Gothic Miscellany: Vowels, Grimm's Law, Syntax, and Devoicing

1. The manuscript spellings <au> and <ai> appear to represent three distinct vowel pronunciations: a short vowel, a long vowel, and a diphthong. The short vowels [o], [e] occur before r, h, and hw, as well as in the reduplicating syllable of Class VII strong verbs; in modern transcriptions these are represented by and , such as baíran, daúhtar, and laílot. The long vowels [o:], [e:] and the diphthongs [aw], [aj] are the source of some dispute (see below); the long monophthong is represented now by no diacritics (au, ai) while the diphthong is shown as áu and ái; thus we see bauan, saian and ráuths, háitan.

Scholars have spent considerable effort in an attempt to explain these differences. An obvious point of comparison is to trace the development of these vowels in related Germanic languages in order to prove the likely Gothic pronunciation. Thus we deduce that the vowel in baíran is short by comparing it with OHG beran; similarly Gothic dáuhtar can be compared with the short vowel in the OHG tohter. The long vowels and diphthongs show a different vowel in related languages: this while Gothic háitan is cognate to ON heita, the monophthong saian is reflected in ON as ; so too we can compare Gothic ráuths, bauan to ON raudr, búa. We can also examine foreign words (most common are Biblical names from Greek) to see how they are represented in Gothic: thus Greek [epsilon] (short [e]) in the name Elizabeth shows up in Gothic as : Aíleisabaíp, and Greek [alpha][iota] (long [e:]) occurs as Gothic ai in Haíbraius. (Examples from Robinson, pp. 63-66).

The fact that Wulfila, who is otherwise extremely exact in his orthographic innovations, did not provide separate symbols for these distinct sounds poses a question for many scholars, and leads some to postulate that there is really only a two-way distinction between the long and the short vowels (long occur before other vowels, while short occur elsewhere). Others note the different Latin spellings for the Gothic tribe: in 300 A.D. Latin scribes wrote Austrogothi, while by 400 A.D. it was Ostrogothi; this, and other linguistic evidence, supports the postulation that at some point before Wulflia (the so-called "pre-Gothic" oeriod) the sounds were disctinct, but had merged by the time of the texts we have preserved.

2. Karl Verner noted a seeming "problem" with the operation of Grimm's Law in the development of the Germanic languages, and postulated a solution , in reality merely a supplement to Grimm's Law that provides an important exception to the sound changes. According to Grimm, the Indo-European consonants [p, t, k, s] developed into Germanic [f, th, x, s] in word-initial, medial, or final position. Verner noted, however, that in many cases these consonants did not remain voiceless, but instead developed into their voiced counterparts [b, d, g, z]. Through deductive reasoning, Verner showed that the Indo-European placement of stress was the deciding factor in this development: if the immediately preceding syllable did not bear primary stress, then the alternate forms (the voiced versions) were produced. (The reasons behind this are complicated; Bennett summarizes it as explaining that stress requires greater articulation and thus produced a fortis (voiceless) consonant, while no stress produced the lenis (voiced) variant.) Voiceless consonant clusters with an initial 's' retained their voiceless quality, so [sp], [st] and [sk] are not affected. Thus, we see [p] -> [f] in IE klépo, Go. hlifa "I steal", but [p] -> [b] in Sk. kapálam, Go. háubith "head"; [t] -> [th] in IE wérto, Go. waírtha "I become", but [t] -> [d] in IE wentós, Go. winds "wind"; [k] -> [x] in Gk. déka, Go. taíhun "ten", but [k] -> [g] in IE plkenós, Go. fulgins "hidden"; and [s] -> [s] in Sk. jósati, Go. kiusith "chooses", but [s] -> [z] in Sk. bhárase, Go. baíraza "you are born". (Examples from Bennett, pp. 55 and 60). Verner thus upheld the regularity of sound change, by accounting for the seeming exceptions to Grimm's Law.

Gothic is unusual among the Germanic languages in regards to Verner's Law, because it lacks many of the clearest examples of this change. Gothic verbal paradigms, for instance, often do not show the morphophonemic alteration caused by Verner's Law, while nearly all their cognates in the other Germanic languages do. Thus, we have Go. waírthan - warth - waúrthum - waúrthans, but OE wearth - wurdon - worden. There are many possible reasons for this peculiarity: Braune/Ebbinghaus merely note that Gothic is not as conservative as other languages, while Bennett posits possible nonnative speaker influence, levelling by analogy (Roberge and other dismiss this, however), and, most likely, an "earlier fixation of primary stress," meaning that the Goths probably lived close to where the Proto-Germanic fixation of stress began, and thus did not fully reflect the changes.

