The Žemaitian Accent Retraction as a Typological Parallel to Olander’s Mobility Law

ABSTRACT

In his 2009 monograph, Thomas Olander presented a new solution to the old problem of the origin of Balto-Slavic mobile accent paradigms. According to his “mobility law,” pre-Proto-Balto-Slavic words accented on the final mora became phonologically unaccented and automatically received ictus on the first syllable of the phonological word, as in *longós > *ˌlāˀngas > Lith. lángas, CS *lǫ̑gъ. The theory is supported by two typological parallels from contemporary Slavic dialects (Podravina and Zaonežje). However, a parallel can also be found in the Baltic material: the Žemaitian accent retraction, a phenomenon well described by Lithuanian dialectologists, shares some features with Olander’s mobility law and the above-mentioned parallels in Slavic, such as a peculiar phonetic realization of the retracted accent, retraction to the first syllable of the phonological word, partial neutralization of tonal oppositions under the retracted accent, and possible origin due to language contact. Moreover, a closer examination of the irregularities in the Žemaitian process could shed some more light on the origins of at least some of the exceptions to the mobility law, as well as of the somewhat different outcomes of this law in Baltic and Slavic.

KEYWORDS:

• Balto-Slavic
• Žemaitian
• historical accentology
• accent retraction
• mobility law

Notes

1 The research presented in this article received funding from the Research Council of Lithuania (LMTLT), Agreement No. S-LIP-18-34. I am indebted to Mikhail Oslon, Steven Young, and the anonymous reviewers for their valuable comments and editorial suggestions.

2 In this paper, the term accent is used in the sense of a phonological prominence of one syllable or mora (i.e., part of a syllable) in a word, without reference to phonetic realization, i.e., with no distinction between stress and pitch accent (cf. similarly, Olander 2009, 11). Metonymically, the term may also be applied to a morpheme that contains an accented syllable (or at least its nucleus).

3 A phonological word may be defined as an excerpt of the utterance defined by one (primary) accent. It may consist of either a form of the lexical word only (e.g., Russ. gólovu "head [acc. sg.]") or a form of the lexical word accompanied by clitics (e.g., Russ. na‿gólovu or ná‿golovu "on[to] the head").

4 Enclinomena are usually defined as “phonologically unaccented” word-forms (i.e., forms without distinctive accent) that receive an automatic (i.e., non-distinctive) ictus on the first syllable of the phonological word (including proclitics), e.g., PSl. acc. sg. *gȏlvǫ, *(nȃ‿)golvǫ. I avoid the notion of “phonological unaccentedness” here for the reasons explained in Section 4.2 (see especially ftn. 32).

5 For the sake of simplicity, mobile consonant-stem nouns are not considered in this paper.

6 Cf. Kim (2002, 130ff.) and Babik (2012) for a criticism of Illič-Svityč’s theory.

7 Cf. Nikolaev (2012) or Kapović (2015) for more recent publications advocating this view.

8 Cf. Jasanoff (2017) for the most recent theory of this type.

9 Olander (2009, 8) defines desinence as a complex of PIE stem-forming suffix and case-marker, e.g., PIE o-stem acc. sg. *-o-m. In this paper, I maintain this term for non-Balto-Slavic or pre-Balto-Slavic forms only. For Balto-Slavic, the term ending is preferred.

10 For criticism, see Kortlandt (2010) or Jasanoff (2017, 114f.).

11 Olander’s notation with the following marks has been maintained in the quoted examples: ˈ (accent on the following syllable or mora), ˌ (automatic ictus on the following “phonologically unaccented” syllable), ˀ (glottalization, i.e., the precursor of the acute tone of the preceding vowel).

12 See Pronk (2018, 557ff.) for an updated overview of the Podravian data.

13 See Ter-Avanesova (1989) for a more exhaustive description of Zaonežje accent retraction.

14 Cf. also maps in Zinkevičius (1966, 447); LKA (vol. 1, Map 1; vol. 2, Map 105); Bacevičiūtė et al. (2004).

15 For more details, see the following selected bibliography on the Žemaitian accent retraction: Grinaveckis 1961, 118ff.; 1973, 39ff.; Zinkevičius 1966, 38ff.; Girdenis 1967, §§ 8f., 37ff.; 1982; Pabrėža 1981; 1984; 1986; 1987; Mažiulienė 1996, 40ff.; Murinienė 2007, 157ff., 198ff.; Girdenis, Murinienė 1998; Murinienė, Girdenis 2001; Atkočaitytė 2000. Cf. also Young 1991, 26ff. for a brief description in English.

16 A characteristic feature of Žemaitian is its various forms of secondary accents (cf. Zinkevičius 1966, 42ff.; see Hock 2015, 16 for a recent interpretation of secondary accents as “pitch phenomena in syllables not bearing the ictus”). Since secondary accents are possible only on posttonic syllables, in all Žemaitian examples with several diacritical marks, only the first syllable with a diacritic is to be interpreted as accented.

17 As well as some South Žemaitian dialects (see Atkočaitytė 2000); cf. ftn. 35.

18 The Proto-Baltic acute is phonetically realized as a broken tone in Žemaitian and marked with the diacritic ^ in transcription.

19 In the transcription used by Lithuanian dialectologists, the letters ọ and ẹ denote the closed o and e vowels, respectively; superscript letters stand for reduced vowels while superscript letters in parentheses mark a recent loss of the vowel (often with traces preserved in the pronunciation of adjacent consonants). For the meaning of ˢ and other prosodic diacritics, see Section 4.1 (including footnotes).

