the disappearance of vowel syncope in disyllabic-stemmed ... · 2. vowel syncope in...
TRANSCRIPT
Háskóli Íslands
Hugvísindasvið
Íslensk miðaldafræði
The Disappearance of Vowel Syncope in
Disyllabic-Stemmed -igr and -ugr Adjectives
A Diachronic Examination of Adjectives in Old Icelandic
Ritgerð til MA-prófs í íslenskum miðaldafræðum
Chad S.D. Laidlaw
Kt.: 061084-4449
Leiðbeinandi: Haraldur Bernharðsson
Febrúar 2017
Laidlaw 1
Table of Contents
Acknowledgements ................................................................................................................... 3
Abstract ..................................................................................................................................... 4
Ágrip ......................................................................................................................................... 5
1. Introduction & Research Goals ............................................................................................ 6
1.1 Introduction .............................................................................................................. 6
1.2 Research Goals ............................................................................................................ 6
2. Vowel Syncope in Disyllabic-stemmed Adjectives ............................................................... 7
2.1 Overview of Vowel Syncope ........................................................................................ 7
2.2 The case of heilagr ..................................................................................................... 10
3. Methodology ...................................................................................................................... 11
4. The Research Corpora ........................................................................................................ 14
4.1 The Primary Corpus .................................................................................................. 14
AM 674 a 4° – c1150-1200 – The Icelandic Elucidarius .............................. 15
AM 673 a II 4° – c1175-1225 – Physiologus ................................................. 15
AM 673 b 4° – c1175-1225 – Placítus drápa ................................................ 16
Holm perg 15 4° – c1200 – The Icelandic Homily Book .............................. 16
AM 645 4° – c1220-1249 ............................................................................. 17
GKS 1157 fol – c1250 – Konungsbók Grágásar ............................................. 18
AM 291 4° – c1275-1300 – Jómsvíkinga saga ............................................... 18
GKS 1009 fol. – c1275 - Morkinskinna ........................................................ 19
AM 519 a 4° – c1280 – Alexanders saga ....................................................... 19
DG 11 – c1300-1325 – Codex Upsaliensis, Snorra Edda ............................... 20
Holm perg 7 4° – c1300-1325 ...................................................................... 20
AM 132 fol. – c1339-1370 – Mǫðruvallabók ................................................ 21
AM 351 fol. – c1350-1400 – Skáholtsbók eldri ............................................. 22
GKS 1005 fol. – c1384-1397 – Flateyjarbók ................................................. 23
GKS 2845 4° – c1425-1475 .......................................................................... 26
AM 343 a 4° – c1450-1475 .......................................................................... 27
AM 152 fol. – 1500-1525 ............................................................................. 28
Oddur Gottskálksson’s New Testament Translation– 1540 ........................... 28
Bishop Guðbrand Þorláksson’s Bible Translation – 1584 .............................. 29
Overview of Primary Corpus ........................................................................ 30
4.2 Njáls saga Corpus ...................................................................................................... 30
Njáls saga Corpus Manuscripts .................................................................... 34
4.2.1.1 GKS 2870 4° – 1290-1310 – Gráskinna ....................................................... 34
4.2.1.2 AM 162 B β fol. – c1300 ........................................................................ 34
4.2.1.3 AM 162 b fol. δ – c1300 ........................................................................ 34
4.2.1.4 AM 468 4° – 1300-1324 – Reykjabók .................................................... 35
4.2.1.5 AM 132 fol. – 1330-1370 – Mǫðruvallabók ............................................ 35
4.2.1.6 AM 133 fol. – c1350 – Kálfalækjarbók ................................................... 35
Laidlaw 2
4.2.1.7 AM 162 B ε, fol. – 1350-1375................................................................ 35
4.2.1.8 AM 162 B α fol. – 1400-1500 ................................................................ 36
4.2.1.9 AM 466 4° – c1460 – Oddabók .............................................................. 36
4.2.1.10 AM 470 4° – 1620-1670 ........................................................................ 36
4.2.1.11 AM 464 4° – 1697.................................................................................. 37
Overview of Njáls saga Corpus research ....................................................... 38
4.3 Laxdæla saga Corpus ................................................................................................. 38
Laxdæla saga Corpus Manuscripts ................................................................ 40
4.3.1.1 AM 132 fol. – 1330-1370 – Mǫðruvallabók ............................................ 41
4.3.1.2 AM 309 4° – 1498 – Bæjarbók í Flóa ...................................................... 41
4.3.1.3 AM 158 fol. – 1630-1675 ....................................................................... 41
4.3.1.4 ÍB 225 4° – 1686-1687 – Copy of Vatnshyrna ....................................... 42
4.3.1.5 ÍB 226 4° – 1680-1699 ........................................................................... 42
Overview of Laxdæla saga Corpus Research ................................................. 43
5. Analysis of Available Data .................................................................................................. 43
5.1 Occurrence of Syncope in the Dataset Over Time .................................................... 43
Overall Percentage of Syncopated vs Unsyncopated Attestations ................. 43
Overall Percentage of Lemmata Demonstrating Syncope ............................. 48
Syncopated Attestations and Syncopated Lemmata ...................................... 49
5.2 Frequency and Syncope in -igr and -ugr Adjectives .................................................. 50
Methodology ................................................................................................ 50
Most Frequent -igr and -ugr Adjectives ........................................................ 52
Syncope and Frequency in Old Icelandic Adjectives .................................... 52
Bynames contributing to Retention of Syncopated Forms ............................ 54
On heilagr ..................................................................................................... 55
5.3 Phonological Environment ....................................................................................... 55
Consonant Clusters Created and Avoided .................................................... 56
5.4 Factors Contributing to Vowel Retention ................................................................. 58
u-Epenthesis ................................................................................................. 58
Analogical Forces in -igr and -ugr Adjectives ............................................... 60
6. Conclusions and Areas for Further Study ........................................................................... 61
Appendix A – The Most Common -igr and -ugr Adjectives ................................................... 63
Appendix A – continued ......................................................................................................... 64
Bibliography ............................................................................................................................ 65
Laidlaw 3
Acknowledgements
Ég veit hvað ég hefi reynt og ég veit hvers virði það er mér
- Vilhjálmur Stefánsson
This thesis, having sat unfinished for some time, owes its successful completion to more than a few
people. At Háskóli Íslands, I owe a debt of gratitude to Torfi Tulinius for encouraging me to return
to the school to complete the Medieval Icelandic Studies programme. Haraldur Bernharðsson has
been an exceptional supervisor, with great support and guidance in these uncharted research
territories. Without his patience and direction, this would have been an impossible task. Stofnun Árna
Magnúsonnar í íslenskum fræðum has been enormously useful as a resource and centre for examining
these otherwise difficult to track down manuscripts. In addition, I would like to thank Jesse Byock for
his material support of office space, as well as his writing suggestions.
My deep and sincere thanks go to my friends and family, particularly my parents John and
Norma, for nudging, encouraging, and sustaining me during my studies and writing. More abstractly,
I must acknowledge and thank my maternal grandparents for inspiring in me a desire to learn more
about my Nordic heritage. I would like to dedicate this work to their memory.
Finally, I must thank my wife, Heather, for her unflagging support for me as I have struggled
with this project. None of this would be written were it not for her belief in me, so much stronger
when I lacked the same faith in myself. I will never be able to sufficiently articulate my gratitude to
her for helping ease this albatross from my neck. My love, always.
Laidlaw 4
Abstract
In classical Old Icelandic, disyllabic-stemmed adjectives regularly exhibited vowel syncope in the
second syllable when inflected with a vowel-initial ending, exhibiting intra-paradigmatic alternation
between a disyllabic and monosyllabic stem. At some point in the transition to Modern Icelandic, this
alternation was lost in adjectives ending in -igr and -ugr. There has been no research into the
behaviour of this syncope pattern over time, making it an intriguing problem to investigate. By
assembling a broad new dataset this research also provides a starting point for deeper explorations of
the matter.
To examine the disappearance over time, data from 30 manuscripts dating from the late
twelfth to late seventeenth centuries has been compiled. Attestations of the adjectives with syncope-
triggering inflection are tabulated to give an overview of the presence of the pattern in the corpus.
Where possible, transmission of target adjectives between copies of Njáls saga, Laxdæla saga, and
some of the Fornaldarsögur Norðurlanda is examined. In addition, a list of the most common target
adjectives is formed by combining various wordlists and data sets.
Syncope is demonstrated to begin its decline in usage toward the end of the thirteenth
century, with almost complete abandonment of it in novel texts by the end of the sixteenth. Scribal
conservatism may account for syncopated adjectives occurring in later manuscripts. Syncope is found
to not be universal within the -igr and -ugr adjectives. There is no clear phonological reason some
disyllabic adjectives never exhibited syncope, though a hitherto undocumented tendency towards
having a medial dental sound is found in the adjectives which never underwent syncope. The
disappearance of the syncope can be characterized as paradigmatic levelling where the alternation of
two stem variants, a disyllabic one and a monosyllabic one, was levelled out in favor of the disyllabic
stem. This paradigmatic levelling is largely contemporaneous with the u-epenthesis in nom. sing.
masc. Even if the u-epenthesis cannot be claimed to have triggered the levelling in the adjectives, it
seems not implausible that the u-epenthesis accelerated the levelling.
Laidlaw 5
Ágrip
Í tvíkvæðum lýsingarorðum í forníslensku féll sérhljóð síðara atkvæðis brott á undan endingu sem
hófst á sérhljóði og í beygingu þessara lýsingarorða víxluðust því einkvæður og tvíkvæður stofn eftir
föstu mynstri. Í lýsingarorðum með -igr og -ugr hurfu þessi víxl þó einhvern tíma á leið til
nútímaíslensku. Þessi breyting hefur ekki verið rannsökuð sérstaklega. Hér verður ráðin á því nokkur
bót, en ætla má að með víðtækri söfnun gagna megi varpa ljósi á breytinguna.
Dæmum var safnað til rannsóknarinnar úr 30 handritum frá tólftu öld og fram undir lok
sautjándu aldar. Myndum lýsingarorða sem sýndu brottfall í fornu máli var safnað og með því gefið
yfirlit yfir hvarf brottfallsreglunnar. Þá var einnig kannað hvernig slíkum lýsingarorðamyndum reiddi
af í nokkrum misgömlum handritum Njáls sögu, Laxdæla sögu og nokkrum fornaldarsögum. Á
grundvelli ýmissa orðalista og textasafna var útbúinn listi yfir algengustu lýsingarorð þessarar
tegundar.
Rannsóknin bendir til að byrjað hafi að draga úr brottfalli síðara sérhljóðsins í stofni þessara
lýsingarorða undir lok þrettándu aldar og að við lok sextándu aldar hafi brottfallsreglan hafi verið svo
gott sem horfin. Brottfall virðist þó ekki hafa verið alveg undantekningarlaust í lýsingarorðum með -
igr og -ugr. Ekki verður bent á neina augljósa hljóðkerfislega ástæðu fyrir því að brottfalls hefur ekki
orðið vart í sumum lýsingarorðanna, en þó má benda á að sum þessara lýsingarorða hafa
tannbergsmælt önghljóð í stofni. Hvarfi brottfallsreglunnar má lýsa sem útjöfnun í beygingardæmi þar
sem víxlum tveggja ólíkra stofnbrigða, einkvæðs og tvíkvæðs, var eytt með alhæfingu þess tvíkvæða.
Þessi áhrifsbreyting virðist að miklu leyti samtíða u-innskoti í nefnifalli eintölu í karlkyni. Enda þótt
ekki verði bent á u-innskot sem ástæðu breytingarinnar virðist ekki ósennilegt að u-innskotið hafi ýtt
undir hana.
Laidlaw 6
1. Introduction & Research Goals
1.1 Introduction
Old Norse, being an inflected language with a robust written corpus, provides many traceable
examples of morphological change over time, which lend themselves to study in the realms of
historical linguistics. While many of the obvious changes have been examined, the advent of
digitization of works, wordlists, and manuscripts has made easier the examination of hitherto
unexplained (or unrecognized) phenomena. The subset of disyllabic adjectives which have a stem-
final -igr or -ugr in Standardized Old Icelandic tended to undergo vowel syncope and exhibit an
alternation between a disyllabic and monosyllabic stem in certain morphophonological conditions.
While this, in and of itself, may be relatively easily explained phonologically, it brings to light a
change that happened sometime before Modern Icelandic became standardized, namely the
disappearance of this vowel syncope from this environment.
At present, there is no comprehensive analysis of the decline of the syncopated form of -igr
and -ugr disyllabic adjective stems. Examinations of individual manuscripts and works make note of
whether these adjectives surface in syncopated or unsyncopated form (e.g. Oskar Bandle’s Die Sprache
der Gudbrandsbiblia, Jón Helgason’s Málið á Nýjatestamenti Odds Gottskálkssonar, and Andrea de
Leeuw van Weenen’s A Grammar of Mǫðruvallabók) however no discussion as to the reason behind
this change is presently available. The results, therefore, contribute entirely new data to the field,
including a small but until now unobserved phonological pattern. Due to the lack of pre-existing
scholarship in the field, this study is heavily reliant upon primary sources, mainly lemmatized wordlists
and facsimile or diplomatic editions of manuscripts.
