1 Introduction

Modern Hungarian shows two kinds of vowel harmony patterns: backness harmony and rounding (or labial) harmony. These have been described and analysed in a number of works over the past decades (for a survey of the phenomena and the literature and for an analysis Siptár & Törkenczy 2007, 63–74). Data representing both types of harmony are given in (1).

Diachronically, backness harmony is the older of the two patterns, going back on comparative evidence most probably as far as Proto-Uralic (Aikio 2022; Suomi 1983). Labial harmony, by contrast, seems to have emerged only in the Old Hungarian period (10th ‒ early 16th centuries AD). How labial harmony appeared in the language is a question treated elsewhere in greater detail (Cser, Oszkó & Várnai 2023b and to appear; for a related case study see Cser, Oszkó & Várnai 2023a); here we focus on related phenomena involving the spreading of labiality that preceded the appearance of progressive labial harmony as found in Modern Hungarian.

In the course of the later Pre-Old Hungarian and the Old Hungarian periods several sound changes led to an increase in the number of front rounded vowels, of which only [y] had originally been part of the vowel system; the new vowels that emerged were [ø øː yː].¹ Of these, the long vowels very frequently appeared in final syllables, resulting from the contraction of vowel+glide sequences, e.g. *[fɛɣ] > *[fɛɥ] > [føː] ‘head’ or [meneɣ] > (*)[meneɥ] > [menøː] ‘go’ partcpl. What is seen in Middle and Modern Hungarian as progressive root-controlled labial harmony was historically preceded by the regressive spreading of labiality onto short front vowels from front rounded vowels. This regressive spreading was highly variable and subject to morphological levelling (for details as well as recent literature again see Cser, Oszkó & Várnai to appear), and was further complicated by the fact that the short final vowels that originally triggered the spreading were subsequently lost before the appearance of substantial written attestations of the language.

Long final vowels triggering the regressive spreading did remain, but there are too few lexical items with both convincing etymologies and ample attestation to allow us to map the process in greater detail. This issue was treated in Berrár (1961), with a critical survey of the literature up to that point as well as the list of words that could possibly be examples of regressive spreading of labiality from final long front rounded vowels. The putative candidates are (in Modern Hungarian form) ünő ‘heifer’, ürü ‘castrated ram’, tüdő ‘lung(s)’, süllő ‘zander’, süldő ‘young animal’, küllő ‘spoke (of wheel)’, Küküllő (name of a river), hüllő ‘reptile’, gyűrű ‘ring’, gyűszű ‘thimble’, külső ‘outer’, sűrű ‘thick’, minő ‘of what kind’, midőn ‘when’, ízű ‘of … taste’, dicső ‘glorious’, idő ‘time’, hírű ‘famed’, hitű ‘of … faith’, cipő ‘shoe’ (Berrár 1961, 32). These words are attested either in the modern dialects or in earlier written sources with a rounded vowel in the penultimate syllable; for many that variant is the current standard form. However, for the period in question, i.e. Old Hungarian, when the regressive labial spreading may have taken place in these words, only two of them are attested in any appreciable numbers; some of them sporadically, some of them not at all. The two words are idő ‘time’ and dicső ‘glorious’ and their derivatives. The reason why they are so amply attested in Old Hungarian is obviously related to their meanings. Nearly all remaining texts from the period are religious in nature: Bible translations and periphrases, theological and spiritual writings. Besides being an everyday concept, time (both in the sense of absolute, linear time and the time appropriate for something) is also a central notion in the Biblical and Christian understanding of history and man's fate; glory is a central attribute of God, and glorification and praise are central elements of man's relation to God.

Berrár (1961) investigated in detail the forms of idő in many of the surviving Old Hungarian codices. She found that in some of them, primarily those that represent the Transylvanian dialect, a rounded vowel appeared in the first syllable of the stem when the stem-final vowel itself was rounded; and an unrounded vowel appeared when the stem-final vowel itself was unrounded. The rounding of the stem-final vowel resulted from systematic sound changes (see above) that had different outcomes in word-final vs. word-internal position, see (2).