20 Ter-Avanesova’s knowledge of the Žemaitian retraction is based only on Laučiūtė’s 1979 paper on Žemaitian prosody, which lacks information about retraction to proclitics.

21 Examples of an utterance-final retraction in some Čakavian dialects, presented by Hock (2015, 114f.), although they confirm his finality effect, reflect a different kind of phenomenon compared with those discussed in this paper because in these dialects, the accent is retracted from the final to the penultimate mora rather than to the initial one.

22 For criticism, see Kortlandt 2019.

23 Explanation: *: place of accent (ictus); L: low pitch; H: high pitch; μ: mora; σ: syllable (see ftn. 25); šakà: standard Lithuanian spelling; [šàkà]: Žemaitian spelling (Lithuanian dialectological transcription).

24 Acute syllables are usually characterized, among other features, by a more abrupt change of pitch (independently of its contour), a higher F₀ amplitude, and a shorter duration than the circumflex ones.

25 Cf. also an alternative auto-segmental interpretation of Lithuanian (southern West Aukštaitian) tones, presented in Dogil (1999b), which combines moraic and syllabic segmentation in one tier, on the basis of the author’s assumption that only the “acute accent in Lithuanian has a strong moraic status, whereas the circumflex accent seems to be implemented on a larger unit, possibly the syllable”; see Dogil (1999a, 283). If we apply this interpretation to North Žemaitian, the above-listed examples would look like (i) and (ii):

Table

26 Perhaps with the only exception of Girdenis (1967), where the retracted accent is marked with a diacritic ' before a stressed syllable, independently of its quantity, e.g., 'šọlẹ̀nẹ̀ ‘of a stave (adj. instr. sg. fem.)’, 'arklinĩŋkùs ‘stables (acc. pl.’) (cf. the regular accent in šọ̀lẹ̀nẹ̀ ‘well [instr. sg.]’, árklinĩŋkùs ‘stablemen [acc. pl.]’).

27 Murinienė (2007, 72, 158).

28 If accented with the regular accent, short vowels may become lengthened under certain conditions (typically in forms without secondary accents) and marked with the diacritic ˢ, e.g., prı̾nẹš^ẹ ‘brought (praet. 3)’ (cf. standard Lith. prìnešė). Under the retracted accent that usually triggers the appearance of secondary accents in post-tonic syllables, such conditions simply do not occur.

29 Murinienė 2007, 167. Note that in some dialects, a specific allotone of circumflex may be marked with the diacritic ˢ (cf. Zinkevičius 1966, 36).

30 This is not to be confused with other uses of ˢ (see ftn. 28 and 29).

31 Cf. Trubetzkoy 1939, 29 [1971, 27].

32 One should not get a false impression that I try to question the reconstruction of “phonologically unaccented” forms (i.e., forms without distinctive accent) in Proto-Balto-Slavic. I simply find the notion of “phonological unaccentedness” not very successful, for at least two reasons. First, not every scholar would identify the terms phonological and distinctive, and even Roman Jakobson, who was the first to reconstruct “phonologically unaccented” forms for Proto-Slavic, noted in his famous paper on Proto-Slavic accent that he used these two terms in the same meaning only for the purpose of that particular article, but he would otherwise prefer using the term phonological in a broader sense, i.e., including not only distinctive but also all other linguistic functions in its definition; see Jakobson (1963 [2002, 671]). Second, even if we make no difference between the terms phonological and distinctive, the term accent (as defined in ftn. 2) still may be applied for both a distinctive (“phonological”) and a non-distinctive prominence of a syllable (otherwise, we would not use this term speaking about languages with fixed accent, where accent is always non-distinctive, but we still call it accent) and the term delimitative accent is simply less complicated and less confusing than ictus on a phonologically unaccented syllable. Proto-(Balto-)Slavic syllables bearing this kind of accent most probably phonetically differed from other “phonologically unaccented” syllables, and there is no need to stress the fact that their accent was non-distinctive; cf. Jasanoff (2017, 56) for similar thinking. Note that the retracted accent of Žemaitian is also sometimes called a “non-phonological accent” by Lithuanian dialectologists; see Girdenis (1967, § 39), Murinienė (2007, 158f.); cf. Girdenis (2014, 24) for criticism.

33 “[…] CŠŽT skiriamos ir atitrauktinio kirčio priegaidės. Jas diferencijuoja visi prozodiniai požymiai (kiekybė, intensyvumas, tonas), tik, žinoma, jie ne tokie ryškūs, kaip esant pagrindiniam (senoviniam) kirčiui.”

34 Or, alternatively, the opposition of glottalized vs. non-glottalized vowels that gave rise to tonal oppositions in later Baltic and Slavic languages.

35 Retraction to the preceding syllable is usually called “Aukštaitian” whereas retraction to the initial syllable is called “Žemaitian” by Lithuanian dialectologists. Note that the border between areas of these two retraction types does not fully coincide with the “official” border between Aukštaitian and Žemaitian. For cases of the Žemaitian type of retraction in Aukštaitian and vice versa (often with both types present in a given dialect), see, e.g., Jonynaitė (1959, 112f.), Zinkevičius (1966, 39), LKA (vol. 2, 123), Atkočaitytė (2000, 154), Bacevičiūtė et al. (2004, 208).

36 Cf. Bacevičiūtė et al. (2004, 226).

37 Note that the acceptance of the influence of substrate languages does not deny the role of other, inner-Lithuanian factors for the origin of retraction, e.g., vowel reduction in unaccented endings, for which see Girdenis (1981, 23; 1982) and Pabrėža (1984; 1986).