1.2 Research Goals
While the difference in form is observed between the oldest manuscripts and later ones, there is no
indication of when the syncope rule fell out of favour with Old Icelandic speakers, and in turn, with
the scribes recording works in the language. Similarly, to the lack of knowledge regarding when the
rule broke down, there is no explanation of why it disappeared, or why it only effected -igr and -ugr
Laidlaw 7
disyllabic adjective stems, and not other disyllabic adjective stems which also exhibited vowel
syncope.
In order to fill the gaps in the linguistic record, I aim to answer two questions:
a) When did vowel syncope cease in -igr and -ugr stemmed disyllabic adjectives?
b) What caused vowel syncope to cease in these stems and not in others?
2. Vowel Syncope in Disyllabic-stemmed Adjectives
2.1 Overview of Vowel Syncope
Disyllabic-stemmed adjectives in Old Norse-Icelandic exhibited an intra-paradigmatic
alternation between a disyllabic stem preceding unmarked and consonant-initial endings, and a
monosyllabic stem preceding vowel-initial endings. For example, in the early manuscripts adjectives
such as auðigr ‘wealthy’ would surface as auðgir when inflected with the vowel-initial nominative plural
masculine1 ending -ir. At some point in time, this syncope stopped being exhibited, and the NPM
form came to be realized as auðigir.2 It is worth noting at this point that the standardized forms
presented as lemmata and examples throughout this thesis follow the conventions of Classical Old
Icelandic spelling and orthography, broadly representing the state of the language circa 1200 and
reflecting the forms present in the Dictionary of Old Norse Prose. 3
Not restricted to -igr and -ugr adjectives, other disyllabic adjectives such as gamall ‘old’ would
exhibit syncope of the final vowel of the stem, and affix the appropriate ending. Examples of gamall
and auðigr are shown in Table 1 and Table 1 below, demonstrating strong adjective inflection. In
addition, weak inflection of both is demonstrated in Table 3 and Table 3, as weak inflectional endings
are all vowel-initial, and therefore potentially trigger syncope. Declensions of the adjectives are readily
available in grammars of Old Icelandic, such as E.V. Gordon’s Introduction to Old Norse (1957, 289-
1 For the sake of brevity, future references to case, number, and gender, will be abbreviated with the following
convention: N(ominative), A(ccusative), D(ative), G(enitive) – S(ingular), P(lural) – M(asculine), F(eminine),
N(euter). Thus nominative, plural, masculine, will be rendered NPM. Comparative and Superlative, when
relevant, are noted as c and s, respectively, following the Case-Number-Gender. 2 Eventually it would be realized as auðugir, though the change in the final medial vowel in the stem falls
outside the scope of this investigation. 3 Ordbog over det Norrøne Prosasprog (‘A Dictionary of Old Norse Prose’), abbreviated ONP, is a substantial
database of individual word citations, invaluable for finding instances of particular lemmata.
Laidlaw 8
292) and Adolf Noreen’s Altisländische und altnorwegische Grammatik (1923, 288-303). Cells
containing forms which exhibit syncope are shaded.
Table 1: Strong inflection of gamall ‘old’ Table 2: Strong inflection of auðigr ‘wealthy’
Num. Case Masc. Fem. Neut. Num. Case Masc. Fem. Neut.
Sing.
N gamall gǫmul gamalt
Sing.
N auðigr auðig auðigt
A gamlan gamla gamalt A auðgan auðga auðigt
D gǫmlum gamalli gǫmlu D auðgum auðigri auðgu
G gamals gamallar gamals G auðigs auðigrar auðigs
Plur.
N gamlir gamlar gǫmul
Plur.
N auðgir auðgar auðig
A gamla gamlar gǫmul A auðga auðgar auðig
D gǫmlum gǫmlum gǫmlum D auðgum auðgum auðgum
G gamalla gamalla gamalla G auðigra auðigra auðigra
Table 3: Weak inflection of gamall Table 4: Weak inflection of auðigr
Num. Case Masc. Fem. Neut. Num. Case Masc. Fem. Neut.
Sing.
N gamli gamla gamla
Sing.
N auðgi auðga auðga
A gamla gǫmlu gamla A auðga auðgu auðga
D gamla gǫmlu gamla D auðga auðgu auðga
G gamla gǫmlu gamla G auðga auðgu auðga
Plur.
N gǫmlu gǫmlu gǫmlu
Plur.
N auðgu auðgu auðgu
A gǫmlu gǫmlu gǫmlu A auðgu auðgu auðgu
D gǫmlum gǫmlum gǫmlum D auðgum auðgum auðgum
G gǫmlu gǫmlu gǫmlu G auðgu auðgu auðgu
When syncope fell out of use the paradigm was realized with a disyllabic stem in all cases. Below,
auðigr is declined in both strong and weak forms in Table 5 and Table 5 respectively.
Table 5: Strong inflection of auðigr without syncope Table 6: Weak inflection of auðigr without syncope
Num. Case Masc. Fem. Neut. Num. Case Masc. Fem. Neut.
Sing.
N auðigr auðig auðigt
Sing.
N auðigi auðiga auðiga
A auðigan auðiga auðigt A auðiga auðigu auðiga
D auðigum auðigri auðigu D auðiga auðigu auðiga
G auðigs auðigrar auðigs G auðiga auðigu auðiga
Plur.
N auðigir auðigar auðig
Plur.
N auðigu auðigu auðigu
A auðiga auðigar auðig A auðigu auðigu auðigu
D auðigum auðigum auðigum D auðigum auðigum auðigum
G auðigra auðigra auðigra G auðigu auðigu auðigu
Laidlaw 9
It should be noted, however that this applies to adjectives that are polysyllabic in the stem, and not
ones that have gained multiple syllables through stem affixation. Adjectives formed through
compounding a monosyllabic stem with various prefixes would still behave in the manner of the
original monosyllabic adjective. Thus, halsdigr ‘fat necked’ or fótdigr ‘fat footed’ inflected in the same
manner as digr ‘fat, big’ and do not exhibit vowel syncope, and forms such as *halsdgan or *fótdgan
were not formed. Admittedly, there are few attested examples of this, as there are few monosyllabic -
igr/-ugr final adjectives. In examining -igr and -ugr final adjectives, it ought to be noted that -ligr
final adjectives are, by and large, excluded from this analysis, due to the tendency for the productive
-ligr morpheme to remain intact and not exhibit vowel syncope. Examples of this include sárligr
‘sore’, shown in Table 7 below, and dýrligr ‘glorious’. That said, málugr ‘talkative’ occasionally
surfaced with an “i” in final vowel position, rather than a “u”, giving an example of an adjective that
is -ligr terminal, while still retaining syncope. The variation between these two forms is relatively free,
and for purposes of glossing, the ONP spelling málugr was chosen. Though this creates the same -ligr
environment that otherwise would not exhibit syncope, the form would be understood to be derived
from the root morpheme mál ‘speech,’ and not as one formed through the affixation of the -ligr suffix.
In addition to -ligr terminal adjectives, some other -igr/-ugr terminal adjectives did not exhibit
syncope an any attestations throughout the corpus of Old Norse literature, despite containing near-
identical vocalic and consonantal environments. One such example, verðugr ‘worthy’ is inflected
below in Table 7.
Table 7: Inflection of sárligr ‘sore’ Table 8: Inflection of verðugr ‘worthy’
Num. Case Masc. Fem. Neut. Num. Case Masc. Fem. Neut.
Sing.
N sárligr sárlig sárligt
Sing.
N verðugr verðug verðugt
A sárligan sárliga sárligt A verðugan verðuga verðugt
D sárligum sárligri sárligu D verðugum verðugri verðugu
G sárligs sárligrar sárligs G verðugs verðugrar verðugs
Plur.
N sárligir sárligar sárlig
Plur.
N verðugir verðugar verðug
A sárliga sárligar sárlig A verðuga verðugar verðug
D sárligum sárligum sárligum D verðugum verðugum verðugum
G sárligra sárligra sárligra G verðugra verðugra verðugra
Laidlaw 10
Like the adjectives formed with the productive -ligr morpheme, some other adjectives that were -igr
or -ugr terminal never exhibited vowel syncope, such as verðugr ‘worthy’ and burðugr ‘high-birthed’.
While these adjectives obviously cannot be examined to directly measure the disappearance of vowel
syncope, they are useful to include in the examination for a few reasons. First, their presence in the
dataset can add to information about the overall prevalence of syncope in this type of adjective, giving
a broad idea of the state and strength of the phenomenon. Secondly, with information about their
frequency and date of attestation, one can make more inferences about the timing of the disappearance
of vowel syncope. These unsyncopated adjectives also raise the point that it is unknown how broadly
syncope in this type of adjectives was applied be speakers of the language at the time. Ultimately, this
is unknowable, as there is no more data available than what is contained within the earliest
manuscripts.
2.2 The case of heilagr
Potentially muddying the examination of vowel syncope is the behaviour of heilagr ‘holy’ which
exhibits both a root alternation between a disyllabic stem and a monosyllabic stem, but also a
monophthongization of the stem diphthong where the stem /-ei-/ alternates with /-e-/ when the
stem-final /-a-/ is syncopated. This syncope occurs in the same environments as with other disyllabic
adjective stems. Below in Table 9, heilagr is shown following strong declension, and examples of
monophthongization and syncope are in shaded cells.
Table 9: Inflection of heilagr ‘holy’
Num. Case Masc. Fem. Neut.
Sing.
N heilagr heilǫg heilagt
A helgan helga heilagt
D helgum heilagri helgu
G heilags heilagrar heilags
Plur.
N helgir helgar heilǫg
A helga helgar heilǫg
D helgum helgum helgum
G heilagra heilagra heilagra
Laidlaw 11
While the existence of a different final vowel in the disyllabic stem and a monophthongization process
are not inherently problematic, the principle difficulty that heilagr presents is that it does not, over
time, lose syncope. The stem allomorphy between heilag- and helg- was, evidently, rather difficult for
speakers to continue, and competing paradigms eventually arose due to paradigmatic levelling, one
with a monosyllabic stem, and one with a disyllabic stem. The existence of both forms, and the
retention of a syncopated form must be addressed in any analysis of this adjective type. The
competing, monosyllabic helg- stemmed declension is below in Table 10. It ought to be noted that
the declension is presented with the NSM epenthetical -u-, which features in later discussion of the
sound changes. Monosyllabic stem forms are attested in ONP with consonant-initial endings for the
first time in the late fourteenth century.
Table 10: Inflection of helgr ‘holy’
Num. Case Masc. Fem. Neut.
Sing.
N helgur helg helgt
A helgan helga helgt
D helgum helgri helgu
G helgs helgrar helgs
Plur.
N helgir helgar helg
A helga helgar helg
D helgum helgum helgum
G helgra helgra helgra
Further complicating the case of the competing paradigms arising from stem alternation in heilagr is
the existence of the nominal forms Helgi and Helga (arising from the form inn helgi ‘the holy’), and
the substantive helgr ‘holiday.’ Special attention is paid to heilagr in §5.2.5.
3. Methodology
The amount of work required to conclusively determine the dating of the disappearance of
vowel syncope exceeds the scope of this thesis. Such a task would require the examination of a very
large cross section of manuscripts, mostly untranscribed ones, and would take an extreme amount of
time. Preliminary investigation of the matter, however, can be accomplished through the examination
of what transcribed sources we have available. Transcribed and lemmatized sources, however, can
only take the investigation so far, and need to be augmented through examination of unpublished
Laidlaw 12
manuscript texts. Broadly speaking, the manuscripts consulted can be broken into four categories
based on research method, herein labelled as Groups A through D, with Group D broken into three
subgroups based on texts consulted.
Group A consists of the following manuscripts: AM 674 4°, AM 673 a 4°, AM 673 b 4°, Holm
perg 15 4°, AM 645 4°, GKS 1157 fol., AM 291 4°, AM 519 a 4°, DG 11 4°, AM 132 fol., & AM 351
fol. (nb. AM 132 fol. is also present in groups D1 and D2). First, lemmatized indices were consulted
wherever possible. These are available for some of the earliest of Icelandic manuscripts, covering a
representative sample of the first few hundred years of the written record. These sources cover the
earliest manuscripts up to the end of the fourteenth century with AM 351 fol. This method, of course,
is not without flaws, mainly the manual nature of searching the wordlists for relevant adjectives, as all
were done by hand, apart from AM 519 a 4°, which was electronically searchable and lemmatized. No
doubt, some examples were likely missed in these works, however the work holds up and covers the
majority of the relevant lemmata contained within the manuscripts.