3. Relying as we must on the few extant Gothic manuscripts we have, it is often hard to determine much about the 'normal' syntax and structure of Gothic. The primary manuscript, Wulfila's Bible, is a translation from the Greek Bible into Gothic; as such, we can assume that Wulfila drew heavily on the Greek constructs in order to express the Biblical concepts, which may well have been foreign to the native Gothic population. Complicating the matter, as Roberge notes, is the fact that the translation even shows some Latin-based influence, so there may be several factors at work. And the other meager manuscripts are no more reliable: documents, records of transaction and other legal writings are hardly fair representations of the daily language of a community, and in addition, many of these were accompanied by their Latin counterparts, and may have been simple translations. Thus we can only draw limited conclusions about Gothic syntax, and even semantics and grammar-based structure may be suspect.

We do see a certain amount of variation, even in the texts we have examined thus far. Word order, for instance, seems to be quite flexible. Significantly, verb placement is variable: sometimes the verb occurs before the subject, sometimes after, sometimes at the beginning of the clause, in the middle, or even at the end. Thus we see ith sa asneus afthliuhith but also jah usgeisnodedun fadrein izos, as well as many other variants. We also see variation in the placement of adjectives and articles. The differentiation between strong and weak adjectives appears consistent, with statements such as ik im haírdeis gods. haírdeis sa goda ... occuring in frequent alternation. Possessive pronouns seem to always appear after the noun they modify, and further modifiers appear to follow, thus we see so daúhtar izos and sunus is sa althiza.

Grammar constructions may have been corrupted by the influence of translation, but it is difficult to say. Notably, optative and imperative forms seem to be almost interchangeable, with only a slight semantic difference to distinguish them: we see the command gibith figgragulth contrasted to the commandment ni maúrthrjáis. There are also many genitive constructions in Gothic that do not occur in later Germanic languages; to a certain extent these may have been influenced by the prevalence of the genitive in Greek, but they may also represent a native Gothic construct.

Finally, semantics can also pose a problem. In even the most basic of texts, the Lord's Prayer, we see both the singular and the plural form of 'heaven': atta unsar thu in himinam (plural), but also swe in himina jah ana aírthái (singular). Certainly the translator was aware of this difference, and may have taken his cue from the Greek original, but the opposition is certainly curious. The same holds true for the distinction between the singular thiuda meaning nation or people, and the plural form thiudos meaning heathens or Gentiles; we see niu jah thái thiudo thata samo táujand, in which case the "Gentiles" are clearly indicated, while in the later speech of Pilate, so thiuda theina probably refers merely to the tribe or nation of people. No matter how much credence we can lend to the quality of Wulfila's translation, in no way can we assume that the phrasings seen in our texts are representative of the actual speech of the Goths.

4. The devoicing of fricatives in final position is a sticking point for many German scholars; in particular the historical time at which this devoicing occured is difficult to pinpoint. Until recently, most scholars had assumed the change to have taken place in the "Pre-Gothic" period. This means that Gothic texts should show a consistent final devoicing. Needless to say, this is not the case; as Roberge notes, there are numerous instances of final voiced consonants (-b, -d, -g, -z) in the extant manuscripts. Roberge therefore posits that the devoicing in fact occured sometime between the time of Wulfila (ca. 311-383 A.D.) and the later scribes who produced the Codex Argenteus from earlier archetypes (thus closer to 500 A.D.).

Roberge reviews earlier solutions to the problem of the extant final voiced consonants: some scholars attributed them to a late dialectal change, in which voicing was retained under certain conditions; others postulated that two different sounds were in fact represented by these orthographical conventions; and still others held to the idea of a regular alternation depending on the quality of the following initial. Thus most scholars saw these voiced fricatives as a scribal deviation from the original archetype. Roberge, however, shows that this cannot be the case, since we see forms that follow none of the above rules; he thus concludes that these forms must be "remnants of a tradition long since abandoned," meaning the pronunciation at the time of Wulfila. Roberge looks at the patterns of final fricatives and concludes that these voiced fricatives (-b, -d, -g, -z) were in fact present in the original archetype, and were copied by the later scribes. Since, however, final devoicing had begun to take effect by the time of these later scribes, there are inconsistencies: in some cases the scribes tried to reconcile Wulfila's spellings to their own pronunciation, but they did so incompletely, thus leaving us with a rather confusing mixture of voiced and voiceless final fricatives.

Written and © Nancy Thuleen in 1997 for German 755 at the University of Wisconsin-Madison.

If needed, cite using something like the following:
Thuleen, Nancy. "Gothic Miscellany: Vowels, Grimm's Law, Syntax, and Devoicing." Website Article. 26 October 1997. <>.