Of the two forms, only the first shows a rounded first vowel, i.e. üdő, in the relevant textual sources.² What this means is that a systematic alternation exists within the paradigm of this lexical item in both vowels, and the two alternations are interrelated. Berrár's (1961, 34) data taken from the Vienna Codex (one showing a highly consistent pattern) include those in (3), with spelling modernized here for the sake of exposition.

Berrár (1961, 37) also reports that Erzsébet E. Abaffy (p.c.) drew her attention to a similar-looking alternation apparently displayed by the derivational family of the absolute stem dics- ‘glory, praise’. Forms with alternating labiality such as dücső [dytʃøː] ‘glorious’ vs. dicsér [ditʃeːr] ‘praise’ could be found both in Middle Hungarian texts and in more recently attested dialects, but the evidence was anecdotal. To our knowledge no systematic study has ever been conducted on the alternation displayed by this lexical family (including dicső ‘glorious’, dicsőség ‘glory’, dicsőít ‘glorify’, dicsőül ‘be glorified’, dicsér ‘praise’, dicsekedik ‘boast’) and no research dedicated to it has been published. We now make an attempt to fill this gap by systematically surveying all the forms in the Old Hungarian codices based on this absolute stem to see whether such a pattern indeed existed and how exactly it worked. The results are interesting because they differ from those obtained for idő in significant ways. By doing this research and publishing its results we pay tribute to the late Erzsébet E. Abaffy (1928–2023), a towering figure of Hungarian historical linguistics for many decades; furthermore, we honour the late László Kálmán (1957–2021), who always insisted on making claims only if they could be verified on a sufficiently strong empirical basis.

In what follows we first describe the morphological structure of the lexical family (section 2), then the way the data were collected (section 3), and the patterns they reveal (section 4). We devote separate discussion to an exceptional form (section 5) before the conclusion (section 6).

2 The morphological structure of the dics- lexical set

The origin of the absolute stem dics- is unknown.³ At the same time, the lexical items based on it all show derivational endings that are well attested and appear to have well-defined functions and morphosyntactic properties. The adjective dicső ‘glorious’ has the ending of present participles (back-harmonizing -ó/-ő), and the words dicsőség ‘glory’, dicsőül ‘be glorified’ and dicsőít ‘glorify’ are clearly based on it. The noun dicsőség has a productive affix that forms nouns from adjectives (back-harmonizing -ság/-ség); the verbs dicsőül and dicsőít end in productive affixes that form intransitive (back-harmonizing -ul/-ül) and transitive verbs (-ít), respectively. The verb dicsér ‘praise’ also ends in a verb-forming affix with high lexical frequency (-Vr showing a variety of harmonizing vowels), whereas dicsekedik ‘boast’ shows a productive reflexive affix-complex (back- and round-harmonizing -kod/-ked/-köd).⁴ The latter two verbs are not based on the adjective but on the absolute stem except for marginal Old Hungarian formations such as dicsőlködik found in the Jókai Codex⁵ where the reflexive affix-complex appears to be attached to the verb dicső(ü)l. The noun dicsőret ‘praise’, found three times in the Weszprémi Codex and once in the Teleki Codex, ends in a noun-forming affix (back-harmonizing -at/-et), but is based on an r-suffixed verb which in turn is apparently based on the adjective dicső rather than the absolute stem. Alternatively it may represent a case of blending influenced by the labial vowel of the adjective and the forms based on it. The more usual form of this noun is dicséret (even in the Weszprémy Codex and in the Teleki Codex), with the -et suffix added to the verb dicsér.

