Open and Closed e in Latvian A Synchronic Approach

This problem is traditionally dealt with by a diachronically oriented formu-
lation which sees the occurrence of open or closed e as depending on the
nature of the following vowel or diphthong (closed front or other) or conson-
ant (palatalized or other). From a purely synchronic point of view this
formulation is open to a number of objections:

(a) one has frequently to refer to previous stages of phonetic development
to determine the earlier nature of a following vowel, e.g. tēvs < *tēvas;
(b) no account is given of vowels in final syllables, e.g., puķe;
(c) no account is taken of loan-words, e.g., bibliotēka;
(d) the loss of the phoneme [r] partially invalidates the distinction
between palatalized and other consonants, e.g., dzeru.

It is also misleading to the extent that it suggests a purely environmental
explanation, whereas the facts would seem to point more to a combination
of environmental and categorial, as far as contemporary usage is concerned.
The present formulation attempts to meet objections (a), (b) and (d) above,
and to a lesser extent (c) also, although clearly a number of problems re-
main here.

Nouns and adjectives

The quality of an e in the final syllable of the stem is determinable by the
environmental rules from the form of the accusative singular:

zēnu, sētu, mežu, ķešu, dēli, mēli,
akmeni, medu, telti, pelēku, svešu

The e of a non-final stem syllable is similarly determinable:
tecīlu, cepuri.

In addition, we must specify that any e occurring in a noun termination
is closed.
Once established from the accusative singular, the quality of an e remains
unchanged throughout the declension, and in the case of adjectives, through-
out the degrees of comparison: hence bērni, bērniem, etc. One exception in-
volves feminine nouns in -s (c f. debesi, acc. sing.) where the expected pala-
talization of the final stem consonant in the genitive plural does not occur:
a stem e in such cases is often open, e.g., debesu, Cēsu, although an analogical
e is perhaps more common.
The relation between the accusative singular and the prediction of e
quality throughout the declension suggests at once that in general terms
each type of declension has its prevalent pattern, and that the declensions
are groupable into two types: zēns‚ sēta, medus on the one hand, dēlis‚, mēle,
akmens, telts on the other. In syllables other than stem finals, e behaves
appropriately under the influence of what follows. The possibility that the
former pattern and the latter may sometimes conflict must be seriously
considered on the evidence of such pairs as vēsture and vēsture, etc., debesu
and debesu, etc. The pair spilvens and spilvens further auggests analogy across
declensional borders (cf .akmens), a possibility reinforced by the use of
-tiņš rather than -iņš as diminutive suffix for such nouns as spilvens and
Diminutives cause no special problem, and are readily explained by the
procedure outlined above. Similarly, most derived nouns:sētnieku, tēl-
niecību, etc. However, nouns formed with the prefixes bez, pret and ne are
exceptional, and will be considered later.
It should be noted that any noun or adjective in -erš ( <erš) would cause
difficulties: we have not found such a type.
As a general rule, we may state that in foreign words e is closed in spite of
environment: bibliotēku‚ šefu, ar(c)hitektu, ektu, etc. However, when the e is
followed by r + consonant, e is often heard: laternu, personu, ercenģeli. While
this may be due to the pronunciation of the language of origin, often Ger-
man, other processes may be involved, and cannot be ruled out a priori. At
this stage we can only suggest these as general tendencies: closer study of
this aspect of the problem is needed before any definitive answers can be


In cardinals e is always closed, in spite of phonetic environment: četrus,
sešus, septiņus
, etc. In all other numerals environment is a reliable guide:
ceturto, sesto, septīto, četrējus, etc.


These exhibit only e: ,es, mēs, tevi, sev, etc.


It appears that all monosyllabic prepositions and postpositions have e:
bez, zem, pret, pēc, dēļ. When used as noun prefixes, bez and pret are not
subject to environmental influence and retain e: bezdarbība, pretstats. Similar
immunity is shown by pret in the verb pretoties.


All monosyllabics seem to have e: bet, jeb, ne ... ne, nedz; so too perhaps for
bisyllabics other than those in -mēr: neba, nekā , but kamēr.


