![Sample our Language & Literature journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5941/TOC-Subject-Sample-2021-Language-Literature.jpg"})
Notes
1 Certain preliminaries, such as the discussions of reconstruction and subgrouping, could have benefited from more reference to the literature on these topics, such as Hoenigswald or Hale.
2 Kufner, ix. The main earlier attempts in English at filling the gap are Coetsem and Kufner, as well as Voyles, of which the former is self-avowedly heterogeneous and misses coverage of many important topics (Kufner, x–xi), and the latter is primarily concerned with theoretical linguistic approaches to historical phonology. The ongoing Linguistic History of English (Ringe; Ringe and Taylor) is of course an extremely valuable treatment of many classic topics in the field, but it is not structured as a reference grammar, and is in any case biased (by intention) towards Old English, especially in the second volume. Effectively contemporaneous with Fulk’s Comparative Grammar is the excellent new collection of essays on Germanic in Klein, Joseph, and Fritz; these are rather broader in their coverage of topics, including substantial treatments of things like syntax and the lexicon, but each topic is still covered somewhat more briefly. We are fortunate to have three excellent and complementary works appear in such a short span of time, and each is worth consulting alongside the others.
3 While this goal is generally admirably met, there are of course some holes. To take just one possible example that may be of particular interest to scholars of Old English: Fulk endorses without comment the old idea that the variation in weak class II preterites in Old English between forms like egsode “terrified” (with o from older u; the form egsude is in fact also attested) and egsade was originally created by a raising of pre-Old English *ō to *ū before a following *u, which occurred in some endings (as in eg(e)sodon < *egisudun), with a change of *ō > a in other contexts. Under this theory—sometimes called van Helten’s rule—the third-person singular *egisōdǣ should have given egsade by regular sound change, and forms like egsode have the o (< *u) introduced by analogy from the plural. Stausland Johnsen, in an article not cited in this book, has now shown convincingly that this old theory cannot be correct. Rather, it is more likely that the development of *ō > u occurred in all medial syllables, regardless of what vowel followed, so that the development *egsōdǣ > egisude > egsode is perfectly regular. The development of *ō to a occurred, by contrast, when no further syllables followed, which was mainly the case in the past/passive participle: *egisōd “(was) terrified” would regularly become *eg(e)sad. This new theory—which we might call Stausland Johnsen’s rule—empirically accounts significantly better for the distribution of vowels in early West Saxon, where a still predominates in final syllables, and o (u) in medial syllables. A few other minor omissions which might nonetheless have been cited as useful recent contributions: the thesis of Johannsson on the Old High German verbal ending -mē̆s (p. 276); the controversial but important work of Dyvik on “breaking” in Old Norse—arguing that the development of older *bergaⁿ “protect, save” into Norse bjarga is due to the following consonant cluster rather than the following (nasalized!) vowel—which would have nicely rounded out the comparisons to parallel vocalic developments in Old English (cf. beorgan); Kümmel (Ungeklärtes *u) on early anaptyctic vowels in Germanic; Gąsiorowski on the possible sound change *Vzr > *V̄r; and the useful synthesis of Hyllestad on the early interactions between Germanic and Celtic. Naturally preferences about just what should have been cited will vary between individuals.
4 The suggestion that Indo-European *o had not become *a in all positions seems exceptionally implausible. Somewhere between a notational question and a matter of substantive reconstruction is the placement of many final segments that were lost early in brackets. This leads to odd anachronisms, such as *sixʷiþ(i) “she sees”, *saxʷ(e) “she saw”, and *sēʒʷun(þ) “they saw” all given as “Proto-Germanic” (pp. 64, 41, 108). In theory, the bracketed sounds are ones that existed in Proto-Germanic, but disappeared earlier, so that the parentheses are a kind of mnemonic to this loss (pp. 274, 277), but of these three forms, this was only true of Proto-Germanic *sixʷiþi (cf. Gothic saiƕiþ). The proper Proto-Germanic forms of the latter are simply *saxʷ and *sēɣun (using <ɣ>, from the International Phonetic Alphabet, rather than Fulk’s <ʒ>, which to most linguists symbolizes the “zh” sound of words like measure, and supplying the delabialization that occurred before *u), with loss of final *-t and *-e happening (in that order) before the Proto-Germanic stage; cf. §5.2 of the present grammar, and the clear discussion by Stiles, 120–22. Such anachronisms are perhaps sometimes intended as artificial shorthands for the sake of exposition or as mnemonics, but in very many cases the brackets produce needless clutter and some hybrid forms appear to sacrifice accuracy for no useful purpose. More explicit recognition of the early loss of *-t (*-þ), for instance, would have saved Fulk from a certain amount of anxiety about why the vowel of the weak preterite ending “*-ðē(þ)” was shortened in Gothic -da (e.g. p. 292).