3 Data collection

We collected all word forms beginning with (Old Hungarian spellings of) dics- or dücs- from the electronic versions of the Old Hungarian codices, which are all included in the Old Hungarian Corpus (henceforth OHC, http://omagyarkorpusz.nytud.hu/en-intro.html, see Simon 2014). Since the majority of the codices are not normalized we needed to make searches for all written variants of the absolute stem, listed in (4).

The variants in (4a) clearly show a non-labial vowel, those in (4b) a labial vowel. The last two spellings in (4b) suggest a lowered vowel (i.e. [døtʃ] rather than [dytʃ]), as does decz in (4a) (i.e. [detʃ] rather than [ditʃ]), though this spelling is extremely rare, with three attestations altogether in a single codex in the entire corpus.⁶ The last two spellings in (4a) suggest a palatalized initial stop: g would be a usual spelling for [ɟ] but not for [d]. In the electronic corpus we also needed to do separate searches for capitalized forms as well as for forms separated at line breaks (coded as “di-@@ch” etc.). In the remainder of the paper we shall continue to represent forms in a modernized spelling, since rendering the highly varied Old Hungarian spellings would only add clutter to the exposition, while the focus is solely on the contrast between rounded vs. unrounded front vowels. Old Hungarian written forms, as culled from the OHC, are given where we deemed it necessary to give precise documentation.

Altogether we found 5,716 data for forms based on the absolute stem in question. We checked several forms in the facsimile editions of the codices on an individual basis if we found them suspect for any reason, but we needed to discard only a handful as instances of mistaken data entry. The majority of the data (4,604) occurred in codices in which the labial variant was not found at all, i.e. in which we encountered no forms written with v, w, o or any of their diacriticized forms. 1,112 data were found in codices that contained both labial and non-labial variants, i.e. forms based both on stems written as in (4a) and on stems written as in (4b). The latter set of codices, i.e. of those containing at least one labial form, includes eight codices altogether, viz. the Debrecen Codex, the Érsekújvár Codex, the Gömöry Codex, the Booklet on the Dignity of Apostles, the Lázár Codex, the Székelyudvarhely Codex, the Teleki Codex and the Thewrewk Codex. It is notable that this set does not contain several codices that show either consistent or partial labialization in the forms of idő (the Vienna Codex and the Vitkovics Codex with consistent, the Apor Codex, the Döbrentei Codex and the Munich Codex with partial labialization according to Berrár 1961, 35). The overlap between the two sets consists in the Érsekújvár Codex, the Székelyudvarhely Codex and the Teleki Codex, all the three with partial labialization in idő. This is a point to which we shall return below. In what follows we look in detail at what these 1,112 word forms show; we will not be concerned with the data found in the codices with no labialization at all.

4 The data patterns

In a summary tabulation, the forms found in the eight ‒ at least partially ‒ labializing codices altogether show the distribution given in Table 1. The two rows of the table distinguish between the vowel variants of the absolute stem: in the upper row the stem is of the dücs- type, with a rounded vowel, in the lower row it is of the dics- type, with an unrounded vowel. The two columns distinguish between the type of the vowel of the immediately following syllable independently of the overall length and morphological composition of the whole word form. Thus in the left column belong forms such as dücső (top left cell) or dicső (bottom left cell) ‘glorious’, with a rounded second vowel; in the right column belong forms such as dücseség (top right cell) ‘glory’, dicsér (bottom right cell) ‘praise’, with an unrounded second vowel.

Even at first approximation the numbers in Table 1 show that (i) the stem has an unrounded vowel in the majority of forms even if the second vowel is rounded (69%), but (ii) a rounded stem vowel is nearly always followed by a rounded vowel in the second syllable (90%). We will discuss below the smallest and most unexpected set of data, when dücs- is followed by an unrounded vowel in an inverse harmonic fashion, but not before we look at the codices individually. Tables 2‒9 show the data tabulated from the codices in decreasing order of the ratio of rounded stem vowels (i.e. dücs- vs. dics-) before rounded vowels in the second syllable.

