This is a re-typed version of “On the Structure of Proto-Uralic” found at https://helda.helsinki.fi/bitstream/handle/10224/4083/janhunen23-42.pdf?sequence=1
Uralic (U), with its two main branches Finno-Ugrian (FU) and Samoyed, is one of the most thoroughly investigated language families. Moreover, the Proto-Uralic (PU) that can be reconstructed on the basis of Common Uralic (CU) comparative material is probably the most ancient unambiguously established parent language in Eurasia. While Proto-Finno-Ugrian (PFU) seems to have existed contemporaneously with some of the early forms of Indo-European. Proto-Uralic must lie considerably farther back in time. Therefore, a condensed review of the basic structural characteristics of Proto-Uralic should be of interest even outside the field of Uralistics. 1
PU possessed at least 16 distinctive consonant units:
labial dental cacuminal palatal velar - plain stops affricates sibilants nasals spirants laterals vibrants glides
1For the impetus for writing this article I am obliged to the Indologist, Dr. Asko Parpola.
Additional units have been proposed on the basis of uncertain etymologies. Indeed, the reconstructed paradigm of 16 phonemes is a minimum comprising only the high and medium frequency consonants of theproto-language, a number of low frequency units remaining technically unreconstructable because of the scarcity of etymological material. Some of the main problems in further elaboration of the reconstruction concern the status of the affricate (*c), the palatal series (*ś, *ń, *ð’) and the spirants (*ð, *ð’).
The affricate *c probably differed from the dental stop *t both in release (affrication) and in place of articulation (cacuminal). It is not clear which of these two phonetic features was phonologically more important, and didalectal differences may also have complicated the picture. If, however, it was cacuminality that was the distinctive characteristic, then it would be plausible to postulate other cacuminal phonemes as well. A cacuminal (retroflex) sibilant *š, at least, was a distinctive unit in PFU, but it cannot be reliably traced using CU etymological material.
Similarly, for the sake of system symmetry, additional phonemes could be proposed in the palatal series. A couple of uncertain etymologies suggest the reconstruction of a palatal lateral *l’ but even other palatal phonemes (stop, affricate) may well have existed in PU, though they must have been of low frequency.
Synchronically the status of the “spirants” *ð and *ð’ appears to be problematic. These phonemes may have originally been related to either the dental stop or the liquids.
The maximal (first syllable) vowel paradigm most probably comprised 8 qualitative units:
back front labial illabial labial illabial high middle low
Only minor problems arise in the qualitative reconstruction, first and foremost the status of the low back vowel (labial *å or illabial *a) and of the illabial back vowel (high *i̮) or middle *e or low *a). The possibility of dialectal differences cannot be eliminated.
The traditional cause of uncertainty and dispute in reconstructing the PU vowel system has been the question of quantitative correlation. Two different phenomena seem to be related to the problem of quantity. First, in Finnic there is a systematic non-combinatory quantitative correlation that can be traced back (as a quantity phenomenon) to Proto-Finno-Permic. Second, Proto-Samoyed had in addition to ordinary single vowels a system of vowel sequences which consisted of any full vowel of the paradigm followed by an invariable reduced vowel segment. A few reliable etymologies, at least, support the hypothesis that there is indeed a systematic correspondence between these two vowel phenomena: Finno-Permian quantity vs. Samoyed vowel sequences. How this correspondence should be interpreted from the point of view of PU is not clear. It seems however, that quantity, as such, was not a distinctive feature of the U proto-language; the question is linked with both the vowel and consonant systems.
Another problem related to both quantity and quality is the question of reduced vowels. In Proto-Samoyed, a reduced vowel *ə̑ was a distinctive unit of the vowel paradigm, and it is possible that it should be derived from an independent PU phoneme. The problem is complicated, however, by the fact that some of the occurences of the Proto-Samoyed vowel seem to be explainable in terms of combinatory phenomena.
The PU syllable was simple in structure, consisting of a vowel nucleus and optional surrounding consonants: (C)V(C). The glides probably had a special status and could join the vowel nucleus without affecting the syllable structure (V = V, Vj, Vw). A word root (free morpheme) consisted of either one or two syllables and always ended in vowel: (C)V, (C)V(C)CV. Derivative elements and grammatical morphemes could be added to the roots by means of suffixation. These suffix elements (bound morphemes) consisted syntagmatically of either a single consonant, -C, or a whole syllable, -CV(C). Suffixes of a more complex structure also occurred: -CCV; these were originally mainly combinations of suffixes.
