Volume 41, No.2 - Summer 1995
Editor of this issue: Antanas Klimas, University of Rochester
ISSN 0024-5089
Copyright © 1995 LITUANUS Foundation, Inc.

Marta Rudzīte. Latviešu valodas vēsturiskā fonētika (The historical phonetics of the Latvian language). Riga: Zvaigzne (1993) pp. 382.

In the foreword (pp. 3-6) the author writes that the book is intended for students either in regular or correspondence divisions who must have a course in the historical grammar of Latvian. The object of the book is to help the student master the fundamentals of Latvian historical grammar and to deepen his understanding of his native language and to further his professional preparation.

In the introduction the author explains (p. 31) that the task of historical phonetics is not to explain the articulatory or the acoustic characteristics of the sounds, but rather to explain which sounds have remained unchanged from the older epochs of the proto-language and which sounds have developed later and in what circumstances. In order to illustrate the conservative nature of the Baltic languages Rudzīte (p. 35) gives the reconstructed word *saūsos 'dry' with the descendant forms in several daughter languages, viz., Lith. saūsas, Latv. sàuss, Old Slavic suxŭ, Gr. haŭos, Middle Low German sōr (cf. English sere). Since the Baltic languages are more conservative than the old dead Indo-European languages, it seemed possible to Endzelins that they remained at the center of the Old Indo-European area.

Next Rudzīte gives a brief review of the sources for the history of the Latvian language, viz. (1) the contemporary Latvian language, (2) old written Latvian writings and (3) related languages (p. 40). In the contemporary Latvian language we find such cognate words as šķelt 'to split' and skaldīt 'id.' on the one hand and šķirt 'to separate' and skara 'panicle'. From this it is clear that we have complementary distribution with sk occurring before non-front vowels and with the sequence šķ arising before front vowels. This is, of course, the well-known method of internal reconstruction. Old writings are important also, but for the explanation of the history of Latvian they must be used with certain restrictions because they are of a late date and with the exception of J. Reiters and G. Elgers their authors were not Latvians. As is well known the first book in Latvian is the Catechismus Catholicorum... which was printed in Vilnius in 1585 (p. 44). The language of the 16th century writings is close to the contemporary central dialect or the standard language. But the orthography based on the Middle Low German way of writing does not completely reflect the sounds of the Latvian language. Rudzīte quotes Endzelīns (Latviešu valodas skaņas un formas) to the effect that a final orthographic e very frequently stands for a, i or u and that the sibilants are not clearly distinguished in the orthography from the shibilants or the affricates. The sibilant s is not clearly distinguished from its voiced counterpart z, nor is the shibilant š clearly distinguished from its voiced counterpart ž, consonantal softening is only rarely marked. In the morphology we encounter forms that today are encountered only in dialects, e.g., the dative plural in -ms (cf. -ms in modern Lithuanian, -mus in Old Lithuanian). We still encounter the illative as a case, cf., e.g., Tew hues touwe Thewe unde mate cenan turret' thou shalt honor thy father and mother (lit. hold in honor)' (p. 45).

In the first half of the seventeenth century the dictionary was published by G. Mancelius, Lettus, das ist Wortbuch / Sampt angehengtem täglichem Gebrauch der Lettischen Sprache... (1638) which is important because of the orthographical improvements. Long vowels in the root are indicated by the letter h, and the consonants s, z, š, žz, c, dz, č and are distinguished (p. 46).

Rudzīte writes further (p. 49) that if there is nothing in the dialects or old texts which would help us solve a problem of Latvian phonology, then one must look for evidence in related languages. Thus, for example, there is nothing in Latvian which could definitively show the origin of the diphthong uo. At this point it is most appropriate to turn to Lithuanian and Old Prussian, both of which have a more archaic aspect than Latvian. Let us take, for example, the word uôzuols 'oak' in which the first uo derives from the mixed diphthong *an and the second from the vowel (which has also become uo in Lithuanian). We can determine this when we compare the Lithuanian cognate ąžuolas and the Old Prussian cognite ansonis (=anzōnis). But in addition to the value of the testimony of the cognate languages Rudzīte rightly points out the value of the testimony of the borrowings in the neighboring Finnic languages particularly in the determination of the earliest form of the East Baltic diphthong ie thus Lith. šiẽnas 'hay', Latv. siens, but Livonian āina, Estonian dialect hain and Finnish heinä (p. 55).