The only text that shows a perfect assimilatory pattern is the brief Booklet on the Dignity of Apostles; here we do not find a single form in which the labiality of the vowels does not match. In the Gömöry Codex the majority of the forms shows the same harmonic pattern, with only 17 representing the dicső-type with an unrounded stem vowel before a rounded vowel in the following syllable, and two forms representing the inverse pattern. In the Székelyudvarhely Codex the forms with a rounded vowel in the second syllable show a rounded or an unrounded stem vowel in equal numbers, but these numbers are fairly low altogether. It is in the Érsekújvár Codex that we have the highest number of relevant forms overall by far. In this text when the vowel in the second syllable is rounded, there is some tendency for the stem vowel to undergo assimilation, but this only results in 40.5% rounded stem vowels; interestingly we also have 15 inverse forms. In both the Teleki Codex and the Debrecen Codex rounded stem vowels are found before rounded second syllable vowels but are a clear minority. In the last two texts, the Lázár Codex and the Thewrewk Codex, there is only one and two forms in each, respectively, with a rounded stem vowel altogether; in the former this is in a harmonizing form, in the latter they are inverse forms.

We thus see a graded pattern of variation in this single lexical family, but the variation, to the extent that it exists at all, points consistently in one direction: a rounded vowel in the second syllable may be preceded either by a rounded stem vowel or an unrounded stem vowel; but rounded stem vowels are almost exclusively found before a rounded vowel. As argued in Berrár (1961) for idő ‘time’, and in Losonczi (1915–1920) for short vowels, patterns like this show quite clearly the effect of the original regressive spreading of labiality among front vowels,⁷ which then came to be replaced by the progressive spreading pattern more narrowly referred to as rounding harmony or labial harmony in the phonological literature (Cser, Oszkó & Várnai 2023b and to appear).

5 The exception

There is one word within the lexical family which shows an inverse pattern: dicsőség ‘glory’. In 19 instances altogether this appears as dücseség, with a rounded stem vowel followed by an unrounded vowel plus the nominalizing suffix -ség. Since the spelling in the Old Hungarian period does not normally distinguish short and long vowels, it is, in theory, possible that the form is dücséség with a long second vowel. There are thus two possible analyses for this form.

If the second vowel was indeed long, it would most likely represent a delabialized form of the participial ending -ő, where the delabialization could possibly be attributed to the long unrounded vowel of the following suffix. In this case, dücséség, or indeed dicséség would be analogous to marginally attested forms such as béség ‘plenty’ from bő ‘wide, plentiful’ or héség ‘heat’ from hő ‘hot’.⁸ The problem is that such a pattern is almost totally unattested for words where there are two or more syllables before the -ség suffix: we hardly ever find a vowel written e where one could plausibly expect a rounded long vowel.⁹ A notable example is found in the Debrecen Codex (292), where within a single sentence one finds the word form gyönyörűségek ‘pleasures’ written as go̗no̗resegek and go̗no̗ro̗segek, with the vowel before -ség apparently varying in labiality. Yet another interesting form is ÿgÿenleseghe (Érsekújvár Codex, 69) for egyenlősége ‘equality poss3sg’, in which the suffix -ség is without doubt preceded by the participle ending that would otherwise be a rounded vowel. It is difficult to make sense of the variation seen in the form written as ku̇ńesege (Guary Codex, 125/13) for what would now be könnyűsége ‘lightness poss3sg’ -- here again the stem shows extensive variability in Old Hungarian and it is not at all clear whether the vowel before -ség etymologically derives from a rounded vowel. The only form that would present a closer parallel to a long-vowelled dücséség, and is attested in substantial numbers, is erősség (‘strength’, from erős ‘strong’), not infrequently written with unrounded vowels throughout.¹⁰