Group B consists of the following manuscripts and books: GKS 1009 fol., Oddur
Gottskálksson’s New Testament translation published in 1540, and Bishop Guðbrandur Þorláksson’s
Bible translation published in 1584. These works were all examined through previous academic study
of the relevant texts, and their observations of adjectives of the same type. As a result, there is less
raw data present, as none of the works consulted contained word lists, but, as will be shown in their
appropriate sections, the relevant data is quite clear.
Group C consists of a single manuscript: GKS 1005 fol., Flateyjarbók. The method of
examination of Flateyjarbók was the most involved of the manuscripts consulted. First, the nineteenth
century diplomatic edition of the manuscript edited by Guðbrandur Vigfusson and Carl Rikard Unger
was consulted. Then, using the OCR available at archive.org, the text of the three volumes of the
diplomatic edition were copied and formed into a single, searchable text file. The text was searched
for vowel-initial inflectional endings preceded by “g” (e.g. -gir, -gan), and while this yielded results
that are not germane to the study (e.g. eigan, tuttugu) all non-relevant forms were combed out of the
data set. It should be noted, though, that the scan of Vol. 2 of Flateyjarbók’s diplomatic edition was
not of particularly good quality, and as such the OCR had quite some difficulty and forms could be
Laidlaw 13
missing from its data. Where possible, the OCR’s results were cleaned, however there is undoubtedly
some missing data. The difficulties that the OCR faced were, at least, evenly distributed throughout
Vol. 2, and should not affect the statistical significance of the results.
Group D consists of three subgroups of manuscripts, each set an examination of the adjectives
in transmission between copies of different texts. Because of their multiple attestations in manuscripts
throughout the corpus, Brennu-Njáls saga and Laxdæla saga were chosen, as well as selections from
the Fornaldarsǫgur Norðurlanda.
Groups D1 and D2 examine copies of Brennu-Njáls saga and Laxdæla saga respectively. D1
consists of the following manuscripts: GKS 2870 4°, AM 162 B β fol., AM 162 B δ fol., AM 468 4°,
AM 132 fol., AM 133 fol., AM 162 B ε fol., AM 162 B α fol., AM 466 4°, AM 470 4°, and AM 464
4°. Group D2 consists of the following manuscripts: AM 132 fol., AM 309 4°, AM 158 fol., ÍB 225
4°, and ÍB 226 4°. To perform the analysis, standardized editions of these sagas were searched for
target adjectives in forms that would exhibit vowel syncope. Then, noting approximate location of
these adjectives in the text, photographic and facsimile copies of manuscripts containing these sagas
were manually searched. When examining a manuscript in this fashion, the saga text itself was focused
on, and the search did not broaden to other texts contained within the manuscript. Due to the manual
nature of both the initial search for target adjectives and the subsequent search for attestations within
the manuscripts, there is some margin of error. While it cannot be claimed that the forms found
constitute the definitive list of such adjectives in those sagas, they are likely to be the bulk of them,
and a representative sample.
Group D3 consists of several manuscripts in which some of the Fornaldarsǫgur Norðurlanda
are attested: AM 152 fol., GKS 2845 4°, AM 343a 4°, and Holm perg 7 4°. The search methodology
for the Fornaldarsǫgur was broadly similar to that of Njáls saga and Laxdæla saga, though with some
differences. To form the initial word list for these a digital copy of standardized editions of the texts
was searched for four of the most common -igr and -ugr adjectives: auðigr ‘wealthy’, nauðigr
‘unwilling’, gǫfugr ‘noble’, and máligr ‘talkative’. Following this, four manuscripts were selected which
showed some overlap of texts that exhibited some of the target adjectives. Not all of the texts
overlapped between the manuscripts, but there exists enough for a comparison to be of some use in
Laidlaw 14
examining both transmission and novel attestation. The same caveat regarding the completeness of
the dataset that applies to the analysis of Njáls saga and Laxdæla saga also applies to the examination
of the Fornaldarsǫgur manuscripts.
4. The Research Corpora
4.1 The Primary Corpus
The primary research corpus consists of manuscript groups A, B, C, D1, and D2, as described
in §3. Findings from each manuscript are broken down in the following way. First, examples are
sorted by lemma, and then again by inflectional ending, as it appears in the manuscripts. The form
the adjective takes in the manuscript or in the glossing text is then given in rather broad transcription.
The following orthographic conventions are employed in transcription. Differing forms of consonants
are represented with their standard Caroline form, for example, all instances of Insular F are
transcribed as Caroline “f.” Consonantal ligatures are also rendered as their two individual letters.
While these differences are of some interest from a paleographic standpoint, they do not play a role
in the phonological environment.
Afterwards, the number of adjectives appearing with the appropriate inflectional endings to
potentially trigger vowel syncope is tabulated, giving a value ‘T’. All targets are then checked to see
if they did in fact undergo vowel syncope, giving value ‘S’. The percentage of potential targets that
exhibit vowel, ‘S%’, syncope is then calculated, using the formula (S/T) * 100. The lemmata are given
in normalized classical Old Icelandic. Note that as part of this overall number, adjectives that are -igr
and -ugr terminal that never have displayed syncope, such as verðugr, are included in the totals. This
is to give an overall picture of syncope in adjectives with the target phonological environment.
Instances of forms of heilagr were noted, but not counted in the overall T-value. This choice
was made due to the regularity of the behaviour of heilagr throughout the corpus of Old Icelandic
texts. Likewise, terms derived from heilagr, (i.e. lǫgheilagr) were omitted from the T-value.
All dates for the manuscripts are taken from ONP. Unless otherwise indicated, when a target
adjective fails to display syncope in the data presented, it is in boldface type.
Laidlaw 15
AM 674 a 4° – c1150-1200 – The Icelandic Elucidarius
Table 11: AM 674 a 4° - c1150-1200
Lemma CNG Form Count
nauðigr DSM “nauþgom” 1
NPM “nauþger” 1
ómáttigr ASM “omotkan” 1
T = 3, S = 3, S% = 100%
AM 674 a 4° is the oldest of the manuscripts consulted, and one of the oldest Icelandic manuscripts
in general. It is made up of 33 leaves and is believed to have been written in Iceland, either towards
the end of the twelfth century, or the very beginning of the thirteenth (Kålund 1894, 92). Despite its
relatively large size, thirty-three leaves, it contains only a few examples of the target adjectives in a
syncope-inducing environment. Data from this manuscript was taken from Ludvig Larsson’s
Ordförrådet i de älsta islänska handskrifterna (1891) and lacked the specifics of orthographical
representation of the words in the source manuscript, with its information being transcribed by
Larsson from a diplomatic edition of the text. Orthographically, it does illustrate devoicing in the
root-terminal “g” to “k” in máttigr when syncopated.
AM 673 a II 4° – c1175-1225 – Physiologus
Table 12: AM 673 a 4° – c1175-1225
Lemma CNG Form Count
almáttigr ASM “almatkan” 2
DSM “alMotcom” 1
auðigr NPM “ꜷþger” 1
T = 4, S = 4, S% = 100%
Also covered in Larsson’s Ordförrådet (1891), AM 673a 4° is a short manuscript consisting of only
nine leaves, written in Iceland at turn of the thirteenth century. (Kålund 1894, 90-91) It adds little to
the data set, but the examples contained show syncope being consistently applied, as well as devoicing
of the root-terminal “g”.
Laidlaw 16
AM 673 b 4° – c1175-1225 – Placítus drápa
Table 13: AM 673 b 4° – c1175-1225
Lemma CNG Form Count
aldrgǫfugr DSN “aldrgꜵfgo” 1
T = 1, S = 1, S% = 100%
Presumably written in Iceland at the turn of the thirteenth century, AM 673b 4°, is made up of five
leaves and once part of a larger manuscript with the preceding AM 673a II 4° (Tucker, 1998, xcii-
cxxv). It adds little to the dataset and is presented here for the sake of completeness. The only attested
adjective appears in syncopated form. (Larsson 1891)
Holm perg 15 4° – c1200 – The Icelandic Homily Book
Table 14: Holm perg 15 4° – c1200
Lemma CNG Form Count Lemma CNG Form Count
almáttigr
ASM
“almátkan” 14
gǫfugr
ASMs “gꜵfgasta” 1
“almatkan” 4 DSFs “gꜵfgosto” 1
“álmatkan” 2 NPF “gꜵfgar” 1
“állmatkan” 1 NPM “gꜵfger” 1
DSM
“almótkom” 25 NPMc “gꜵfgare” 1
“almottkom” 2 NPMc “gꜵfgari” 1
“álmótkom” 1 NPMsw “gꜵfgosto” 1
“almóttkom” 1 NSFc “gꜵfgare” 4
“almotkom” 2 NSFc “gofgari” 1
auðigr
APM “auþga” 1 NSFs “gǫfgost” 1
ASM “auþgan” 1 NSMs “gꜵfgastr” 2
DPM “auþgom” 1 illmálugr NPM “illmólger” 1
“ þgom” 1 lastauðigr DPM “lastaúþgom” 1
NPM “auþger” 1 lostigr APM “losca” 1
bakmáligr NPM “bacmólger” 1 minnigr NSFc “miNgare” 1
dreyrugr ASM “dreýrgan” 1 ógǫfugr APM “ gǫfga” 1
fjǫlkunnigr ASM “fiolkungan” 1 ómáttigr NPM “ómótker” 1
gáðugr ASM “gróþgan” 1 ǫflugr NPM “oflger” 1
góðmálugr NPM “góþmólger” 1 T = 84, S = 84, S% = 100%
Holm perg 15 4°, The Icelandic Homily Book (Íslenska hómilíubókin), a sizable manuscript of 102
leaves, contains 62 homilies and saga excerpts, and shares a number of texts with the AM 619 4°, the
Laidlaw 17
Old Norwegian Himily Book (de Leeuw van Weenen 1993, 1-7). It contains a large number of
examples of appropriate adjectives, all of which display syncope in relevant environments. Predictably,
due to the nature of the book, many heilagr-derived adjectives are attested, and display syncope in
the ending and alternation between the pure vocalic stem and diphthong. (Larsson 1891)
AM 645 4° – c1220-1249
Table 15: AM 645 4° – c1220-1249
Lemma CNG Form Count Lemma CNG Form Count
almáttigr
ASM
almatcan 3
fjǫlkunnigr
(cont.)
NPMw fiolcvŋo 4
almátcan 1 NPMw fiolkungo 3
allmatkan 1
NSMw
fiolcvŋe 1
ASMw almatca 3 fiǫlcvɴge 1
DPM almatcom 1 fiolcvŋi 2
DSM
almatkom 5
gǫfugr
APNw gaufgo 1
almátkom 1 gøfgo 1
almotkom 3 ASM gøfgan 1
GSM almategs 1 ASMw gøfga 1
GSM almatex 3 DPM gøfgom 1
almattegs 1 NPM gøfger 1
auðigr
DSF auþegre 1 NPMs gøfgaster 1
GSM auþegs 1 NPMsw gøfgusto 1
NSF auþeg 1 NSF gøfvg 2
fjǫlkunnigr
ASM fiǫlcvŋan 1 NSM gøfvgr 4
fiolcvŋan 2 NSMs gøfgastr 1
DSMw fiokvŋa 1 NSMw gøfge 1
GPMw fiolcvŋo 1 kviðugr NSF qviþog 2
GSM fiolcvnniŋs 1 máttigr NSM mattegr 1
GSMw fiolcvŋa 1 ógǫfugr ASM ogǫfgan 1
NPM fiolcvŋer 2 T = 49, S = 49, S% = 100%
AM 645 4° is of similar subject material to Holm perg 15 4°, though it is a collection of saints’ lives
rather than a homiliary, and has a similar size of dataset. It is made up of 66 leaves, and believed to
have been written in or around Skálholt in the early thirteenth century. (Kålund 1894, 51-52) All -
igr and -ugr lemmata attested are also found in The Icelandic Homily Book, and exhibit syncope in
Laidlaw 18
the predicted places. Likewise, devoicing of /g/ still appears in forms of almáttigr, with the devoiced
segment surfacing as both “c” and “k”, in free variation. Differentiating the paleography somewhat,
/-ng-/ surfaces as “ng”, “ɴg” and “ŋ”, though the varying declensions of fjǫlkunnigr. (Larsson 1891)
GKS 1157 fol – c1250 – Konungsbók Grágásar
Table 16: GKS 1157 fol – c1250 – Konungsbók Grágásar
Lemma CNG Form Count Lemma CNG Form Count
almáttigr ASM “almátkan” 1 lǫg-heilagr
(cont.)