It is well known the primordial Latvian intonation system is retained only in part of the central dialect where we have the falling intonation (e.g., Latv. [acc. sg.] rùoku 'hand' beside Lith. rañką), the sustained intonation (e.g., liẽpa 'linden tree' beside Lith. líepa), and the broken intonation (e.g., dzîvs 'living' beside Lith. gývas). Rudzīte furnishes a very handy chart on p. 110 showing that the falling intonation has merged with the broken intonation in the Livonian dialect and that the sustained intonation has merged with the falling intonation in the High Latvian dialect.

Rudzīte (p. 85) quotes approvingly from S. Bernštein and V. Kiparsky to the effect that from the functional point of view the sounds should be studied with regard to their place in the system, rather than individually. She notes (p. 117) that regarding the Proto-Baltic vocalic system there is a divergence of views. She provides one possible reconstruction on the basis of Stang and Zinkevičius (to which I would pretty much suscribe also):

Next she gives the view of Kazlauskas:

   Stressed position     Unstressed position

First I might remark that from the phonemic point of view the vowels in stressed position would not differ from those in unstressed position, since they would be in complementary distribution. But Rudzīte quotes Kazlauskas to the effect that during the course of time the two systems would have become a single system:

As I have frequently written, the long vowel system is typologically extremely unlikely. It violates the typological constraint that vocalic systems do not ordinarily have more back vowels than front vowels. Considerations of both articulatory and acoustic economy are advanced to support this notion. In articulatory terms there is more space in the front of the mouth than in the back and in acoustic terms the distance between the second and third formants is less for the back vowels than for the front vowels.

Therefore I think that Rudzīte is to be applauded for also giving Levin's schema, 1975, 151, which, in my view, presents a much more reasonable and typologically plausible Common East Baltic vocalic system:

In this schema e derives from the monophthongization of *ei and æ derives from Indo-European *ē.

Rudzīte (p. 119) writes that from Endzelins' writings one might reconstruct the following Proto-Baltic vowel system:

Such a system seems typologically plausible, although the long vowel system could be written as a triangle with short and long variants except for *ō which would only have a long variant. Rudzīte mentions that some scholars assume that the long low vowel usually represented as Proto-Baltic *ā was rounded and differed from the Proto-Baltic *ō only by degree of tongue height. Whether *a was accompanied by rounding or not would, of course, have no phonemic significance as long as it was kept separate from the-putative higher vowel. The eventual development of Indo-European etymological *ā to ō (e.g., in Lithuanian) does not afford a reason for the assumption of earlier rounding in proto-Baltic *ā. This can be satisfactorily explained in terms of chain shifts of vowels, see, e.g.. Levin, 1975, 152. Here Levin only considers the Lithuanian evidence, but the Latvian dialect evidence could be analyzed also in terms of chain shifts.

Rudzīte writes further (p. 119) that there are many opinions concerning the diphthongs, particularly the long diphthongs. Since this book reflects primarily the views of Endzelins, Rudzīte quotes him to the effect that the long diphthongs which had been retained in the Baltic proto-language were shortened in unstressed syllables faster than in stressed syllables. There the proto-Baltic diphthong *ōu in unstressed syllable passed to *ou and then to *au, whereas in stressed syllable *ōu turned into East Baltic *ō > uo with the loss of the tautosyllabic u. Thus parallel forms in au and uo were formed, so that we can explain thereby Lith. kaũpas 'heap, pile' and kúopa '(military) company.' Other such doublets include, e.g., dúoba 'hollow,' dáuba 'ravine'; sliuõgti 'to climb up a tree without branches,' sliaũgti 'id.', etc., see Skardžius, 1943, 478-479. I still hold to the view (expressed in 1963) that the au:uo alternation is analogical to the ablaut pattern established with ei/ai:ie, but see also Kuryłowicz, 1956, 123-124 and Karaliūnas, 1987, 149ff. Such a view does not require the assumption of long diphthongs, an assumption which I find highly doubtful.

The Indo-European origin of each vowel along with cognate words in a number of Indo-European languages is given (pp. 122-203), but we also find various conditioned sound changes, e.g., -ivi- < -uvi- exemplified by zivs 'fish' (cf. zuve in Curonian dialects and Lith. žuvìs). Another common change in High Latvian dialects is the dissimilation of u to i if the following syllable contains u, ū or au, cf. Latv. tirgus 'market' beside Lith. turgùs (p. 127).