If, however, the vowel before -ség in this word was short rather than long, the form dücseség fell into a well-populated class of forms where the stem (in late Old Hungarian at least) was a monosyllable, and could be a noun, an adjective or even a verb, such as bölcsesség ‘wisdom’, szüzesség ‘virginity’,¹¹ vereség ‘defeat’, nyereség ‘gain’, eleség ‘food’; and formally very similar words could be based on disyllabic stems, e.g. békesség ‘peace’, epeség ‘sadness’. Many such forms were certainly lexicalized already in Old Hungarian, e.g. feleség ‘wife’, etymologically based on fél ‘half’. Note also the word gyűleség ‘hatred’ (e.g. gÿwleseeg), often found in the codices, which varies with gyűlöség (e.g. gÿwlo̗seegh; Érsekújvár Codex, 95vb) and gyűlölség (rarely, e.g. go̗lo̗lseg; Tihany Codex, 123v, with go̗lo̗segh in the very next sentence), but whose morphology is unclear.¹²

The upshot is that we cannot say with certainty if the forms with an unrounded vowel before -ség were dücseség or dücséség. What we can say, however, is that they represent a stage in the variable lexicalization process at least in the language variety that is represented by the Érsekújvár Codex, possibly also by the Gömöry Codex. It is variable because the labiality of the second vowel does not always spread to the stem vowel; but it shows lexicalization because the rounded stem vowel is no longer strictly dependent on a following rounded vowel.

6 Conclusion and remaining issues

We have shown that Erzsébet E. Abaffy's intuition reported in Berrár (1961, 37) was correct: the lexical family built on the absolute stem dics- indeed shows alternations induced by the regressive spreading of labiality from a rounded vowel following the stem. Its behaviour is thus similar to that of the forms built on idő ‘time’, but also significantly different in terms of grammatical status: in idő, the alternation was intra-paradigmatic, i.e. it appeared within inflection (as well as appearing across derivation); in the dics- lexical family, it seems to have appeared only across derivationally related words. Another difference concerns the distribution of the variant forms across the codices: as was indicated earlier in the paper, the variation the two sets of words or word forms show is not coextensive in terms of attestation. The Érsekújvár Codex, the Székelyudvarhely Codex and the Teleki Codex show partial labialization both in idő and the dics- family; the Vienna Codex and the Vitkovics Codex show consistent labialization in idő but none in dics-; the Apor Codex, the Döbrentei Codex and the Munich Codex show partial labialization in idő but none in dics-; the Gömöry Codex shows dominant labialization in dics- but none in idő; the Debrecen Codex shows partial labialization in dics- but none in idő; the Lázár Codex and the Thewrewk Codex show marginal labialization in dics- but again none in idő (see Figure 1). The Booklet on the Dignity of Apostles shows consistent labialization in dics- but since the only forms of idő it contains are two occurrences of idején, it is impossible to tell if the language variety of its author/scribe included üdő or idő. The different distributions may, of course, also result from etymological differences in that idő may have been borrowed with a rounded rather than with an unrounded vowel at least in some dialects of Pre-Old Hungarian (see above).

View Full Size

Labialization in idő vs. dics- in the Old Hungarian codices (note that all other codices would go into the none/none slot, not filled here)

Citation: Acta Linguistica Academica 71, 1-2; 10.1556/2062.2024.00731

• Download Figure
• Download figure as PowerPoint slide

It is also evident that since several of the codices were written by several hands each (with overlaps between different codices), a full investigation of the variety represented by these texts should involve an investigation into the individual hands. It remains for future research to establish how exactly the variety within the codices corresponds to the hands. It also remains for future research to trace the later variation in the words belonging to the dics-family; Middle Hungarian data in which rounded vowels are followed by unrounded ones¹³ point to an interesting interplay between lexicalization with labial vowels and the harmony patterns appearing in affixes, still involving a great deal of variation.

Conflict of Interest

András Cser is a member of the Editorial Board of the journal and was not involved in the review process in any capacity.