ASF “lǫg-helga” 2
eykt-heilagr ASM “eykt-hælgan” 1 “lǫg-hælga” 1
heilagr
NSMw “hælgi” 2 NPF “lǫg-hælgar” 2
DSM “helgum” 1 nón-heilagr ASM “nón-helgan” 2
“hælgum” 1
óheilagr
DSM “óhelgum” 1
ASM “helgan” 2 ASM “óhelgan” 1
ASF “helga” 1 NPM “óhelgir” 1
NPM “hælgir” 3 NPF “óhelgar” 1
hǫfugr ASF “hofga” 1 nauðigr ASF “nauðga” 2
lǫg-heilagr
DSM “lǫg-hælgum” 2 rúm-heilagr
ASM “rúm-helgan” 18
ASM “lǫg-helgan” 4 NPM “rúm-helgir” 1
“lǫg-hælgan” 1 T = 4, S = 4, S% = 100%
GKS 1157 fol., a recording of the Grey Goose law codes, made of 93 leaves (Finsen 1852, Forord),
gives a robust dataset, showing syncope to be a fully enforced rule. (Beck, 1993) That being said, the
vast majority of the attested adjectives are heilagr and its derived forms, which have always shown
vowel syncope. Devoicing is still present in máttigr-derived adjectives.
AM 291 4° – c1275-1300 – Jómsvíkinga saga
Table 17: AM 291 4° – c1250-1300
Lemma CNG Form Count Lemma CNG Form Count
auðigr
ASNs “ꜷþgazt” 1 blóðugr ASM “bloþgan” 1
GSMw “ꜷþga” 1 gǫfugr ASM “gꜷfgara” 1
NPNs “ꜷþguz” 1 málugr NPM “malgir” 1
NSNs “ꜷþgaz” 2 T = 8, S = 8, S% = 100%
AM 291 4ᵒ, of a similar age to GKS 1157 fol., is a parchment manuscript made up of 38 leaves, and
the earliest attestation of the legendary Jómsvíkinga saga. (Kålund 1888, 538) It, like manuscripts
before it shows a robust and active syncope rule, although only a few examples of the target
Laidlaw 19
environment. These examples cover a broad range of lemmata, however, showing the overall
applicability of the rule. (Larsson 1956)
GKS 1009 fol. – c1275 - Morkinskinna
Morkinskinna, literally ‘Rotten Parchment’ (on account of its poor condition) is a parchment
manuscript made of thirty-seven leaves. It contains a history of the kings of Norway from ca. 1025 to
1157, and holds many examples of skaldic poetry in the various þættir contained within (Finnur
Jónsson, 1932, III-XL). Information on GKS 1009 fol. was taken from a Ph.D. dissertation on the
manuscript which does not include a complete wordlist (Kjeldsen, 2013). In that investigation, a broad
range of -igr and -ugr terminal adjectives were attested, covering 26 lemmata, and all adjectives exhibit
vowel syncope in the relevant conditions. In total, syncope occurs a total of 27 times, 17 in strong
declension, and 10 in weak, though it is not noted how many of those attestations of syncope are in
heilagr type adjectives, which do occur in the manuscript. Additionally, nauðigr, minnigr, ráðigr, and
saurigr, only occur in syncopated form. (Kjeldsen, 2013, A.3.1.1)
AM 519 a 4° – c1280 – Alexanders saga
Table 18: AM 519 a 4° – c1280
Lemma CNG Form Count Lemma CNG Form Count
ánauðigr NPM “ánauðgir” 1 nauðigr
ASF “nauðga” 1
auðigr NSFc “auðgari” 1 APF “nauðgar” 1
gǫfugr NPM “gǫfgir” 1 skuldugr NPM “sculldugir” 1
heilagr NPM “helgir” 1 stǫðugr NPM “stǫðugir” 1
DPF “helgum” 2 T = 10, S = 8, S% = 80%
Written in the late thirteenth century, AM 519 a 4ᵒ is made up of 37 leaves (Kålund 1888, 673), and
shows a strong tendency towards syncopation, however there are two examples counter to this trend,
which bear some additional study (de Leeuw van Weenen, 2009). These are the first examples of the
disappearance of syncope within the corpus covered in this investigation. The appearance of skuldugr
‘owing’ in AM 519 4ᵒ is the first attestation of the word in the ONP catalogue of Old Norse-Icelandic
literature. In this light, the fact that it appears solely in unsyncopated form bears notice. Similarly,
this is one of the first appearances of stǫðugr ‘steady’ in the corpus, and it is worth pointing out that
its initial appearance does not display syncope in the places on might expect it. Neither skuldugr nor
Laidlaw 20
stǫðugr go on to display any syncope in future attestations of the words. It remains possible that these
words were never syncopated in speech prior to their attestation in Alexanders saga, however given
the paucity of information, that cannot be certain.
DG 11 – c1300-1325 – Codex Upsaliensis, Snorra Edda
Table 19: DG 11 – c1300-1325 – Codex Upsaliensis, Snorra Edda
Lemma CNG Form Count Lemma CNG Form Count
blóðugr DSN “bloðgu” 1 gǫfugr NSMsw “gǫfgasti” 1
harðmóðugr NPNw “harðmóðgu” 1 jafnhǫfugr NPM “jafnhǫfgir” 1
heilagr
DSM “halgum” 1 máttigr
NSMs “mátkastr” 1
ASM “helgan” 1 NSMw “mátki” 1
NSMw “helgi” 3 þróttǫflugr ASF “þróttoflga” 1
GSMw “Helga” 3 T = 7, S = 7, S% = 100%
NSMc “helgari” 1
DG 11, the Uppsala Codex, is the oldest surviving copy of Snorri Sturluson’s Prose Edda. Syncope is
still strong in DG 11, with no relevant examples exhibiting lack of syncope. (Grape et al., 1977) Of
some note may be harðmóðugr ‘hard-mooded’, in one of the only attestations of the word, one might
expect it to show the absence of syncope, given the initial appearance of target adjectives in similarly
dated manuscripts. Also, þróttǫflugr ‘mighty’ may bear further examination, as it creates a complex
consonant cluster.
Holm perg 7 4° – c1300-1325
Table 20: Holm perg 7 4° – c1300-1325
Lemma CNG Form Count
verðugr NSFw “verðuga” 1
nauðigr NPM “nauðgir” 2
gǫfugr DPMs “gaufguztum” 1
DPM “gaufgir” 1
ógǫfugr NPM “ogaufgum” 1
kunnigr ASM “kunnigan” 1
T = 7, S = 5, S% = 71.43%
Holm perg 7 4° is the first of the Fornaldarsǫgur Norðurlanda manuscripts consulted, and the following
texts were examined: Hrólfs saga Gautrekssonar, Ásmundar saga kappabana, and Ǫrvar-Odds saga.
Laidlaw 21
While the data set is small, there are some notable breaks in syncope. Foremost, kunnigr appears
unsyncopated for the first time in the target environment, and never (in subsequent manuscripts)
appears to regain syncope. Somewhat downplaying this, however, is that previous attestations of
kunnigr-derived adjectives were all declensions of fjǫlkunnigr, which possibly exhibited a different
syncope pattern. This is also the first appearance of verðugr in the examined corpus, and the adjective
never displays syncope in any attestations. This is in-line with observations from ONP, where data
regarding verðugr goes back as far as the early thirteenth century.
AM 132 fol. – c1339-1370 – Mǫðruvallabók
Table 21: AM 132 fol. – c1339-1370
Lemma CNG Form Count Lemma CNG Form Count
alblóðugr ASM “alblóðgan” 1
kunnigr NPM “kunnigir” 1
ASF “alblóðga” 1 NSNs “kunnigaz” 1
auðigr
ASM “augan” 1
málugr
NPF “málgar” 1
APM “augir” 1 NSFw “málga” 3
NSMw “auðgi” 7 GSFw “malgu” 2
NSMc “auðgari” 2 nauðigr
ASM “nauðgan” 1
NSMs “auðgastr” 1 DSM “nauðgum” 1
ASMs “auðgastan” 1 óheilagr
NPM “óhelgir” 1
NPMs “auðgastir” 1 APM “óhelga” 1
djúpauðigr NSFw “djúpauðga” 2 ómáttugr NPMc “ómáttgari” 1
gagnauðigr NPF “gagnauðgar” 2 ónauðigr NSMc “ónauðgari” 1
heilagr
ASM “helgan” 1 saurugr NPF “saurgar” 1
NSMw “heilagi” 1 stórauðigr NPM “stórauðgir” 1
NSMw “helgi” 2 T = 34, S = 32, S%=94.12%
ASFw “helgu” 1
AM 132 fol., Mǫðruvallabók, a significant 200-leaf saga manuscript likely written in Mǫðruvellir
(Kålund 1888, 94-97), much as contemporary works such as DG 11 and AM 519 4ᵒ, displays a
tendency towards syncopation, however there are some notable exceptions to the syncope rule. (de
Leeuw van Weenen, 2000) Also, heilagr fails to syncopate in one instance, giving the rare form heilagi.
Per ONP, heilagi appears a couple of other times in the overall corpus around the same time, and may
contribute to the competing paradigms of heilagur and helgur which appear later in Icelandic.
Laidlaw 22
Mǫðruvallabók contains a copy of Njáls saga, and transmission of the adjectives contained within is
further discussed in §4.2.
AM 351 fol. – c1350-1400 – Skáholtsbók eldri
Table 22: AM 351 fol. – c1350-1400
Lemma CNG Form Count Lemma CNG Form Count
fákunnigr ASF “fákunniga” 1
nauðigr DSM “nauðgum” 1
NPM “fákunnigir” 1 ASF “nauðga” 1
heilagr
NSMw “heilagi” 2
skyldugr
ASM “skyldugan” 1
DSM “helgum” 4 NPM “skyldugir” 2
DSF “heilagri” 2 NSMc “skyldugari” 1
DSN “helgum” 5
óheilagr
DSM “úhelgum” 1
ASM “helgan” 1 NPMw “úhelgu” 3
ASF “helga” 10 NPM “úhelgir” 1
ASF “heilaga” 7 NPF “úhelgar” 1
GPM “helgra” 2 kunnigr NSNs “kunnigast” 1
GPM “heilagra” 6
T = 9, S = 2, S% = 22.22% lǫgheilagr
DSM “lǫghelgum” 3
ASM “lǫghelgan” 1
Skálholtsbók eldri, AM 351 fol., is a law book made up of 133 leaves (Kålund 1888, 285-286), and it
differs from Mǫðruvallabók in that syncope has almost completely fallen apart, although only across
three lemmata, two of which are derivatives of kunnigr. There are not many target lemmata, however,
which, given the few examples of target adjectives in the other law book examined, GKS 1157 fol., is
not completely surprising. In adition, skyldugr ‘obligated’, which is related to skuldugr ‘owing’ as
attested in AM 519 4ᵒ, never exhibits syncope in any of its future attestations in the corpus.
Laidlaw 23
GKS 1005 fol. – c1384-1397 – Flateyjarbók
Table 23: GKS 1005 fol. – c1384-1397
Lemma CNG Form Count Lemma CNG Form Count
alblóðugr NPM “alblodgir” 1
gǫfugr
(cont.)
NPMsw “gofgazsti” 1
allsvǫldugr DSM “allzuolldugum” 1 NPMsw “gofguzstu” 2
almáttigr DSM
“almaattigum” 1 NSFs “gaufguzst” 1
“almattigum” 1 NSMc “gofgari” 1
NSM “almattigur” 1 NSMs “gaufgazstr” 1
ástúðigr NPM “astudigir” 1 NSMs “gofgazstr” 13
auðigr APM “audga” 4 NSNs “gofguzst” 1
auðigr
ASM “audgan” 1 jafngǫfugr ASM “iafngofgan” 1
ASMc “audgara” 1 kunnigr NPM “kunnigir” 2
NPF “audgar” 1 liðugr APM “liduga” 1
NPM “audgir” 6 málugr NPM “malgir” 1
“audigir” 1
máttigr
APM “mattuga” 1
NPMw “audgu” 1 DSM “mattugum” 1
NSMs “audgazstr” 2 NPM “mattugir” 1
“audgaztr” 1 minnigr NPM “minnigir” 1
NSMw “audge” 2
nauðigr
ASF “naudga” 5
NSMw “audgi” 9 ASM “naudgan” 2
NSNs
“audgazst” 1 ASM “naudgan” 1
“audgazst” 1 ASM “naudgan” 2
“audguzst” 2 ASM “naudgann” 1
blóðugr
ASM “blodgan” 1 DPM “naudgum” 1
“blodgann” 1 NPM “naudgir” 2
DSM “blodgum” 1
ǫflugr
ASM “oflgan” 1
NPF “blodgar” 2 DSM “auflgum” 2
“blodgar” 1 GSF “auflgrar” 1
burðugr NPMw “burdugu” 1 NSMs “auflgazstr” 1
dádǫflugr ASM “dádoflgan” 1 ógǫfugr NSMc “ogofgare” 1
djúpauðigr ASFw “diupaudgu” 3 ómattugr NPM “mattugir” 1
NSFw “diupaudga” 5 sálugr NPM “salugir” 1
fjǫlkunnigr DPM “fiolkunnigum” 1 skammminnigr NPM “skamminnigir” 1
Laidlaw 24
Lemma CNG Form Count Lemma CNG Form Count
fjǫlkunnigr NPM “fiolkunnigir” 2 stǫðugr
APM “stoduga” 1
gǫfugr
APM “gaufga” 2 ASM “stodugan” 1
“gofga” 4
stórauðigr
ASM “storaudgan” 1
APMs “gaufgazsta” 1 ASMs “storaudgazstann” 1
“gofgazsta” 2 DPM “storaudigum” 1
ASM “gofgann” 2 NPM “storaudgir” 1
ASMc “gofgra” 3 sverðauðigr ASM “suerdaudgan” 1
“gofgra” 1 synðugr
APM “synduga” 1
ASMs “gofgazstann” 1 DSM “syndugum” 2
DPMs “gofgauzstum” 1 þurftugr
APM “þurftuga” 1
DPMsw “gaufguzstu” 2 NPM “þurftugir” 1
DSM “gaufgum” 3
úverðugr DSM “vverdugum” 1
“gofgum” 4 NPM “uuerdugir” 1
NPF “gofgar” 1 vanmáttugr ASM “uanmattugan” 1
NPM “gaufgir” 3 velauðigr ASM “uelaudgan” 1
“gofgir” 16
verðugr
DPM “uerdugum” 1
NPMs “gofgazstir” 4 NPM “verdugir” 1
“gofgazstir” 1 NPMw “verdugu” 2
T = 181, S = 142, S% = 78.45%
Flateyjarbók is unique in the corpus, as there is much more data surrounding its composition than
many of the other texts. The linguistic trends in Flateyjarbók contradict the those of those manuscripts
in some ways. Composed at the end of the fourteenth century, a bit later than Mǫðruvallabók and AM
351 fol., Flateyjarbók was likely compiled at Víðidalstunga, or possibly the Benedictine monastery of
Þingeyrar. (Ashman Rowe, 2005, 11) The manuscript has two scribes, Jón Þórðarson and Magnús
Þórhallsson, with work on it being begun by Jón before completion of the work by Magnús. Data
from Flateyjarbók was culled from the diplomatic edition edited by Guðbrandur Vigfusson and Carl
Rikard Unger (1860-1868).