5 Cf. Prokosch, 171.
6 Fulk reports the opinion of Ringe, in Ringe and Taylor, 141f., that Proto-Germanic had *kʷemanaⁿ, and the reduced forms are due to phonological developments in Northwest Germanic. Alternatively, Proto-Germanic may have had only *k(w)umanaⁿ, which was replaced by the e-grade forms analogically, this being a very common vowel in the presents of strong verbs of classes 1–5; see Seebold, 315; Rix and Kümmel, s.v. *gu̯em-.
7 Compare Rix and Kümmel, s.v. *u̯ed-, and, especially, Wodtko, Irslinger, and Schneider, s.v. *u̯ed- and *u̯eh₁-r-, n. 1. Fulk also gives a dubious lengthened grade in *rei̯-rēi̯-é-ti “trembles”, which is not in keeping with the standard view that such Indo-European lengthened grades are generally morphologically much more restricted than some earlier research thought. LIV suggests the reconstruction *h₃réi̯-h₃roi̯H instead (see Rix and Kümmel, s.v. *h₃rei̯H- and nn. 9 and 10, with references). This update of *ēi to *oi, far removed as it first appears from Germanic, is actually of some direct relevance to the formation of the class III weak verbs—the class to which the descendent of this Indo-European verb belongs, cf. Gothic in-reiraida “trembled”—which are now thought to contain reflexes of Indo-European *oi in the stem rather than *ē(i̯e).
8 Streitberg, 236f.; Flasdieck, 57ff.; Dahl, 133; Sievers, 206 (§252, Anm. 2); Campbell, 234.
9 The problem is that Proto-Germanic *ôz ought to develop into Old High German -o and Old English -a, but the ending in question is actually found as -a and -e, respectively, as if from bimoric *-ōz. Rewriting Old High German geba as gebā does not really help matters.
10 Rix, 132 (§143); Schaffner, 368.
11 Stiles, 139, n. 17; Bammesberger, 102f. (§4.2.3.3); Schaffner, 368.
12 Though it must be said that Fulk’s suggestion (§6.9 and n. 3), following Marchand, that Kluge’s law operated only after the breakup of Proto-Germanic and the divergence of Gothic is impossible to credit. Like Verner’s, Kluge’s law depends on the position of the Proto-Indo-European mobile accent—a conditioning which Fulk does not doubt—and so, if it took place at all, it did so before the fixing of the accent on the root (a sound change which itself preceded a number of other Proto-Germanic sound changes, and can hardly be regarded as particularly late). A reference back to Kluge’s law in the discussions of class II and IV weak verbs might have also been helpful.
13 Fulk is, perhaps, a little overzealous in fighting his corner on certain details, such as with the rather improbable theory that the medial r of Old High German preterites like ki-scerot “cut” is some sort of hiatus filler, from an earlier *skeōt. Compare the discussion by Ringe, in Ringe and Taylor, 88–92, who is deeply indebted to Fulk’s 1987 article, but who is also more open to the kinds of irregular reductions, e.g. of -scerot from *ske-skrōt (from *ske-skraut), advocated by Jasanoff. It is nonetheless clear that Fulk’s overall approach has been deeply influential, and the discussion in the current grammar is valuable.
