one generalized type of ending: every Common Slavic ending had to contain at least one vowel, and had to begin with that vowel.
Thereafter we can distinguish three periods in the development of Slavic declensional types:
1) The period of morphological distinctions, discussed above. These distinctions are reflected in Old Church Slavonic monuments and, to some extent, in the oldest monuments of other Slavic languages.
2) The period of redistribution of declensional types according to genders, characterizing almost all contemporary Slavic languages.
3) The period of decline of declensional stems, reached conspicuously in Bulgarian.
The transition from the first Common Slavic period to the second was furthered principally by the loss of the final vowels ŭ (Ъ) and ỉ (Ъ).
This phonetic process introduced, on the one hand, forms with a zero ending and established on the other hand, the phonetic identity of the stem with the nominative case in many substantives of the first group.
The phonetic process, furthermore, prevented the preservation of the old morphological distinctions of declensional types, causing their regrouping and redistribution. One of the principal results of this process was establishment of a connection of particular endings (including the zero ending) with distinctions of gender.
Various semasiological and lexical features now took precedence over purely morphological ones. Thus there was a gradual shift from a purely morphological to a semasiological and lexical distribution of the declensional types.
Along with the new declensional types based on semasiological and lexical distinctions, there continue to exist residues of the old morphological distinctions. In this context it should be noted that:
1) Some old types merged. Original group II, continuing the proto-Indo-European consonantal stems, was absorbed by group I. In general, the original first and second subtypes of group I predominate over the other subtypes. Subtype 3 merged with 2a, subtype 4 with ib.
2) The original, purely morphological differences were utilized for new semasiological purposes. It is enough to mention the distribution of the endings -a and -u (y) in the abl-gen. sg. in Russian, Polish, and other Slavic languages. These endings come from types 2 and 3, but are now used to convey different meanings. Similar distinctions are attached to the various endings of the nom. pl. which differentiate between the personal and nonpersonal, animate and inanimate genders, and so on.
3) In addition, there evolved purely morphological distinctions, but along different principles than in the old declensional types. Thus, for example, several Slavic languages distinguish declensions with final hard and soft consonants, with stems ending in consonants which are modified according to meaning or case (according to psychological association) and stems ending in consonants which are not so modified. In some stems the vowel preceding the final consonant or consonant cluster remains the same in all forms (of the declension) without undergoing psychologically determined changes, and in others the final vowel is subject to change (e.g., Polish rόd - rodu, Czech buh - boha, Slovenian bug - bogá, Ukrainian nič - nočy). In languages with short and long vowels the stems display quantitative alternation: some stems retain the same vowel in all forms, while in other stems there appears a short or long vowel depending on the declensional form.
In all Slavic languages the stems may alternate as a result of the split of the old vowels ŭ (Ъ) and ĭ (Ъ) into zero and full vowels; one and the same stem may appear with or without the vowel (e.g., Russian son– ║ sn-a, pen ║ pn-ja. . . . , Polish pień ║ pń-a. . .). Finally, in languages with a morphologically determined stress, there are stems with a fixed stress, as opposed to stems with a mobile stress.
The development of stems reached its extreme point with the full merger of declensional types, the fusion of stems and endings into a single morphological unit, the appearance of one general case (1 casus generalis) and, where the need to retain the inflection remained, the replacement of all endings by syntactically separable suffixes and case determinants, i.e., prepositions. This stage of development is tantamount to the complete disappearance of the category of declensional stems in the old sense and the appearance of a new kind of stems. These new stems combine not with endings, but with prefixes which are at first syntactically separable (in free combination) and later on syntactically inseparable from the stems (bound combinations and the germ of new declensional forms based on new principles).
Among the Slavic languages, Bulgarian has almost reached the stage outlined above. In that language the only distinctions preserved are those of the singular and the plural and the continuation of the old vocative case with a special ending, i.e., the case which in proto-Indo-European had only a bare stem without any formal marker, i.e., without an ending.
The transition from a synthetic or centralized to an analytic or decentralized morphological structure has been accelerated to a significant degree by the mixture of ethnically different tribes.
1. I wish to remind the reader that this article was written in 1870, at a time of the absolute dominance of the theory of three original vowels, a, i, u, which lay at the basis of Schleichers Compendium.