The occurence of of consonant phonemes in the syllable and within the word was restricted by only a few syntagmatic rules. The velar nasal *ŋ and, in view of the lack of relevant etymological material also the vibrant *r and the spirant *ð never occured word-initially. The spirants *ð and *ð’, as well as the palatal nasal *ń have not been attested in syllable-final position. At the boundary of two, syllables, the most typical consonant combinations were those of obstruent + obstruent and nasal + obstruent, but many other types of combinations also occured. Among the unrecorded and probably syntagmatically impossible combinations were those of the type obstruent + sonorant. Also, no reliable evidence exists for combinations of two identical segments (geminates).
Two important phonotactic restrictions governed the occurence of vowels in non-first syllables. First, vowel harmony - one of the typological features of many of the present-day U languages - allowed a vowel to contain either back vowels or front vowels. Second, the actual paradigm of vowels in non-first syllables was limited to three units: the two low vowels *å and *ä and a higher vowel. A major problem in reconstruction concerns the paradigmatic identity of the non-first syllable high vowel. According to Finno-Permian evidence this vowel seems to have been identical with *i of the first syllable (though traditionally written as *e) but the eastern groups, the Samoyed languages in particular, suggest the reconstruction of a more neutral phoneme. Indeed, it seems preferable to avoid identifying the unit with any of the qualitative units in the maximal paradigm and to use instead a distinct, *ə (neutral reduced vowel) for the segment in PU. As a result, the rules for vowel occurence bring the number of possible vowel combinations in the first two syllables down to 16:
first syllable u, o å, i̮ ŭ, e ä, i
second syllable å } ä } ə
A consequence of the phonotactic restrictions is that only two distinctive stem types could occur in the word roots: å/ä-stems and ə-stems. A characteristic of the latter is that the stem final high (reduced) vowel ə could alternate with zero before suffixes comprising a whole syllable provided the restrictions on consonant distribution were not violated. The phenomenon (the so called consonant stem) is actually one of the very few morphophonemic alternations that can be reconstructed in PU.
Another consequence of the restrictions in vowel distributions is that in suffixes containing a low vowel in the lexicon, the vowel segment could be realized either as a back vowe (*å) or as a front vowel (*ä) depending on the vocalism of the root.
Vowel distribution is interrelated with a non-distinctive suprasegmental characteristic of the U languages: the initial stress. The PU stress pattern divided the word in two-syllable sections with initial stress, with the main stress on the first section of the word: (C)V́(C)CV/CV̀(C)CV(C)/. This phenomenon is best preserved on the periphery of the language family (Finnic-Lapp, Samoyed), where it has convergently led to important phonotactic and morphophonemic developments (esp. so called “consonant gradation”).
Parts of speech
The most obvious material characteristic dividing words into functional classes in PU seems to have been the distinction according to the number of syllables in the word root. Basic deictic elements, used in pronominal and auxiliary functions, formed a class of monosyllables, whereas the bulk of the lexicon, the so called ”notation words” were disyllabic. In grammatical behaviour, however, no basic difference existed between two groups, except that in the paradigms of some pronouns there was suppletion.
Using morphological and syntactical criteria, two parts of speech, the noun and the verb, can be distinguished in PU. it is true that rather abundant evidence suggests that the distinction had been somewhat less strict in Pre-U. In fact, several PU derivative and inflectional suffixes could be affixed to both nominal and verbal stems. Also, there existed a small group of word roots, the so called nomenverba, which could morphologically and syntactically act both as nouns and verbs, in semantically closely related functions. However, in PU most of the lexicon was already unambiguously divided into nouns and verbs, and both parts of speech did have a range of morphological and syntactic characteristics of their own.
As for further classification, morphological criteria are not sufficient to serve as a basis for distinguishing any subclasses (such as “adjectives” and “numerals”). Also, no evidence exists of any separable group of indeclinable words (“adverbs”). For instance, space relationships were expressed by regularly declined spatial nouns, used both independently and in postposition constructions (as nominal postpositions). Undoubtedly, however, there existed in PU some kind of extra-grammatical group of utterances (“interjections” and the like).