14 Ringe and Taylor, 68f.
15 The only really unfortunate thing is that Fulk presents his view as the communis opinio, commenting that the aorist theory is “now usually” the explanation adopted by scholars. This is no longer true, whatever popularity the aorist theory enjoyed among previous generations: from a glance at the references cited over the course of Fulk’s discussion, it is apparent that there have been few, if any, serious defences of the aorist theory in the past four decades—aside from Fulk’s own discussion in Hogg and Fulk, 222, which presents the aorist theory as relatively uncontroversial—in contrast with the array of rejections of this idea in favour of the subjunctive origin that have appeared in more recent years. There are one or two other places where Fulk’s judgements about what is “normally” or “usually” thought do not seem entirely accurate. For instance, commenting on the three “laryngeal” consonants of Proto-Indo-European, he claims that Greek is “more commonly” now regarded as producing a single schwa from the “vocalization” of any laryngeal, which usually developed into *a, paralleling the development in many other Indo-European languages. This is not true, and the “triple reflex of schwa” is the mainstream view in Greek and Indo-European linguistics, and is indeed a crucial component of the trilaryngealistic model that Fulk himself follows for Proto-Indo-European; cf., among others, Rix, 72; Tichy, 41; Rau, 175f.; Fortson IV, 62; Beekes and de Vaan, 148; Bubenik, 641.
16 That is, based especially on Homeric Greek, Vedic Sanskrit, and Old Avestan. The “Aryan” part does not refer to the distorted and racialized sense once far too popular in European culture, but to the Indo-Iranian (better: Indo-Iranic) branch, of which Sanskrit and Avestan are the oldest representatives; cf. Fortson IV, 209f.
17 On which now see the excellent book by Willi.
18 Fulk’s focus is evident in the index verborum, which includes only some dozen Hittite words, and just two from Tocharian (both from Tocharian B). There are a few points where a little more attention to these languages might have been useful: when considering possible o-grades in the verbal root *√dʰeh₁- “set, put” (whence English do), Tocharian B tättā- should probably have been mentioned as possible evidence (see Malzahn, 650), as should Luwian dai (see Morpurgo Davies, 225, n. 47). Also compare Rix and Kümmel, s.v. *dʰeh₁-. Here we do have a possible example of Germanic grouping with these more archaic branches against its more immediate European neighbours—though it might still group with Indo-Iranic, since Vedic dádhāti and Young Avestan daσāiti could go back to either *dʰe-δʰeh₁-ti or *dʰe-dʰoh₁-ti. Nonetheless, Fulk’s arguments against a particularly archaic position for Germanic are generally well grounded and convincing.
19 Fulk (p. 159) does comment on this, but only well into his first chapter on morphology. It is worth noting that nearly every paradigm in the book contains such reconstructions, not merely the few for which explicit caveats are given.
20 Magnús Snædal, 1010.
21 Kiparsky, 353.
22 In this case, the exhorbitant costs typical of an academic book were assumed by various bodies at Indiana University, Bloomington rather than by the reader. The PDF may be found online: https://doi.org/10.1075/sigl.3 The price tag of the printed book remains, typically but unfortunately, high, beyond the reach of most students and independent scholars.
