Umlaut: From Common Germanic to Dutch

Abstract

This article deals with the Germanic umlaut processes in Common Germanic and Old West-Germanic. Section 1 presents a definition of umlaut and Section 2 gives an overview of different kinds of umlaut processes, including primary i-umlaut and secondary i-umlaut. Their phonological effects are discussed in Sections 3 and 4 attempts to offer an explanation for the various umlaut effects, e.g. why Common Germanic Umlaut precedes the Old West (and East) Germanic Umlaut and why the first Umlaut was limited to short vowels while the second one also applied to long vowels and diphthongs.

This article1 deals with the Germanic umlaut processes in Common Germanic2 (CG)3 and Old West-Germanic (OWG). Section 1 presents a definition of umlaut and Section 2 gives an overview of different kinds of umlaut processes. For OWG umlaut, a distinction must be made between primary i-umlaut and later secondary i-umlaut. Their phonological effects are discussed in Sections 3 and 4 attempts to offer an explanation for the various umlaut effects. An important question is why the Common Germanic Umlaut precedes the Old West (and East) Germanic Umlaut and not vice versa. And also why the first Umlaut was limited to the short vowels while the second one also applied to long vowels and diphthongs.

1 Definition

The German term 'Umlaut', proposed by Jacob Grimm and sometimes used interchangeably with 'metaphony', refers to a change in a stem vowel usually due to a vowel (e.g. i, a) but also a j (which originally was often an allophone of i) in the next syllable. The change causes assimilation between the stem vowel and the other vowel or j, which means they acquire one or more distinctive features in common, e.g. [±high] or [±back]. This is what we refer to as diachronic umlaut; see e.g. Fulk (2018) and references cited there for Old Germanic in general and van Loey (1970), van Loon (2014) or van Bree (2016) for Dutch. Synchronically, umlaut can cause a vocalic alternation within a paradigm or between morphologically related forms. Examples are German tragen 'to carry'- er trägt 'he carries', Du. recht 'right' - richten 'to direct', wand 'wall' - wenden 'to turn'. Another example is Du. sprookje [o:] 'fairytale'- spreuk [ø:] 'proverb', although this might be due to so-called "spontaneous palatalization" (see the next section). Alternations in a paradigm, such as in the German example, are referred to as synchronic umlaut. In the form trägt, the vowel indicated by <ä> in part indicates the 3rd person singular indicative. In the pair Mutter-Mütter 'mother-SG vs. PL', umlaut alone expresses the morphological category of plural. Finally, umlaut can also refer to the reflection in the spelling of synchronic umlaut, like the diacritic in the vowel letters <ä>, <ü>.

In the next section we will present an overview of the umlaut processes that were active in the Germanic languages (see e.g. van Coetsem 1997). Because Old Dutch has not been well preserved we will mainly use examples from Old Saxon, which was a relatively closely related language. These umlaut processes were crucially influenced by consonants: certain consonants or consonant clusters could block or trigger the process. For instance, a-umlaut from PGm. u to CG. o was blocked before a nasal plus consonant, cf. OSa. gibundan - giworpan (the past participle of the verbs bindan 'to bind' and of werpan 'to throw', respectively), but the same combination could also act as a trigger, since it could raise PGm. e to i: cf. OSa. bindan to werpan. I-umlaut of a to e is blocked, by, among others, a geminate [?:] (from PGm. ??j): Du. lachen 'to laugh'; cf. Got. hlahjan; <h> = ?).

2 Overview

In this section we present an overview of the umlaut processes that took place in Germanic, specifically two such processes in Common Germanic and two Old West-Germanic ones (which are also assumed for Old North-Germanic). The first two involve isolated dialect differences, but we assume, following van Coetsem (see also Kylstra 1983, contra I. Fausto Cercignani), that these processes started at the beginning of the Common Germanic period, in which a strong accent had moved to the initial syllable. These first two umlaut processes can be referred to as rl-umlaut, since they consist of raising or lowering, and the latter two processes as fb-umlaut, since they involve fronting or backing. There is an implicational relation: a language with rl-umlaut also has fb-umlaut, but the reverse is not true, so rl-umlaut is more usual (see Buccini 1988, 1995).4 For fb-umlaut we can disregard backing, because it is almost completely irrelevant for Dutch. Let us discuss the umlaut processes in turn:

1. Common Germanic i-umlaut of e to i before i, j (in isolated cases also before u) in the next syllable: OSa. birid 'he bears' - beran 'to bear' (PGm. ?beran-; Du. ont-beert - ont-beren 'to lack', with OSL of i and e to e). This umlaut also plays a role in the development of eu to iu: OSa. kiusid 'he chooses' - kiosan 'to choose' (for kiosan, see 2 below; PGm. ?keusan-; Du. kiezen - hij kiest 'to choose-he chooses').5 Nasal and consonant (or geminate nasal) could also act as a trigger for the raising of e to i: cf. OSa. bindan to werpan.6

2. Common Germanic a-umlaut, caused by a non-high vowel in the following syllable (especially a) from i to e and from u to o: cf. OSa. wer 'man' (PGm. ?wiraz; Du. weer-wolf 'man-wolf' with lengthening before r), OSa. dohtar 'daughter' (PGm. ?duhtar; Du. dochter). A nasal-consonant cluster (or geminate nasal) or j prevent a-umlaut: OE. nistian 'to nest' (PGm. ?nistjan; Du. nestelen, by analogy to nest, PGm. ?nista-),7 Osa. gibundan (PGm. ?bundan-; Du. gebonden 'bound' with a later change from u to o), OSa. fullian 'to fill' (Pgm. ?fuljan; Du. vullen with a later change to [ø]). OSa. bindan 'to bind' (PGm. ?bendan; Du. binden) and OSa. fillian 'to skin' (j > i) (PGm. ?felljan, cf. OSa. fel 'skin'; Du. villen 'to skin' - vel 'skin') show that e becomes i in spite of the presence of a in the following syllable. Note that i/j (as well as nasal plus consonant) is the stronger factor, compared to a. A-umlaut also clearly plays a role in the change of eu to eo > io: OSa. kiosan, but cf. stiurja 'pay', in which the influence of j prevails (eu > iu). Old Saxon and Old High German have goma 'man', as expected, but also guma, in which the m may have prevented umlaut. Note that Proto-Germanic only had a short back vowel u (PIE o developed to PGm. a), so the CG. o created by a-umlaut originally was an allophone of /u/. There was a danger that Common Germanic umlaut would merge the PGm. phonemes /i/ and /e/ into one phoneme with two allophones, but because (among other things) there were a considerable number of exceptions to a-umlaut of i, this did not happen.8

3. Old West-Germanic primary i-umlaut (by i or j in the next syllable) from a to e: OSa. wendian 'to turn' - -wand (PGm. ?wandjan- - ?wand-; Du. wenden - wand). This process is blocked by a velar geminate; the velar articulation seems to prevent the process from applying (cf. lachen, OSa. hlahhian, in the previous section). Umlaut does not apply when the conditioning factor occurs in the third syllable, which explains the difference between OSa. edili 'noble' (factor in the second syllable; Du. edel) and adali 'nobility' (no factor in the second syllable; Du. adel) (see van Loon 2014: 137). Finally, umlaut does not apply when the conditioning factor is part of the second member of a compound or of a relatively independent suffix: OSa. (eli)landig 'in another country = foreign' (cf. Du. (el)lendig 'miserable', with e probably as a result of...