As Rudzīte points out (pp. 147-148) there is a problem concerning the development of Indo-European *eu into Baltic. This a root such as *leudh- is represented in Latvian by ļaudis, Lith. liáudis 'people' (cf. Old High German liut), but the noun *teutā is represented by Latv. tàuta 'nation,' Lith. tautà (cf. gothic Þiuda). In the first example we have *jau as the reflex of *eu and in the second example we have au. Rudzīte quotes Endzelīns' view that since there were no ablaut variants for the words tàuta, Gothic Þiuda, Latv. làuks 'having a blaze (of horses),' Gk. leukós 'white,' the (unstressed?) eu in position before non-palatal (i.e., non-front) vowels was changed to au before the development of iau from *eu. Rudzīte notes the occurrence of eu in Old Prussian texts, e.g., peuse 'fir tree,' beside Gk. peúkē 'pine tree' and Old High German fiuhta and mentions that Endzelīns and others believe that here we do not have to do with the retention of Indo-European *eu, but rather that in the Pomesanian dialect of Old Prussian *įau passed back to eu. I would argue that the phonemic distinction between *iau and *au lies only in the nature of the preceding consonant. The sequence *įau < *eu can only be preceded by a palatalized consonant and *au by an unpalatalized consonant. Therefore in the sequences *eu and *au in post-consonantal position the contrast between /e/ and /a/ is neutralized. This is, of course, different from the situation of ei and ai, where in proto-East Baltic the distinction between /e/ and /a/ was probably maintained, cf. Latv. teikt 'to tell, to say' vs. taisīt 'to make.' Because of the recent palatalization of consonants by following front vowels in Lithuanian a case could be made for finding Lithuanian /e/ to be contrastive with /a/ only in word initial position, since in Lith. teĩkti 'to give, offer' the initial consonant is palatalized in contrast with the initial unpalatalized consonant of Lith. taisýti 'to repair.' But I think that all of the comparative grammars of the Baltic languages miss the fundamental point that etymological *eu and *au do not contrast their vocalic quality phonemically, but rather the nature of the preceding consonant (which may of course be ;' in word initial position, cf. Lith. jaũ 'already').

A useful feature simplifying information for the learner is the use of a diagram with plusses or minuses in boxes denoting Livonian, Central and High Latvian dialects showing where certain phonetic changes have taken place, e.g., following the heading ę > a; e > a, we find the notation:








This means that the change does not occur in Livonian or Central (Vidus) dialects but that it does occur in High Latvian (Augšzemnieku), e.g., vacs as opposed to standard vęcs 'old,' bàda 2 as opposed to standard będa 'care, trouble' (p. 140, 225). Similarly the passage of ei to ai occurs only in High Latvian dialects, e.g., maîta for standard meîta 'daughter; girl,' kraiss for standard kreiss 'left,' etc. (p. 146, 227).

Rudzīte remarks that in Latvian dialects as in other languages we encounter a tendency for diphthongs to become monophthongs (p. 242). She notes further that in certain of the High Latvian dialects there is a tendency for ie and uo to be monophthongized as ī and ū respectively. She writes further that the monophthongization of ie and uo is more recent than the diphthongization of the old ī and ū the new i (< *ie) and ū (< *uo) are not diphthongized. In my view this is merely a phenomenon of a chain shift. She also quotes A. Breidaks' 1979 view that the Latgalian-Selonian i and ū developed directly from Common East Baltic *ē (< (ei) and *ō (< ō). Breidaks' view seems to me to be better motivated than the usual view. It harmonizes well with the theory presented by Levin, who also does not ascribe the diphthongization of *o and *ē2 to Common East Baltic. This notion does, however, entail the assumption that the Lith. 1 sg. forms in -au, e.g. (pres.) sak-aũ 'I say,' (pret.) lip-aũ 'I climbed,' etc. could not be derived directly from *-ã + úo, *-ẽ + úo, but must derive directly from *-āō, *-ēō (Stang, 1942, 226) or else have some other source as suggested below.

The 1 dual endings, Slavic -va, -vĕ and Baltic (Lith.) -va give us evidence of an original bilabial spirant, perhaps alternating with *m in the 1st person (see Erhart, 1970, 54). There is a chance that in Baltic there existed a 1st sg. ending *-u. The similarity between the Tocharian 1st sg. lak-au 'I see' and Lith. sakaũ 'I say' is very striking (see Schmalstieg, 1975, 170 with references).