23 There are inevitably a few typos and small errors of fact, few of which are of any real significance. I list them here in hopes that they may be corrected in a future printing or edition: the merger of short a and o was only opposite in Slavic, not Baltic (p. 6); that Wulfila “certainly knew the Runic alphabet” is an overstatement, since the claim that any Gothic letters derive from runes is dubious (p. 22); ςυναγωγῇ should be συναγωγή (p. 20); “ultima” should be “penult”, “penult” should be “antepenult” (p. 35); Avestan saēni- can hardly reflect *k̑h₁-i̯-n-, which would give ˣsin- or the like, but comes from something like *k̑eh₃-in- (Mayrhofer, s.v. ŚĀ) (p. 50); *k̑h̥₁i̯d- should be *k̑eid- or *k̑oid-, cf. Sogdian ’n-s’yδ, Ossetic Iron sidyn, Digoron sedun, etc. (Cheung, Vocalism, §§0.7.2.3, 0.7.2.5; Verb, s.v. *said²) (p. 50); *deh₂i̯u̯ers should be *deh₂iu̯ers, with no *ā ever arising, as suggested by, e.g., Sanskrit devár-, not ˣdāivár- (p. 50); in *laiþjánaⁿ the accent would never have stood there, cf. *loitéjonom (p. 52); Hittite has been omitted from the tables in §3.7; *o did not occur in any environment in Proto-Germanic (p. 55); forms like Segimerus may have more to do with Latin phonology or folk etymology than Germanic phonology (p. 58); PGmc *ē > NWGmc *ǣ > OE ǣ requires more sound changes elsewhere, and so is not clearly “simpler” (p. 61); *baðjaz should be *baðjaⁿ (p. 65); Bergelmir is only evidence for iʀ-umlaut (p. 66); the change of ǫ to á is only graphic, and the resulting vowel was probably /ɔː/ (see Hreinn Benediktsson) (p. 68); ā also appears several centuries earlier than Ellestad on the Thorsberg and Vimose Chapes (p. 70); Norse eru may not be an example of lowering (p. 71, cf. 325); the first two rows of the table in §4.18 are repeated from p. 77 at the top of p. 78; on this table, the long vowel reflexes (from *Vnx sequences) of short vowels are inconsistently given, and umlaut reflexes for *ō are missing, as is ý as a reflex of *eu in Norse; the relevance of Anglian milc for Gothic syncope is opaque (p. 80); *i and *u need not have developed in parallel in Gothic (p. 81); Proto-Germanic *-âi does not fit with the assumptions of Proto-Germanic phonology laid out elsewhere, and it is unclear why anything more complex than *-ōi > Gothic -ai is needed (p. 85); the lack of fronting in the OE genitive plural -a is because it was formerly trimoric, not from nasalization (p. 85); *-ōⁿ was never trimoric, it just remained unshortened in early OE (p. 87); the partial failure of *-st-, *-sk- clusters to heterosyllabify is only potential evidence of syllabification rules, not that they formed a “unitary phoneme” (pp. 91, 94); the lack of syncope in Mercian lȳtelu is not an “exception” (see Goering) (p. 92); *s is missing from the table of PIE consonants (p. 99); “allophones of the voiced aspirates” should be “allophones of the voiceless stops before *h₂” (p. 99); Ossetic is “Iranian” but not “of Iran”: the linguistic term “Iranic” has been suggested precisely to avoid confusions of this sort (Kümmel, Iranic) (p. 100); uncer is a further important possible example of Germanic *k from a laryngeal (p. 101); *dheu̯b- should probably be *dʰeubʰ- (p. 103); the etymology of igqis does not entail a synchronic structure /kw/ (p. 104); vawžaka- should be vaβžaka- (and in general the notation of Avestan forms in the book reflects Bartholomae rather than the more philologically precise transcriptions now standard; see Hoffmann’s important article) (p. 105); an old root accent in auhns is implied, though not guaranteed, by the voiceless fricative as well as the lack of gemination (p. 116); for a recent accentual approach to Holtzmann’s law, see Kroonen, xxxviii–xl (p. 118); the f of þarft is not from devoicing, but from *pt (p. 122); Thöny argues that final devoicing only began late in the Early Runic period (p. 125); final *n is regularly retained in Old Frisian monosyllables such as in (p. 133); “paraadigm” should be “paradigm” (p. 142); the reasons for doubting the simple scenario of PGmc *-ōi > Gothic -ai, PGmc *-ai > Gothic -a (both > West Germanic *-ǣ) are not clear from this discussion (p. 147); it is no longer the “usual formulation” to group all polysyllables with heavy stems under Sievers’ law (p. 150); in the paradigm in §7.11, the Old Icelandic accusative and genitive singulars have been swapped; holtijaz is more likely a nominative adjective than a genitive noun (cf. final z, not ˣs) (p. 152); it is not clear that mawi goes back to a heavy stem on any relevant scale (a two-phoneme *ɣw in PGmc is at least very speculative) (p. 