The Third Grammatical Treatise (TGT) is a vernacular treatise on language and rhetoric, based in part on Latin models. It is transmitted in mss A, B and W (see Section 4.1.4 above), as well as in AM 757 b 4° (c. 1500), a copy of W that contains no poetry. The treatise is named for its position among four treatises on broadly grammatical topics in the Codex Wormianus (AM 242 fol, W). According to ms. A(8v), TGT was written by Óláfr hvítaskáld ‘White Skald’ Þórðarson (d. 1259), a nephew of Snorri Sturluson (see Óláfr’s Biography in SkP II, 656). It was most likely composed after 1242, when Óláfr returned to Iceland from mainland Scandinavia. Björn Magnússon Ólsen suggests a narrower range of c. 1245-52 for its composition (TGT 1884, xxxv-xxxvii), but this is based on speculation about poorly documented events in Óláfr’s life (see Wills 2001, 7).
The treatise is in two parts. The first, Málfræðinnar grundvǫllr ‘The Foundation of Grammar’, as it has been known since Rask’s edition (SnE 1818, 297), derives ultimately from Priscian’s Institutiones Grammaticae Books I-II (Keil 1855-80, I, 7-242), with additional material primarily on the runic alphabet. It contains two fragments of skaldic poetry, both known from other sources: SnSt Ht 86/5-6 (illustrating end-rhyme (runhenda) in the section on the syllable) and Þorm Lv 22/4I (illustrating conjunction in the section on parts of speech).
The second part, termed Málskrúðsfræði ‘The Lore of Rhetoric’ by Sveinbjörn Egilsson (SnE 1848, 181), contains almost all the poetry in the treatise, some 121 fragments. The main source of Málskrúðsfræði is Donatus’s Ars maior Book III (Holtz 1981, 653-74), illustrating errors (vitia) of language (barbarisms and solecisms). Óláfr, like other medieval grammarians and rhetoricians, identified deviations from normal usage as either vices or virtues, and considered deviations in poetic language as justifiable in terms of literary effects or poetic licence. The Latin models for this part of the treatise use classical Latin verse citations to illustrate the figures, whereas Óláfr uses Norse poetic citations. Between Donatus and late medieval grammarians lies an extensive medieval commentary tradition, to which Óláfr is clearly indebted, although the exact influence is unclear. Despite direct sources having been identified for similar works such as FoGT (see Introduction to FoGT in this volume) and Ælfric’s grammar (Förster 1917), no such sources for TGT have yet been discovered. Micillo (1999) suggests that extant Hiberno-Latin commentaries on Donatus are likely to be sources for the second section of TGT. These include Murethac, In Donati Artem Maiorem (CCCM 40), the anonymous Ars Laureshamensis (CCCM 40A) and, more convincingly, Sedulius Scottus, In Donati Artem Maiorem (CCCM 40B). Three stanzas here attributed to Óláfr (Ólhv Frag 6-8) appear to be inspired by supplementary examples found in Sedulius but not in Donatus. The last part of the treatise shows clear influence from the Doctrinale of Alexander de Villa-Dei (see Introduction to FoGT below) and other minor sections show influence from the Doctrinale in the structuring of information about rhetorical figures (Wellendorf in FoGT 2014, xli-xliv; Wellendorf [forthcoming]). Additionally, a number of parallels in glosses to the Latin grammatical tradition have been identified, most extensively in the notes to Björn Magnússon Ólsen’s (TGT 1884) edition.
The stanzas edited in this and other volumes of SkP are the following, in the order they appear in the two parts of the treatise.
Málfræðinnar grundvǫllr: (ch. 5) SnSt Ht 83/5-6, (ch. 9) Þorm Lv 22/4I
Barbarisms (ch. 11): Egill Arkv 15V (Eg 111), Hár Lv 1/1-2I, Auðunn Lv 1, Glúmr Gráf 12/5-8I, Arn Frag 7, Anon (TGT) 1, Eil Þdr 4, Anon (TGT) 2, ESk Lv 10, SkrautO Frag 1, StarkSt Frag 1, Anon (TGT) 3-4, Bjhít Lv 15/5-6V (BjH 20), Eskál Vell 30/1-2I, Ólsv Love 1, Anon (TGT) 5, Eyv Lv 8/1-2I, ESk Lv 11, Anon (TGT) 6-7
Solecisms (ch. 12): Anon (TGT) 8, Þjsk Hák 2I, SnSt Frag 1, Anon (TGT) 9, Anon Hafl 1, Arn Hryn 3/3-4II, Anon (TGT) 10, ÞjóðA Lv 1II, Hskv Útdr 12II
Other faults (ch. 13): Ólhv Frag 1, Sigv Nesv 1/1I, SnSt Ht 28/3-4, Anon (TGT) 11-12, Kolb Lv 7, Ólhv Frag 2, Arn Hryn 2II, Guðbr Frag 2, SnH Frag 1, Anon (TGT) 13-14, ÞjóðA Frag 4II, Guðl Lv 1, Ólhv Frag 3, Anon (TGT) 15, Ólsv Love 3, HSt Frag 2
Metaplasm (ch. 14): Egill Frag 1, Anon (TGT) 16, Bragi Frag 4, KormǪ Lv 65, Arn Hryn 1II, Anon (TGT) 10/1, 2/2, Eil Þdr 4/4, HólmgB Lv 8/7-8V (Korm 43), Sigv Lv 23/6I, Anon Bjúgvís 1, Sigv ErfÓl 26I, Anon (TGT) 1/2