Flateyjarbók exhibits patterns broadly counter to that of manuscripts of the same age.
Uniquely, syncope is broadly applied in the total number of adjectives, but only over a small subset
of lemmata. Of the thirty-one lemmata, syncope is only attested in twelve of them, and if lemmata
Laidlaw 25
are broken down to their root words (e.g. djúpauðigr as a subtype of auðigr) only six of the eighteen
present show syncope.
Table 24: Lemmata of Flateyjarbók, syncope-exhibiting lemmata in bold
Lemmata ‘Root Lemmata’
allsvǫldugr
alblóðugr
almáttigr
ástúðigr
auðigr*
blóðugr
burðugr
dáðǫflugr
djúpauðigr
fjǫlkunnigr
gǫfugr
jafngǫfugr
kunnigr
liðugr
málugr
máttigr
minnigr
nauðigr
ógǫfugr
ómattugr
sálugr
skammminnigr
stǫðugr
stórauðigr
sverðauðigr
synðugr
þurftugr
úverðugr
vanmáttugr
verðugr
ǫflugr
auðigr*
ástúðigr
blóðugr
burðugr
gǫfugr
kunnigr
liðugr
málugr
máttigr
minnigr
nauðigr
sálugr
stǫðugr
synðugr
verðugr
vǫldugr
þurftugr
ǫflugr
L = 31, S = 12, S% = 38.71% L = 18, S = 6, S% = 33.33%
All occurrences of the adjectives ǫflugr, three of gǫfugr, two of blóðugr, and one each of djúpauðigr,
stórauðigr, nauðigr, and sverðauðigr are in sections of poetry, where the presence of syncope could be
attributed to syllabic retention and constraints in the skaldic form. All instances but one of auðigr-
type adjectives appear syncopated, which might be expected if they were by-and-large in poetic form,
however only three (“diupaudga” - 3ra26, “storaudgazstann” - 3ra45, “suerdaudgan” - 43va29) are
contained in poetry. The majority of them are, however, bynames, inflected in the weak form. In
addition, gǫfugr appears to exhibit the syncope rule more than other adjectives, with comparative
forms such as “gofgra” appearing, despite the lack of a syncope-inducing environment.
Flateyjarbók contains Jómsvíkinga saga, previously attested in AM 291 4° which is about a
century older. Small changes in spelling exist between the two, and not all words present in the earlier
text are attested in the latter one. The superlative forms of auðigr in AM 291 4° are present, with small
changes in the consonants in the ending. The form of malgir appears as “malgir” in both manuscripts.
Laidlaw 26
With respect to the two scribes, their representations of syncope are in agreement. The only
lemma which exhibits both syncopated and unsyncopated form in the target environment is auðigr,
and only in the nominative masculine plural. The only form which fails to show the expected syncope
is “audigir” at 99va46, in the hand of Jón Þórðarson. The relevant sentence at 99va45-46 reads:
“Þeir voru a ungum alldri vinsælir menn ok vel | audigir.”
(‘They were young men, blessed with friends, and rather wealthy.’)
The next attestation of auðgir follows at 101ra11-12, reading:
“Þeir bræðr voru aller menn rikir audgir ok kynstorir foruit|ra ok kærir
konungi”
(‘The brothers were all powerful and wealthy men, of noble birth and wise, and
dear to the king.’)
These two proximal examples show no features giving reason for choosing a syncopated form over
an unsyncopated form, existing in similar syntactic and semantic positions.
GKS 2845 4° – c1425-1475
Table 25: GKS 2845 4° – c1425-1475
Lemma CNG Form Count
auðigr NSMw “audg”[…] 1
blóðugr NPF “blodgar” 2
DPN “blodgum” 1
fjǫlkunnigr NPM “fiolkunigir” 2
“fiolkunnigir” 1
gagnauðigr NSNs [...]“udigast” 1
gǫfugr
NPMs “gafgaztir” 1
ASM “gofgan” 1
NPM “gofger” 1
margmálugr NSMc “margmalgari” 1
nauðigr ASM “naudgan” 1
NPM “naudgir” 1
ókunnigr NPM “okunnigir” 1
T = 15, S = 11, S% = 73.33%
Laidlaw 27
GKS 2845 4°, made of 73 leaves was, likely written in West or Northwest Iceland (Jón Helgason,
1955, ix-x). It was the second of the Fornaldarsǫgur Norðurlanda manuscripts reviewed, and the
following texts within were consulted: Norna-Gests þáttr, Hálfs saga og Hálfsrekka, Gǫngu-Hrólfs saga,
and Herverar saga og Heiðreks. Interestingly, given the late composition of the text, syncope is largely
present, despite its breakdown in earlier manuscripts. Similar to late 14th century manuscripts, kunnigr-
derived adjectives strongly resist syncope, while auðigr, gǫfugr, málugr, and nauðigr retain it. Norna-
Gests þáttr was previously, and contains the forms “gafgaztir” (15r14) and “fiolkungir” (15v19),
corresponding to the forms “gofgazstr” (46rb32) and “fiolkunnigir” (46va4) that were seen in
Flateyjarbók.
AM 343 a 4° – c1450-1475
Table 26: AM 343 a 4°– c1450-1475
Lemma CNG Form Count
auðigr NSMc “audgari” 1
blóðugr NPF “blodgar” 1
gǫfugr
NPM “gaufgir” 1
ASFsw “gaufguztu” 1
APM “gaufga” 1
nauðigr ASF “naudga” 2
T = 7, S = 7, S% = 100%
AM 343a 4° is the third Fornaldarsǫgur manuscript consulted, and is made up of 110 leaves. The
manuscript contains a number of texts, with the following being examined: Ketils saga hængs, Ǫrvar-
Odds saga, and Áns saga bogsveigis (Kålund, 578-579). Vowel syncope is attested in all located forms,
which is somewhat unusual given the age of the manuscript and the decline of syncope in earlier
manuscripts. Attestations of nauðigr contained within occur in in Ǫrvar-Odds saga, which was covered
in Holm perg 7 4°, and the syncopated form could be attributed to scribal conservatism carried over
from an earlier text. That said, the only attested target adjectives are ones that have previously
demonstrated a strong preference for syncope, which could also account for their forms.
Laidlaw 28
AM 152 fol. – 1500-1525
Table 27: AM 152 fol. – 1500-1525
Lemma CNG Form Count
blóðugr NPF “blodgar” 1
fjǫlkunnigr NPM “fiolkunnigir” 2
APM “fiolkunniga” 1
kyndugr NPM “kyndugir” 1
lostugr NPM “lostigir” 1
lýðskyldugr NPM “lydskyldugir” 1
nauðigr NPM “naudigir” 1
ókunnigr NPM “okunnigir” 1
ASM “okunnigan” 1
T = 10, S = 1, S% = 10%
AM 152 fol. is made up of 201 leaves is the final Fornaldarsǫgur manuscript consulted. 200 of the
leaves date to the early sixteenth century, and one leaf dates to the early fourteenth century, however
the single fourteenth century leaf contains homily text, rather than a saga. (Kålund 1888, 105-106)
The manuscript shows minimal syncope, it being only attested in blóðugr. The comparative ending of
gǫfugr surfaced as consonant-initial “gofugra”, rather than the vowel initial form “gaufgara” as in AM
291 4°, or “gꜵfgare” from Holm perg 15 4°. All target lemmata occur beforehand in the dataset and
had previously exhibited, except for kyndugr, which has never exhibited regular syncope. Most non-
syncopated forms have inflectional endings that take the form of brevigraphs.
Oddur Gottskálksson’s New Testament Translation– 1540
Oddur Gottskálksson’s translation of the New Testament has some special distinction as the first
printed Icelandic language book. It is made up of 330 pages, and contained many examples of the
target adjectives, and an almost complete lack of syncope. By the time of Odds NT, u-epenthesis in
the NSM ending was fully realized, and greatly increased the potential for syncope, by creating a new
vowel-initial inflectional ending. From Jón Helgason’s investigation (1929) of the text, only four
examples of syncope are attested, out of a total of 121 target adjectives:
Laidlaw 29
Table 28: Nýja Testamenti Odds Gottskálkssonar – 1540
Lemma CNG Form Count
bakmálugr NPM “bakmalger” 1
máttigr ASM “matkan” 1
ASM “matkann” 1
ǫflugr NPM “oflger” 1
T = 121, S = 4, S% = 3.3%
Of the attested adjectives, máttigr is most common both in syncopated and unsyncopated form in the
text. While it appears syncopated twice, it occurs eighteen more times as potentially syncope triggering
declensions of máttigr, and another ten as declensions of almáttigr. The appearance of syncope is
limited to the ASM ending, however not universal to it, as the form mattugan appears once at Fil.
4:426. In addition, ǫflugr also appears in unsyncopated form eight times, though never with the -er/-
ir NPM ending, with the phonologically most similar being the NSM -ur, which occurs 3 times. There
is only the one attestation of bakmálugr, however, and no direct comparison. The root adjective málugr
also surfaces with a prefix as fjǫlmálugir, showing vowel retention in the NPM form which was missing
from bakmálgir.
Both ǫflugr and málugr have been seen to retain syncope even while it was disappearing from
other adjectives, while máttigr had previously exhibited vowel retention in Flateyjarbók, where málugr
appeared syncopated.
Bishop Guðbrand Þorláksson’s Bible Translation – 1584
Guðbrand Þorláksson’s translation of the entire bible into Icelandic, known as Guðbrandsbiblía, much
like Oddur’s New Testament which came before it showed an overall lack of syncope. There is no
total available of the number of target adjectives within, but it can be assumed to be higher than the
117 present in Oddur’s New Testament translation. In Oskar Bandle’s discussion of Guðbrandsbiblía
(1956), the following forms are the only syncopated ones identified:
Laidlaw 30
Table 29: Guðbrandsbiblía – 1584
Lemma CNG Form Count
bakmálugr NPM “Bakmalger” 1
ǫflugr NPM “Øflger” 1
gǫfugr
NSMw “gỏfge” 1
APM “Gøfga” 1
“Gỏfga” 1
ógǫfugr NPMw “Ogøfgu” 1
máttigr NSMs “Maatkastur” 1
Notably, both “Bakmalger” and “Øflger” are carried over from their appearances in Odd’s NT in
Rom. 1:30 and Heb. 11:34, respectively. Interestingly, both instances of “matkan” in Odd’s NT (Heb.
4:14 and 10:21) do not surface as such in Guðbrandsbiblía, displaying instead the non-syncopated
form, though “Maatkastur”, not present in Odd’s NT, surfaces in syncopated form. Both gǫfugr and
ógǫfugr previously were shown to retain syncope longer than other adjectives.
Overview of Primary Corpus
The primary corpus above give a very broad dataset, covering a large period of time. The data appears
to bear out the assertion that vowel syncope was widely present in the earliest Old Icelandic works,
and was almost entirely absent from the later texts. The dataset is not completely consistent over time,
with fewer exemplar texts towards the end of the period studied, thinning the overall amount of data
as time goes on. Still, the results appear quite clear, and are discussed further in §5.1.