156); Sanskrit vr̥kyàm rather reflects *u̯l̥kʷíh₂m̥ or the like (p. 157); the genitive singular of jō-stems is potentially distinct in original length from that of the plain ō-stems (Schaffner, 368) (p. 157); PIE *-on-m̥ is robustly attested alongside *-en-m̥ in the older branches (p. 170); in the first paradigm in §8.2, the tonic forms *n̥h₁u̯e (*n̥h₃u̯e?), *mebhi, and *n̥smei̯ are missing their accent marks (p. 181); vit has a boldface v- (p. 182); raised *ik could be regular in unstressed position (p. 182); PGmc *wīz could itself be from “*u̯es, with raising to *wiz when unstressed, and lengthening when restressed” (p. 184); *unʀar-cannot regularly become ór- (would give ˣuðr-), which must either be from *unsar- or involve some other analogy (p. 189); eis could reflect *ei̯es (Sihler, 391f.) (p. 191); Erbebnis should be Ergebnis (p. 193); *þanō would give OE ˣþonu (p. 194); Early Runic hino (and the precise equation of hi-no = Gothic hi-na = OE hi-ne, all < PGmc *hi-nōⁿ) suggests a greater age for the inflected *hi- demonstrative than implied, and creates problems for some of the suggestions here (pp. 198–99); long *-ī < *ij, not short *-i < *j (p. 214); the section header is repeated at end of §9.12, obscuring part of note 1; PGmc *seƀun, not *siƀun (p. 226); “anvetovalic long vowel” should be “antevocalic short vowel” (p. 227); voicing in Gothic ainlibim is not informative about Germanic stress, only the IE accent (p. 228); completion is a property of verbal aspect (p. 242); the injunctive is also a distinct formation in Avestan (p. 243); the contrast between λϵίπω and *√leikʷ- is unclear (p. 248); the change of *-h₂e to simple *-a is post PIE, as the laryngeal is required in IIr. to block Brugmann’s law (p. 250); *-oi̯h₁- should be *-oi̯h̥₁- (p. 252); causitivity is not part of aspect (p. 254); Schumacher’s “bigetun rule”, rather than the aorist theory, is perhaps now the “commonest explanation” (p. 257); Verner’s law only tells us about the IE accent, not the position of Germanic stress (p. 261); if *xaƀjan- is from *√kh₂p-—as suggested by, e.g., Rix and Kümmel, s.v. *keh₂p-—no full-grade *a is required (p. 263); for skaidan, now compare Rix and Kümmel, s.v. *sk^ʰei̯d- (p. 265); remove second mark for note 4 (p. 265); sēgon is surely the phonological development and sāwon analogical (p. 271); *-is and *-iz should be *-is(i) and *-iz(i) (better, *-isi and *-izi) (p. 274); *-þ(i) should be *-iþ(i) (better, *-iþi) (p. 275); iʀ-umlaut is a complicating factor in the discussion of verbal umlaut in §12.24 note 1; “fricative consonant” should be “obstruent consonant” (p. 277); Runic -da should be -de (p. 294, correct on p. 303); “*-þ- (> *-t-)” should be “*-t-” (GL never affected second obstruents in clusters, and cf. wissa) (p. 300, also 321f.); *salƀōðēþ should be *salƀōðē(þ) (better, *salƀōðē) (p. 304); “ƕarbōīīdēdun” should be “ƕarbōdēdun” (p. 305); since the longer Ingvaeonic forms must have been formed centuries before any surviving Old Saxon texts, they would be real archaisms by the time of Old Saxon verse (p. 308); Pinault’s rule poses a problem for the Bennett model of class III weak verbs advocated for here (p. 311); ai should be ē in the Ingvaeonic paradigm of *xaƀ- (p. 312); preterites like OHG hapta are plausibly archaisms, from PGmc *xaƀða- < *kəpətó-, itself < *kh₂p-h₁-tó- or the like (p. 313); Gothic magt is just the proper spelling of /maxt/, ˣmaht being only for /maht/ (p. 317); Gothic aih is strong evidence, since h does not synchronically represent the voiceless counterpart of g (p. 321); “sj. (opt.)” should be “sj.”: the true PIE subjunctive is meant (p. 322); roots should not be cited in the zero-grade (p. 323); “-mi” should be “*h₁és-mi” (p. 323); *izunþ should be *izun(þ) (better, *izun or *ezun) (p. 325); for earlier suggestions of *ui̯ > *ii̯, see Rix and Kümmel, s.v. *bʰu̯eh₂-, with references (p. 327); if ēode < *ē-udǣ, with regular *u < medial *ō (Stausland Johnsen), then many of the objections to seeing the endings as class II do not hold (p. 336).