Lexical figures (ch. 15): Ólhv Frag 4, Anon (TGT) 17-18, Ólhv Frag 5, Glúmr Eir 1, Anon (TGT) 19, SnSt Ht 15/7-8, 16/1, SnSt Ht 40/1-4, Anon (TGT) 20-5, Ólhv Frag 6, Anon (TGT) 26, SnSt Ht 73, Hfr Lv 11/1-2V (Hallfr 14), Egill Arkv 24VIII (Eg 120), Grí 47/1-2, Þul Sea-kings 1, Máni Lv 5, Anon (TGT) 27
Tropes and metaphors (ch. 16): Anon (TGT) 28, Eyv Hál 9/5-8I, SkrautO Frag 2, Mark Lv 1/5-6, Anon (TGT) 29, Ormr Woman 4/1-2, Þjóðólfr Frag 1, Mark Frag 2, ÞjóðA Sex 28II, Anon (TGT) 30, Ólhv Frag 7, Anon (TGT) 31-2, SnSt Ht 5/3-6, Anon (TGT) 33, ÞKolb Eirdr 7/5-8I, Anon (TGT) 34, Leiðólfr Frag 1, Ólhv Frag 8, Anon (TGT) 35, Egill Arkv 25V (Eg 121), Anon (SnE) 13, Anon (Nj) 3/8V (Nj 64), Anon (TGT) 36-7, Sigv Berv 12/1-4II, ESk Geisl 1/1-4VII, Sveinn Frag 1, Anon Kúgdr 1, Sveinn Norðrdr 3, Gestumbl Heiðr 25/1-3V (Heiðr 72), Egill Lv 46V (Eg 130), Anon (TGT) 38, Egill Arkv 24V (Eg 120), SnSt Lv 5, Ník Kristdr 1 and Ólhv Frag 9.
Around two-thirds of the stanzas in the treatise are not cited elsewhere, and of those the majority are not attributed to a named poet by Óláfr (Gísli Sigurðsson 2000). Wills (2006) has an extensive discussion of the provenance and classification of these stanzas. To a greater extent than in other prosimetrical works whose poetry appears in this volume, the citations in TGT are fragmentary, often reduced to single lines and couplets. This practice is no doubt influenced by the Latin grammatical literature it draws upon, which typically uses single-line poetic citations. The citations in TGT are eclectic in subject and provenance; they include early skaldic poetry, þulur, encomiastic material and poetry also found in Íslendingasögur and fornaldarsögur.
The present edition substantially reorganises Skj’s presentation of the poetry in TGT. It includes two stanzas which Skj inexplicably omits, namely Anon (TGT) 4 and StarkSt Frag 1. Anon (TGT) 12 was attributed to ESk Øxfl in the Arnamagnæan edition of SnE (SnE 1848-87, III, 365) and this attribution, despite lack of evidence, has been accepted by subsequent editors. It is included here as anonymous. Stanza 49 in TGT is attributed to an ‘Óláfr’ in TGT and this Óláfr is identified by Björn Magnússon Ólsen (TGT 1884, 191; adopted by subsequent editors) as Óláfr hvítaskáld Þórðarson, the treatise’s author. However the poet in question is more plausibly Óláfr svartaskáld ‘Black Skald’ Leggsson (Wills 2006, 1057; see also Introduction to Ólsv Love), and the stanza is edited in this volume as Ólsv Love 3. In the present edition nine anonymous stanzas have been attributed to Óláfr hvítaskáld Þórðarson (Ólhv Frag 1-9). The rationale behind this reclassification is that Óláfr must have been responsible for the adaptation of the Latin source material, and there are close parallels to these nine TGT stanzas in the Latin sources (see Notes to Ólhv Frag 1-9). It is likely, therefore, that Óláfr composed these stanzas himself, modelling them on the examples in his Latin source(s). It could well be that other unattributed stanzas in TGT are the work of Óláfr, but this cannot be ascertained.
Mss A, B and W are independent witnesses to the text. All later mss are copies of the three independent medieval redactions and are therefore not used in this edition. Ms. B is damaged and illegible in places; all previous editions have used the unreliable transcriptions of B in SnE 1848-87, II. The present edition supplies readings from ms. 744ˣ (an early eighteenth-century copy made of B when B was less damaged), where B is not clearly legible. Although the independent witnesses (mss A, B, W) also contain versions of Skm, the stemmata for the two works are different. It is clear that A is the best ms. for TGT, as established by Björn Magnússon Ólsen (TGT 1884, lxii). The relationship between B and W is less obvious (see the detailed discussions in TGT 1884, lv-lxiii and Wills 2001, 52-6).
Editions of TGT include the following: Rask (SnE 1818, with W as the base text); Sveinbjörn Egilsson (SnE 1848, based on A); the Arnamagnæan SnE edition (SnE 1848-87, II; based on W with a Latin translation, but also including transcriptions of A and B); Björn Magnússon Ólsen (TGT 1884; a diplomatic text based on A plus a transcription of W); Finnur Jónsson (TGT 1927; a normalised text based on Björn Magnússon Ólsen’s edition); and Krömmelbein (TGT 1998; a reproduction of the text in TGT 1884 with a German translation and notes). TGT 1998 has not been used in the present edition. SnE 1848-87, TGT 1884 and TGT 1927 provide separate translations of and commentary on the stanzas in the treatise. Lucy Collings’ unpublished MA thesis (1967) on TGT contains an English translation of the prose in the second part of the treatise.