4.2 Njáls saga Corpus
There are many manuscripts of Njals saga, making it an ideal text to track the transmission and form
of -igr and -ugr adjectives. An examination of a normalized edition of the text revealed nine adjectives
of the target type with appropriate inflectional endings. The electronic text was in Modern Icelandic,
but the changes in spelling between normalized Old Norse and Modern Icelandic are predictable and
were accounted for in the search. (Njáls saga, Netútgáfuna)4 The forms and the order in which they
are found follow in Table 30 below.
4 The electronic edition of Njáls saga was linked through Sveinn Yngvi Egilsson’s research project, the
Variance of Njáls Saga
Laidlaw 31
Table 30: Syncopated adjectives in Njáls saga
a) málgir f) nauðgan
b) auðgir g) fjǫlkunnga
c) ómálga h) auðgi
d) málgar i) fjǫlkunngastr
e) auðgir
After locating the adjectives, a list of target manuscripts containing Njáls saga was formed. These
cover a broad range of dates, and various locations within standard stemma of Njáls saga. The
manuscripts range from earliest attestations of the saga, and post-medieval manuscripts copied from
earlier versions.
Table 31: Njáls saga manuscripts
MS Date Nickame Abbreviation
GKS 2870 4° 1290-1310 Gráskinna Gr
AM 162 B β fol. 1300 – β
AM 162 B δ fol. 1300 – δ
AM 468 4° 1300-1324 Reykjabók R
AM 132 fol. 1330-1370 Mǫðruvallabók M
AM 133 fol. 1350 Kálfalækjarbók K
AM 162 B ε fol. 1350-1375 – ε
AM 162 B α fol. 1400-1500 – α
AM 466 4° 1460 Oddabók O
AM 470 4° 1620-1670 – 470
AM 464 4° 1697 – 464
Given the complexity of relationships between different versions of Njáls saga, presented below in
Figure 1 is a simplified stemma of Njáls saga. Relationships between texts are preserved, though texts
not directly preceding ones studied are omitted. This is done with the aim of simplifying for the reader
the general relationship and lineage of any given text. Texts that are examined are underlined. The
stemma is drawn from Einar Ólafur Sveinsson’s work, Studies in the Manuscript Tradition of Njálssaga
(1953, 170-171). For in-depth discussion, Einar Ólafur Sveinsson’s scholarship is to be preferred.
Laidlaw 32
Figure 1: Simplified stemma of Njáls saga
Archetype
┏━━━━━━━━━━┻━━━━━━━━━━━┓
*V *X
┏━┫ ┏━━━┳━━━┳━━┫
*Z
┃
┃
*Y
┃
┃
R
ch.38-90,
141-159
O
ch.143
-159
K
ch.19
-156
*x1
┃
┃
┃ ┣━━┳━━━┓ ┏━━━━┳━━━━━┫
G
*y1
┃
┃
M G
ch.138
-159
*x3
┃
┃
*
┃
┃
*x2
┃
┃
┃ ┏━━┳━━━━┫ ┣━━┓ ┣━━┳━━┓
470
┃
┃
O s
ch.20
-142
K
ch.1-
116
O
ch.1-
19
R
ch.1-38,
19-141
α β ε δ G
ch.138
┗━━━━━┫
464
Following from that, their spelling in the Njáls saga manuscripts is located below in Table 32. Not all
target adjectives are attested in each of the manuscripts, due to some manuscripts being fragmentary,
or simply not containing the attested phrase, somewhat limiting the ability to study their transmission.
Generally, the bulk of the adjectives are attested, though the fragments of AM 162 B, by their nature,
contain at most two of the targets. Cells which contain examples where syncope is not present are
shaded.
Table 32: -igr and -ugr Adjectives in Njáls saga
Normalized GKS 2870 4° AM 162 B β fol. AM 162 B δ fol. AM 468 4°
a) malgir malgir malgir - malgir
b) auðgir audgir - auðgir audgir
c) ómálga ...|ga - - omalga
d) málgar malgar - - malgar
e) auðgir - - - audgir
f) nauðgan naudgan - naudgan naudgan
g) fjǫlkunniga fiolgvnga - - fiolkunniga
h) auðgi - - - audgi
i) fjǫkunnigastr fiolkyngaztr - - fiolkunnigaztr
Laidlaw 33
Normalized AM 132 fol. AM 133 fol. AM 162 B ε fol. AM 162 B α fol.
a) malgir - *margir - malger
b) auðgir - audgir - -
c) ómálga - o malga - -
d) málgar malgar malgar - -
e) auðgir augir audgir - -
f) nauðgan naudgan naudgan - -
g) fjǫlkunniga fiolkunniga fiolkunga fiolkuniga -
h) auðgi audgi audgi - -
i) fjǫkunnigastr fiolkunnigaztr fiolkunnigaztr - -
Normalized AM 466 4° AM 470 4° AM 464 4°
a) malgir malgir mäluger mälúgir
b) auðgir audgir auduger auðigir
c) ómálga omalga ǫmälga Ǫmalga
d) málgar malgar mälgvr mälugir /
mälgar
e) auðgir audgir audgir auðgir
f) nauðgan - naudgan nꜷðgann
g) fjǫlkunniga fiolkuniga fiolkunnga fiólkúnga
h) auðgi audgi audge auðgi
i) fjǫkunnigastr *fiolkunigr fiǫlkunnugastr fiólkúnnigaztr
As can be expected with texts from the turn of the fourteenth century, the earliest attestations of the
saga display a robust syncope rule. As was seen in other manuscripts from the early 14th century,
kunnigr-derived adjectives are the first to cease undergoing syncope. And, similar to other texts,
syncope fails to take off in other lemmata, with the exceptions of some attestations of málugr, and
auðigr. Again, the set of adjectives auðigr, málugr, nauðigr, and ómálugr consistently display syncope
in the target environments. In general, the syncope seems to continue long after the rule has been
abandoned in other texts, though inconsistently, but this may again be attributed to scribal
conservatism. Though the scribe would have applied his own mental grammar and presumably
avoided terms which were completely ungrammatical sounding, these syncopated terms must have
been at least recognizable, though likely archaic sounding.
Laidlaw 34
Njáls saga Corpus Manuscripts
The manuscripts used to investigate Njáls saga cover approximately four centuries of
transmission of the text, and have varying numbers of attestations of the target adjectives. Njáls saga
has a rich history of transmission, and thesources and lineage of particular versions of the saga in
addition to discussing the manuscripts in which they are contained are touched on in brief.
4.2.1.1 GKS 2870 4° – 1290-1310 – Gráskinna
Gráskinna is the earliest of the Njáls saga manuscripts consulted, it is made up of 121 leaves,
and the only text within is a copy of Njáls saga (Kålund 1900, 55-56). It contains all the adjectives
found in the normalized text except (e) auðgir, and (h) auðgi. Example (c) ómálga was divided across
a line, and missing all letters prior to the terminal -ga. Gráskinna does contain minor addirions to it
dating into the sixteenth century, however none of the examples are found within them. Gráskinna
does not have any descendant texts in this investigation, though the chapters 138 through 159, which
contain fjǫlkunngastr at the end of chapter 155, have a somewhat different placement in the stemma.
4.2.1.2 AM 162 B β fol. – c1300
AM 162 B fol. contains a number of fragments of Njáls saga, of which fragment β is the first
examined. Fragment β, which consists of a single leaf, textually most closely related to fragment α
despite a difference of at least a century, contains only a single example, fully syncopated. (Kålund
1888, 118)
4.2.1.3 AM 162 b fol. δ – c1300
Fragment δ, made up of 24 leaves, is of a similar date to β, and bears its descent from a slightly
different version of the text, being closer related to fragment ε and chapter 138 of Gráskinna, GKS
2870 4°. The attestations of syncope and spellings of them between Gráskinna and fragment δ are
identical. (Kålund 1888, 118-119)
Laidlaw 35
4.2.1.4 AM 468 4° – 1300-1324 – Reykjabók
Reykjabók, from the early fourteenth century, is made up of 93 leaves, and was written in
Iceland. (Jón Helgason 1962, xi) It contains examples of all adjectives identified, and illustrates the
breakdown of syncope in kunnigr-derived adjectives.
4.2.1.5 AM 132 fol. – 1330-1370 – Mǫðruvallabók
Mǫðruvallabók, previously discussed in §4.1.12 retains the syncope rule only in kunnigr-type
adjectives, as in Reykjabók. The examples present were previously included and discussed in the
overall account of Mǫðruvallabók. Mǫðruvallabók’s version of Njáls saga is most closely related to
Gráskinna’s version. Due to scribal error, auðgir surfaces as “augir”, completely lacking the medial
dental “ð”.
4.2.1.6 AM 133 fol. – c1350 – Kálfalækjarbók
Kálfalækjarbók, a mid-fourteenth century manuscript contains ninety-five leaves, and is
contains only a copy of Njáls saga. (Kålund 1888, 97-98) All of the target adjectives are present,
except for the initial málgir, which is replaced with “margir” in the text, likely due to the scribe
misreading “l” as “r” in the exemplar text. AM 133 fol.’s text of Njáls saga is a sister version to
chapters 38-90 and 141-159 of Reykjabók and chapters 143-159 of AM 466 4° (discussed in §4.2.1.9).
Interestingly, of the two attestations of fjǫlkunnigr, one exhibits syncope, while the other does not.
The syncopated form “fi lkunga” occurs in chapter 104, and is not related to the attestation in sister
texts. The unsyncopated form “fiolkunnigaztr” occurs in chapter 155 and its similarity to the form in
Reykjabók “fiolkunnigztr”.
4.2.1.7 AM 162 B ε, fol. – 1350-1375
Fragment ε of AM 162 B, fol. is made up of 8 leaves. (Kålund 1888, 119) It contains only a
single attestation of fjǫlkunnigr, realized in syncopated form as “fiolkuniga”. In the stemma, fragment
ε is most closely related to fragment δ, however the text of the two do not overlap.
Laidlaw 36
4.2.1.8 AM 162 B α fol. – 1400-1500
Fragment α is made up of only 2 leaves, is sister to fragment β, and cousin to the other
fragments. (Kålund 1888, 117) The only attested adjective within is “malger”, which is in-line with
the other most closely related texts, although orthographically it is distinct by not using a brevigraph
for the inflectional ending.
4.2.1.9 AM 466 4° – c1460 – Oddabók
Oddabók, written in Iceland in the mid fifteenth century, is made up of 57 leaves, and contains
only a defective copy of Njáls saga. (Kålund 1888, 652) Chapters 1-19 most closely related to
Kálfalækjarbók and Reykjabók, and contain the first three taget adjectives. The spellings, as well as
their orthographic depicition, are more in line with those of Reykjabók. Chapters 20-142 of Oddabók,
containging examples 4 through 8 (except 7, nauðgan, which is unattested), are only distant cousins
to Kálfalækjarbók and Reykjabók, and are more closely related to Mǫðruvallabók. Spellings between
Oddabók and Mǫðruvallabók are consistent, excepting the form of auðgir in Mǫðruvallabók which is
missing the dental due to scribal error. The final chapters of Njáls saga in Oddabók (143 onwards) are,
in contrast to the rest of the saga, located in a different part of the stemma, though still related closely
to similar sections of Kálfalækjarbók and Reykjabók. The final target adjective of Njáls saga, kunnigastr,
is not attested in the superlative form, surfacing in the positive as “fiolkunigr” instead.
4.2.1.10 AM 470 4° – 1620-1670
AM 470 4° is a paper manuscript from the mid-seventeenth century in the Gullskinna
tradition5, made up of 160 pages. (Kålund 1888, 654) While outside of the scope of medieval
manuscripts per se, it nonetheless is of use to examine for what it can tell about scribal conservatism
and the overall state of disyllabic adjectives. It is not precisely clear where it lies in the stemma (Einar
Ólafur Sveinsson 1953, 171), though there is some evidence that AM 464 4° is a descendant text. All
target adjectives are attested, and forms which earlier exhibited syncope in some cases do not. The
attestation of málgir and the first attestation of auðgir surface as “mäluger” and “auduger”, respectively,
5 Texts derived from those copied from Gullskinna, a now lost and widely copied manuscript. For further
discussion, see ‘A New Stemma of Njáls saga’ by Ludger Zeevaert, Alaric Hall, et. al.
Laidlaw 37
the latter of which is one of the only instances of auðigr surfacing in an unsyncopated form in the
research corpus. The second attestation of auðgir surfaces as “audgir”, showing syncope in a word-
form that was hitherto unsyncopated in the text. It is worth noting that “audgir” appears as a marginal
insertion to the main body of the text on 56v, and may be syncopated for the sake of space saving, or
possibly it was copied from a version of the text in which it was originally syncopated. Similarly,
“naudgan” and “fiolkunnga” are both marginal, and may have been syncopated for the same reasons.