The Baltic 1 sg. ending -u could also be compared with the Lydian 1 sg. pres. kantor-u 'I entrust,' kov 'I swear' (see Ivanov, 1963, 156 and Ševoroškin, 1967, 38-41). In Lycian we encounter one verb with a present tense nadau the subject of which is amu T (Neumann, 1969, 389). In theory the Slavic 1 sg. aorist ending and the Hittite 1 sg. preterite ending -un could also figure in this equation, although I am personally inclined to derive the Slavic and Hittite endings from *-om.

Rudzīte gives the commonly accepted Brugmannian schema for the Indo-European consonants (p. 264), although she also mentions the well-known Hopper-Gamkrelidze-Ivanov glottalic theory which envisions a completely different scenario for the development of the Indo-European stop consonants. Probably it is better for the beginning student to learn first the traditional system so that he can use the older texts books. After he becomes familiar with the traditional system, he can then learn the typological reasons for preferring the glottalic theory, although from the point of view of the Baltic languages it hardly makes any difference.

On p. 287 Rudzīte notes that Latvian c derives form *ki or *k before a palatal vowel, thus Latv. 1st sg. pres. lęcu 'I jump' and cits 'other' as opposed to Lith. lekiù 'I fly about' and kìtas 'other.' Although the results are different the phonemic identifications in Lithuanian are the same, viz. the merger of *k plus front vowel with *ki. Because of difficulties in the interpretation of the Old Prussian orthography we can't tell whether similar mergers occurred there are not. What this means, however, is that from the phonemic point of view the merger could be considered a Common Balto-Slavic inheritance, since in Slavic also *k plus front vowel and *ki merge as č, thus, e.g., 1 sg. pres. plačo 'I cry' (< *pzak-j-o>) and 3 sg. pres. pečetь 'he bakes' (< *pek-et-). This could also be true for the voiced counterpart *g plus front vowel and *gj (see p. 290).

Although because of my theoretical stance I would have preferred to see the chain shifts showing the development of entire vocalic systems of the various Latvian dialects presented here, I believe that this book will prove useful for present and future generations of students interested in Latvian historical phonology. I am sure that, like myself, others in the field will have occasion to make use of this eminently practical book. Its author is to be thanked for making available in a simple and easily understandable form many of the details of the history of the Latvian sound system.


Breidaks, A. 1979. Iz istorii Latgalo-Selonskogo vokalizma. Baltistica 15.124-129. 
Erhart, Adolf. 1970. Studien zur indoeuropäischen Morphologie. Opera Universitatis Purkynianae Brunensis Facultas Philosophica. 
Ivanov, V.V. 1963. Xettskij jazyk. Moscow, Izdatel'stvo vostočnoj literatury. 
Karaliūnas, Simas. 1987. Baltų kalbų struktūrų bendrybės ir jų kilmė. Vilnius, Mokslas. 
Kuryłowicz, Jerzy. 1956. Lotewska-litewska monoftongizacja ai, ei > ie. Biuletyn polskiego towarzystwa językoznawczego 15. 113-125. 
Levin, Jules. 1975. Dynamic linguistics and Baltic historical phonology. General Linguistics 15. 144-158. 
Neumann, Günter. 1969. Lykisch. Handbuch der Orientalistik, Erste Abteilung. Der nahe und der mittlere
Osten, Zweiter Band.
Keilschriftforschung und alte Geschichte Vorderasiens, Erster und zweiter Abschnitt. Geschichte der Forschung,
Sprache und Literatur. Lieferung 2. Altkleinasiatische Sprachen. Leiden, Cologne, E.J. Brill. 
Ševoroškin, V.V. 1967. Lidijskij jazyk. Moscow, Nauka. 
Schmalstieg, William R. 1963. Lithuanian doublets in -uo- and -au-. The Slavic and East European Journal 7.291-292.
 - - - 1975. The Baltic first person singular ending -u. General Linguistics 15. 168-175.
Skardžius, Pranas. 1943. Lietuvių kalbos žodžių daryba. Vilnius, Lietuvos Mokslu Akademija. 
Stang, Chr. S. 1942. Das slavische und baltische Verbum. Oslo, Jacob Dybwad.

William R. Schmalstieg 
The Pennsylvania State University