Monosyllabics seem to have e, e.g., vēl, še, te, sen, although pērn is a notable
Derived adverbs retain the e quality of the adjective, noun or preposition
on which they are based, e.g., lēni, lēnām, vēlu; piemēram; pēcāk. As for
conjunctions, the termination -mēr has ē: tomēr, vienmēr. Diminutives follow
environmental rules, e.g., lēnītiņām.

Particles and interjections

These appear on the whole to have e: jel, re, nez, nē, ne. Like bez and pret, ne
used as a prefix is not subject to environmental influences, cf. neass, nelāga,
egcept when immediately followed by e or ē: neesmu, neesam, neērts.


As it was possible to ascribe dominant quality patterns to nouns on the basis
of declensional type, so a similar division is useful for verbs, but on the basis
of tense rather than of conjugational type.
(a) Infinitive. E is here closed, unless the environment requires e: cf.
celt, nest, redz‚ redzēt, meklēt, but melot, lēkāt.
(b) Present. E is open, unless the environment requires e. Concerning the
environment, however, it must be noted that r counts as a "closing" con-
sonant for First Conjugation verbs only, that the termination (cf. meklē)
is closed, and that First Conjugation verbs lacking -i in the second person
singular must be deemed to behave as if an -i were present. Thus redzu,
meloju, nesu, but meklēju, ceļu, dzeru; redzi, (tu) nes, meklē‚, cel, dzer, but
melo; redz, (viņš) nes, but meklē, ceļ, dzer. We leave aside here verbs like pētīt
which are undergoing a change of conjugational type.
(c) Imperatives. The same rule applies (with the same additional notes
as for the present, e.g., redzi, redziet, nes, celiet, but melo, melojiet. Speakers
who do not follow the standard -at/-iet distinction between the indicative
and the imperative will have corresponding differences in the quality of e.
(d) Imperfect. In the First Conjugation, e is always closed; in the other
conjugations it is closed unless the environment requires e: metu, cēlu,
redzēju, meklēju, but meloju, lēkāju.
(e) Future. E is closed unless the environment requires e: metīs, cels,
redzēs , meklēs, but melos.
(f) Conditional. E is open unless the environment requires e: mestu
celtu, redzētu, meklētu
, but secinātu.
(g) Past participle passive. E is open, unless the environment (excluding
inflections) requires e: celts, rdzēti, but medīts. Cf. the rules for adjectives
(h) Past participle active. E is open unless the environment requires e:
cēlusi, dzērušu
, but redzējis, skrējusi.
(i) Present participles in -ošs and -ot. The quality of e is as for the first
person singular of the present: tekošs, redzot, but dzerot.
(j) Present participle in -am(s). The quality of e is as for the first person
plural of the present: redzam, melojam, but dzeram.
(k) Present participle in -dams. E is open unless the environment requires
e: nesdams, mezdams, redzēdams, but medīdams.
(I) Conjunctives. The present conjunctive takes its e quality from the
first person singular of the present indicative; the future conjunctive from
the first person singular of the future indicative: redzot but dzerot, dzeršot
but melošot.
(m) Debitives. The quality of e is as for the third person of the present
indicative: jānes, jāredz, but jādzer, jāmeklē.
The above considerations hold for reflexive as well as for active forms.
The negative particle ne (q. v.) has its own rules, independently of the verbal
form to which it is attached.
Clearly, the various verbal forms above fall into two major groups;
open unless necessarily closed and closed unless necessarily open. However,
the forms of the present and imperfect, with their rather more complex
behaviour, prevent such generalizations from being totally valid, although
the ad hoc adjustments that are required to handle these two tenses can be
simply stated, as in (b) and (d) above.
But even though a combination of the environmental and categorial
approaches can provide a better synchronic description than can the
environmental approach alone, certain difficulties remain. Nothing short
of an exhaustive list of adverbs, prepositions, conjunctions and interjections
can provide a basis for generalization about these classes. The problems
arising from verbs undergoing a change of conjugation represent additional
complications. More seriously, there appears to be no non-diachronic way
of distinguishing foreign loan-words.
The first of these is, in principle, easily resolved. With the passage of time,
the second may well resolve itself, but, unless widespread analogical changes
bring foreign loan-words into line with native patterns, the third is likely to
represent a very substantial stumbling-block to any description of the
Latvian e phonemes fully compatible with the exigencies of synchronic

T.G. Fennell