4.2.1.11 AM 464 4° – 1697
AM 646 4° is the youngest text examined, is another post medieval paper manuscript, made
up of 162 pages. (Kålund 1888, 651) It is a copy of Kálfalækjarbók, though orthographic representation
of vowels and attestation of syncope is not preserved completely between the two works. Portions of
the text, were copied over from AM 137 fol., and a paper manuscript, presumed to be AM 470 4°.
The first attestation of málgir, which appears in Kálfalækjarbók as “margir”, is corrected back to
“mälúgir”, likely due to the scribe having access to another exemplar text containing a form of málgir.
The second attestation surfaces as “mälugar”, unsyncopated in comparison to “malgar” as it appears
in Kálfalækjarbók. Shown below in Figure 2 is a detail of the attestation of “mälugar”. Additionally to
the unsyncopated form there is a marginal note referencing it as a lexical variant6 of “mälgar”, and
that the scribe followed the original manuscript to the best of his ability.7 This suggests some
philological interest in the matter and that “malgar” (as the scribe diligently copied) was probably
seen as an archaic variant.
Figure 2: Detail of AM 464 4°, 37r24
6 Surrounding “mälugar” are brackets, referenced on 154v as denoting “variantes lectiones” 7 MB is explained on 154v as “membranam quam quan potui diligentia secutus sum”
Laidlaw 38
Overview of Njáls saga Corpus research
The data coming out of the Njáls saga Corpus presents a somewhat different picture than that of the
Primary Corpus, highlighting the power of scribal conservatism and the retention of syncopated forms
despite their decline elsewhere. The fact that scribes retained syncopated forms from the exemplar
texts suggests that they at least understood them to be not completely ungrammatical, and likely
understood them to be archaic, but reasonable, variants.
4.3 Laxdæla saga Corpus
Like Njáls saga, Laxdæla saga is preserved across a large number of manuscripts, broadly covering the
Classical Old Norse period. There are a number of fragments dating back to the thirteenth century,
though the oldest manuscript to preserve the saga in full is Mǫðruvallabók. The methodology to find
adjectives in Laxdæla saga is identical to that used for Njáls saga in §4.2. The forms and the order in
which they are found follow below in Table 33.
Table 33: Syncopated Adjectives in Laxdæla saga
a) auðgi i) alblóðgan q) gǫfgum
b) gǫfga j) málga r) gǫfgasta
c) auðgastan k) gǫfga s) helgi
d) auðgi l) helgi t) helgi
e) gagnauðgar m) malga u) helga
f) gagnauðgar n) malga v) gǫfgast
g) auðgari o) málgu w) gǫfgustu
h) gǫfgastr p) málgu
After locating the adjectives, a list of target manuscripts containing Laxdæla saga was formed in the
same manner as was used for Njáls saga. There are fewer full copies of the saga extant, and even fewer
in manuscripts which have been analyzed or made electronically searchable. The manuscripts
consulted follow below in Table 34.
Table 34: Laxdæla saga manuscripts
MS Date Nickname Abbreviation
AM 132 fol. 1330-1370 Mǫðruvallabók M
AM 309 4° 1498 Bæjarbók í Flóa B
Laidlaw 39
AM 158 fol. 1630-1675 158
ÍB 225 4° 1686-1687 225
ÍB 226 4° 1680-1699 226
Though there are not as many manuscripts consulted as were for Njáls saga, their relationships are of
some import. Based on Einar Ólafur Sveinsson’s description of the Laxdæla saga manuscripts in the
Íslenzk Fornrit edition (1934, lxxvi-lxxxii), a simplified stemma follows in Figure 3.
Figure 3: A Simplified Stemma of Laxdæla saga
Archetype
┏━┻━━┓
Y Z
┏━━┫ ┣━┓
M D1 * *
┃ ┃ ┃
V*/225 B *
┏━┫
226 158
It is worth expanding on two entries in the stemma which are not consulted. D1 is AM 162 D1 fol.,
and is a fragmentary version of only five leaves. While not consulted for this analysis, it is likely to be
the mother text for the now lost Vatnshyrna manuscript, which was burnt in the Copenhagen Fire of
1728, of which ÍB 225 4° is Árni Magnusson’s paper copy. While ÍB 225 4° dates to the late
seventeenth century, Vatnshyrna was likely composed toward the end of the fourteenth century in
Northern Iceland by Magnús Þórhallsson (McKinnell 1993, 689-690). While ÍB 225 4° must, of
course, be considered as a seventeenth century text, its linguistic forms may be thought of as late
fourteenth century.
Laidlaw 40
Table 35: -igr and -ugr Adjectives in Laxdæla saga
Standardized AM 132 fol. AM 309 4° AM 158 fol. ÍB 225 4° ÍB 226 4°
a) auðgi “ꜹðgi” “audgi” “audgi” “ꜹþgi”
b) gǫfga “gavfga” *“virduliga” “gaufga” “gǫfuga”
c) auðgastan “audgaztan” “audgastan” “audgazstan” “ꜹþugastan”
d) auðgi “avdgi” “audgı” “audgi” “ꜹþgi”
e) gagnauðgar “gagnauðgar” *“gagnsamar” “gagnaudgar” *“gagnsamar”
f) gagnauðgar “gagnauðgar” “gagnaudgar” “gagnaudgar” *“ꜹþugar”,
“gagnsamar”
g) auðgari “auðgari” “audigri” “audug” “ꜹþug”
h) gǫfgastr “gaufgaztr” “gǫfugastr” “gaufazstr” “gǫfugastr”
i) alblóðgan “albloðgan” “alblodúgann” “alblodugann” “alblǫþugan”
j) málga “malga” “malga” “maalga” “mälga”
k) gǫfga “gaufga” “gofga” “gaufga” “gǫfuga”
l) helgi “helgi” “helgi” “helgi”
m) málga “malga” “mälga” “maalga” “mälga”
n) málga “malga” “m:” “maalga” “mälga”
o) málgu “malgu” “malgu” “m lgu” “mälgo”
p) málgu “malgu” “malgu” “maalgu” “mälgo”
q) gǫfgum “gaufgum” “gaufgum”
r) gǫfgasta “gaufgazta” “gofgasta” “gøfugasta” “gaufgazsta” “gǫfugasta”
s) helgi “helgi” “helgi” “helgi” “helgi” “helgi”
t) helgi “heilagi” “h·” “helgi” “helgi” “helgi”
u) helga “helga” “helga” “h.” “helga” “helga”
v) gǫfgust “gaufguzt” “gofguz” “gøfgust” “gaufgazst” “gǫfugust”
w) gǫfgustu “gaufguztu”
Early copies of Laxdæla saga kept syncope in most forms, however these broke down somewhat in
later copies. Overall, however, the use of syncopated forms continued after the syncope rule could be
seen to break down. As with the transmission of adjectives in Njals saga, this may be due to the
tendency of scribes to be orthographically conservative.
Laxdæla saga Corpus Manuscripts
As with the discussion of transmission in Njáls saga, the manuscripts consulted, including size and
origin, where known, are briefly examined. There are not as many manuscripts consulted for Laxdaæla
saga as there were for Njáls saga, however the dataset is somewhat larger, containing nineteen target
Laidlaw 41
disyllabic adjectives, and four instances of heilagr. The target adjectives cover five lemmata, two being
forms derived from the root auðigr.
4.3.1.1 AM 132 fol. – 1330-1370 – Mǫðruvallabók
Mǫðruvallabók has been discussed at length already in §4.1.12 and adjectives within it follow the same
pattern as seen with Njáls saga in §4.2.1.5. The attestations within Laxdæla saga were all included in
the initial discussion of the manuscript. In all target adjectives, syncope is applied, though it is worth
noting that one instance of heilagr fails to syncopate, surfacing as “heilagi”, rather than the
monopthongized form “helgi”. Given the eventual emergence of competing heilagr paradigms, this
could be an early example of such a development.
4.3.1.2 AM 309 4° – 1498 – Bæjarbók í Flóa
AM 309 4° from the end of the fifteenth century only contains excerpts from Flateyjarbók, notably
Ólafs saga Tryggvasonar, in addition to portions of Eyrbyggja saga, and Njáls saga, and the later part
of Laxdæla saga, over eight leaves. (Kålund 1888, 544-546) Believed to be written by Jón Oddsson,
most of the target adjectives are missing from this portion of the text, with only two target attestations
of gǫfugr present. Both instances of gǫfugr appear in the superlative and exhibit syncope, which given
that portions of the overall manuscript were copied over from Flateyjarbók, which exhibited a strong
preference towards syncope in declensions of gǫfugr, is not entirely surprising.8
4.3.1.3 AM 158 fol. – 1630-1675
Believed to have been written by Þorsteinn Bjǫrnsson at Útskálar, AM 158 fol. is a large paper
manuscript of 140 pages, and was once part of a larger work, along with AM 121 fol., AM 181 a-h,
k-l fol., and AM 204 fol. (Kålund 1888, 111-112) The dataset from AM 158 fol. differs significantly
from that of Mǫðruvallabók, in that two of the target adjectives are supplanted by altogether different
ones. One instance of gǫfga (‘noble’) is replaced with virðuliga (‘worthy, fine,’) and gagnauðgar (‘very
productive’) is replaced with gagnsamar (‘hospitable.’) As would be expected, neither of the
8 A further discussion of the comparison between Flateyjarbók and Bæjarbók í Flóa may be found in Bára
Yngvadóttir’s BA thesis “Forrit og eftirrit : handritin AM 162 E fol. og AM 309 4to borin saman”
Laidlaw 42
replacement adjectives exhibit syncope, as virðuligr, being -ligr terminal, would never have
syncopated, and the root adjective of gagnsamr, samr, is monosyllabic and also would not have
exhibited syncope.
4.3.1.4 ÍB 225 4° – 1686-1687 – Copy of Vatnshyrna
ÍB 225 4° is a 172-page post-medieval paper copy of Laxdæla saga, copied by Ásgeir Jónsson and Árni
Magnusson, from the lost Vatnshyrna text, which is part of the Y-Group of Laxdæla saga, given the
stemma of Einar Ólafur Sveinsson. (1934, lxxvi-lxxxii) All target adjectives, except for the attestation
of alblodugaṅ, exhibit syncope, which is in line with the forms in Mǫðruvallabók, the closest related
Y-group text. Mǫðruvallabók does contain the unsyncopated form “heilagi”, which is not carried
forward into ÍB 225 4°, instead surfacing as “helgi”, and alblóðgan surfaces in ÍB 225 4° unsyncopated
as “alblodugaṅ”. While the manuscript itself dates from the end of the seventeenth century, the
original text, which we may assume to be faithfully copied, is likely from the end of the fourteenth.
One might expect a higher incidence of syncope in the text, given the overall amount of it in the
corpus at the time of the composition of Vatnshyrna. The overall lack of syncope is likely indicative
of the saga being faithfully copied from an earlier version of the text, likely AM 162 D1 fol., though
the bulk of that manuscript is now lost.
4.3.1.5 ÍB 226 4° – 1680-1699
ÍB 226 4°, the last Laxdæla saga manuscript consulted, is another post-medieval paper copy, made of
137 pages. Believed to be copied by the priest Jón Halldorsson in Hítardalur, it is one of the Z-group
texts, and is more closely related to AM 158 fol., and AM 309 4°, based on the scholarship of Einar
Ólafur Sveinsson. (1934, lxxvi-lxxxii) It contains the most vowel retention of all the manuscripts,
with all of the retention from AM 158 fol. present, as well as the replacement of gagnauðgar with
gagnsamar. Virduliga does not show up, having been returned to gǫfga, however one later instance of
gagnauðgar was replaced with “ꜹþugar” and “gagnsamar”. It seems likely, therefore, that ÍB 226 4°
was copied from a text which used AM 158 fol. as a source in addition to another, assuming that it is
a faithful copy, rather than a synthesis of two differing texts.
Laidlaw 43
Overview of Laxdæla saga Corpus Research
Like the data coming from the examination of Njáls saga, the Laxdæla saga data shows a tendency
towards the retention of syncopated forms late into the period of syncope decline. This again suggests
that scribes considered the syncopated forms to be lexical variants which were at least understandable
to the readers and, presumably, to the speakers of the language.
5. Analysis of Available Data
5.1 Occurrence of Syncope in the Dataset Over Time
Broadly speaking, it can be observed that syncope of -igr and -ugr adjectives had fallen out of use for
new texts effectively by the end of the sixteenth century. Works such as Odds Nýja Testamenti and
Guðbrandsbiblía, new translations of non-Icelandic texts, were shown to not use syncope where
otherwise it would be expected, with few exceptions. Given the body of evidence, some closer
examination of the trends is in order.
Overall Percentage of Syncopated vs Unsyncopated Attestations
The first, perhaps most obvious choice of analytical approach is to chart how much syncope appeared
in each of the manuscripts consulted in the primary corpus. This approach avoids the manuscripts
selected for transmission studies of Njáls saga and Laxdæla saga, primarily due to the heavy prevelance
of scribal conservatism preserving syncopated forms long after said forms were no longer attested in
novel works. The results of charting vowel syncope in the manuscripts in the corpus follow below in
Figure 4, presented with a two-point moving average9.
9 A two point moving average is a statistical tool which smooths out short term fluctuations and highlights
long term trends in a dataset.
Laidlaw 44
Figure 4: Presence of Syncope in the Manuscript Corpus
The data shows, perhaps as expected, a general decline in syncope with AM 519 4°, and continuing
along, somewhat unevenly to Oddur Gottskálksson’s New Testament and Guðbrandur Þorláksson’s
Bible translation. AM 343a 4°, a collection of fornaldarsögur and riddarasögur stands out in this
analysis with 100% of the attested adjectives displaying syncope, unlike the surrounding manuscripts.
The texts contained within, while no doubt copies from earlier manuscripts, are not particularly old,
and perhaps indicative of a high degree of scribal faithfulness to their exemplar texts. While the 100%
figure is high, it is spread across only seven examples and four lemmata, and not a large data set in
and of itself. In addition to AM 343a 4° appearing quite high, AM 351 fol., Skálholtsbók eldri, appears
quite a bit lower than the surrounding texts. It was composed at a time of high Norwegian influence
in Iceland, and is related through that to the Norwegian landslǫg, and as such may contain Old
Norwegian influence on the application of syncope. This is outside the scope of this investigation but
nonetheless of some interest.
Guðbrand’s bible translation was given an estimated value of 1 percent, as the total occurrence
of the target adjectives is unknown. This estimate is arrived at by taking the only two syncopated
forms that appear both in Oddur’s New Testament and Guðbrand’s Bible translations, and assuming
that the overall prevelance of syncope is the similar across both the Old and New Testaments. This
number is then rounded down slightly, influenced by the extremely limited attestation of syncope,
and the overall size of the Old Testament compared to the New. It seems likely that a figure as high
as 1% is, in fact, an overestimation, though without a searchable electronic copy this remains inexact.
100% 100% 100% 100% 100% 100% 100% 100%
80%
100%
71%
94%
22%
80%73%
100%
10%3% 1%
0%10%20%30%40%50%60%70%80%90%
100%
Laidlaw 45
While this approach is useful for showing the trend within the corpus, it lacks precision around
timing, as some of the manuscripts could be ascribed to a rather broad range of compositional dates.
Accounting for the, the manuscripts were sorted into eras based on possible composition date and the
average value of syncope was calculated again. For example, GKS 1009 fol., Morkinskinna, dated from
1250 to 1300 was sorted into the categories 1200-1250, 1250-1300, and 1300-1350. After each
manuscript is assigned to its applicable eras, the average amount of syncope present in each era was
calculated, by averaging the manuscripts and texts present. The manuscripts and the eras into which
they are sorted is shown below in Table 36. This approach serves to account for uncertainty around
the composition of the manuscript, as well as to smooth out the data set somewhat. In addition, by
extending the date as early as possible, this method aims to take into account the time at which the
scribe would have learned the language and word forms. Admittedly, it does extend the range of
manuscripts that are not very precisely dated, giving them more weight than those which are more
accurately known. Nevertheless, ascribing them to as broad a range of dates as possible will better
show trends within the corpus. The averaged amounts of syncope exhibited in the manuscripts for
each era is presented further below in Figure 5.
Table 36: Corpus Manuscripts in Dating System
Manuscript
1150
–
1200
1200
–
1250
1250
–
1300
1300
–
1350
1350
–
1400
1400
–
1450
1450
–
1500
1500
–
1550
1550
–
1600
AM 674a 4° X X
AM 673a II 4° X X
AM 673b 4° X X
Holm perg 15 4° X X
AM 645 4° X
GKS 1157 fol. X X
AM 291 4° X X
GKS 1009 fol. X X
AM 519a 4° X X
DG 11 X X
Holm perg 7 4° X X
AM 132 fol. X X
AM 351 fol. X X X
GKS 1005 fol. X
Laidlaw 46
Manuscript
1150
–
1200
1200
–
1250
1250
–
1300
1300
–
1350
1350
–
1400
1400
–
1450
1450
–
1500
1500
–
1550
1550
–
1600
GKS 2845 4° X
AM 343a 4° X X
AM 152 fol. X X
Oddur’s New Testament X
Guðbrand’s Bible X
Figure 5: Averaged Presence of Syncope in Corpus
Much more clearly than in Figure 4, Figure 5 illustrates the decline of syncope beginning at the end
of the thirteenth century. While the trend certainly begins then, it fails to gain real traction until the
beginning of the sixteenth century, with texts such as AM 152 fol., and Odds Nýja Testamenti, before
finally losing syncope almost altogether by the end of the century with Guðbrandsbiblía.
This depiction of the data available does appear somewhat suspect, though. A spontaneous
decline, even if somewhat foreshadowed by earlier declines in syncope seems unlikely, given then
usual slow nature of sound change. Within this examination, the precipitous decline in the sixteenth
century may be somewhat attributed to the presence of AM 343a 4° in the dataset, where its somewhat
out of place 100% syncope occurrence skews the data from 1400-1450 and 1450-1500 upwards
significantly. If AM 343a 4° is omitted from the dataset, the trend is made a little clearer, as seen in
Figure 6, below.
100% 100% 92%
73.55%66% 65%
55%
7%1%
0%10%20%30%40%50%60%70%80%90%
100%
Laidlaw 47
Figure 6: Averaged Presence of Syncope in Corpus, excluding AM 343a 4°
Without AM 343a 4°, the average curves out in a somewhat smoother manner, and this might be a
justifiable approach to the data. If AM 343a 4° is assumed to primarily contain texts copied from
earlier manuscripts, and if they are presumed to exhibit a high degree of scribal conservatism, the
manuscript may be excluded due to a lack of novel attestations of the sagas contained within. As was
seen with both the Njáls saga and Laxdæla saga manuscripts, scribal conservatism is a powerful force,
and lead to longstanding retention of forms.
While this approach to interpretation of the data may be tempting, it has some significant
pitfalls. First, and most obviously, “problematic” data ought not to be excluded for the sake of making
a point. While the omission cleans the curve from 1400 to 1550, there is still a significant and rapid
decrease in syncope, albeit between 1400-1450 and 1450-1500, fifty years earlier than in the previous
calculation. In addition, the dataset is relatively sparse towards the end of the period examined, and
the removal of AM 343a 4° reduces the periods of 1400-1450 and 1450-1500 to two and one
examples, respectively. Furthermore, other manuscripts that contained examples from the
Fornaldarsögur Norðurlanda would, presumably, be subject to the notions of copying and scribal
conservatism, yet their inclusion in the analysis still points toward a decline in vowel syncope. While
the trendline present in Figure 6 without AM 343a 4° included in the dataset is interesting, and shows
the decline of syncope along a smoother curve, it is ancillary to that of Figure 5 and only included for
the sake of argument.
100% 100% 92%
73.55%66%
48%
10% 7%1%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Laidlaw 48
Overall Percentage of Lemmata Demonstrating Syncope
Another approach to analyzing the trend of syncope decay is to track the number of lemmata
exhibiting syncope over time. To do this the total number of lemmata in each text studied in the
main corpus is taken, and divided into that is the number of lemmata that exhibit syncope in at least
one attestation. In this way, the preference towards high-frequency words in the count of overall
attestations levelled, and a somewhat different picture emerges.
Figure 7: Percentage of Total Lemmata Displaying Syncope
As expected, starting with AM 519 4°, the trend is towards vowel retention. As with figure 3, numbers
for Guðbrandsbiblía are estimated based on overall size of the text, the limited examples of syncope
contained within, and generously rounded up. Using the search feature of the text database of
Árnastofnun (corpus.arnastofnun.is), twenty-five target lemmata10 were pulled from the body of the
Icelandic translation of the Bible. Working on the assumption that these (at least) would be present
in Guðbrandsbiblía, the number of lemmata that were shown to exhibit syncope in §4.1.19 was divided
into this total. These are all taken from a list of the most common -igr and -ugr disyllabic adjectives
(§5.2.1), and their forms created through prefixation. In the same manner and same justification as
10 almáttigr, ánauðigr, ástúðigr, auðigr, bakmálugr, blóðugr, fámálugr, fjǫlkunnigr, gǫfugr, kunnigr, liðugr,
málugr, máttigr, minnigr, nauðigr, ǫflugr, ógǫfugr, ókunnigr, ómáttugr, óverðugr, skuldugr, skyldugr,
stǫðugr, stórauðigr, verðugr
100%100%100%100%100%100%100%100%
66%
100%
60%
91%
25%
39% 38%
100%
14% 18% 20%
0%10%20%30%40%50%60%70%80%90%
100%
Laidlaw 49
in Figure 5, the data can be smoothed out to rough period of composition for clearer analysis, depicted
in Figure 8.
Figure 8: Average of Total Lemmata Displaying Syncope
Somewhat different than the findings in §5.1.1, Figure 8 shows syncope breaking down at a somewhat
slower rate, with a significant holdover into the sixteenth century. The figures in the sixteenth century
are likely higher than their actual number, but will need much closer examination to bring them down
to their likely totals.
Syncopated Attestations and Syncopated Lemmata
While independently interesting and generally similar, the results of analysis of total number of
attestations of syncope and total number of syncopating lemmata are somewhat different. Below in
Figure 9 the two results are overlaid on each other.
Figure 9: Comparison of Overall Syncope Attestations and Syncopated Lemmata
100% 100% 88%68%
52% 54% 57%
19% 20%
0%
20%
40%
60%
80%
100%
0%
20%
40%
60%
80%
100%
Overall Syncope Occurrence Total Lemmata Syncopated
2 per. Mov. Avg. (Overall Syncope Occurrence) 2 per. Mov. Avg. (Total Lemmata Syncopated)
Laidlaw 50
It is of some interest that while the trends in overall amount of syncope and number of lemmata
exhibiting syncope begin similarly, they diverge in the latter half of the fifteenth century. This is
partly due to the small number of texts in that era, as well as the effect of AM 343a 4°, which displayed
syncope over many lemmata, skewing the total upwards. Directly comparing the overall occurrence
of syncope and total number of lemmata syncopated shows quite clearly the trend in Old Norse
towards vowel retention. Beginning with the late thirteenth century, and continuing steadily towards
the end of the sixteenth, vowel syncope fell out of use, before effectively disappearing altogether.
5.2 Frequency and Syncope in -igr and -ugr Adjectives
One fact that has arisen from the data presented is that syncope and retention was unevenly applied
throughout the -igr and -ugr adjectives, with some losing syncope earlier than others, and some never
displaying syncope in the first place within the available data. The uneven nature of the application
of syncope is curious for a number of reasons, both lexical and phonological. A brief lemma-specific
analysis follows, focusing on the most frequent -igr and -ugr terminal adjectives, followed by an
examination of the phonological environments created by and avoided by vowel syncope and vowel
retention. Some of the adjectives also exhibited alternation in the dataset between -igr and -ugr, and
the difference in vowel was likely due primarily to differing interpretations by scribes and speakers of
the vowel, and conditioned somewhat by the surrounding phonological environment.
Methodology
All potential adjectives were taken from the research corpus, and sorted by root lemma. To come to
the ranking, three separate sources were consulted and averaged. First, the lemmatized corpus of the
Íslendingasǫgur available at snara.is were searched. Some adjectives, such as skuldugr, were not attested
in the corpus of the Íslendingasǫgur, but were not eliminated from the search at this point. Given the
large number of texts searched in this manner, it gives a broad picture of frequency. The database
only searches in Modern Icelandic normalized editions of the texts, but all spellings were corrected
back to Old Norse for the dataset. Prefixed versions of the adjectives were searched using prefixes
attested in ONP.
Laidlaw 51
The second step was using the same adjective list and searching the electronic database of the
Íslenskt textasafn of Árnastofnun, available at corpus.arnastofnun.is. The same set of adjectives were
put through the search function and their frequency recorded. Adjectives missing from the database
at this point were excluded, as insufficient data could be collected. The selection of texts searched is
somewhat broader, featuring Íslendingasögur, Fornaldarsögur Norðurlanda, Heimskringla, Snorra Edda,
Heilagramanna sögur, Íslendingabók, Landnámabók, as well as collections of kvæði and þættir.11 The
text database, while not lemmatized, uses a form generator, wherein one can input an adjective and
get declensions of it. As a text base, items may possibly be missed due to misspellings or adjectives
not appearing in the editions used in the searchable text. For example, auðgastr, which appears in
Njáls saga, does not show up in the Íslenskt textasafn search at all. If the goal were direct comparison
of the data sources, this would be troubling, but overall frequency results from this database were
similar to those of Snara.
The third step was to check the total attestations of each adjective available in ONP. The
electronic wordlist of ONP has both advantages and disadvantages over the other databases examined.
The chief advantage it has is a high degree of cross-reference, where prefixes to target adjectives are
linked through the entries in the word list. This makes calculating a total number of attestations
including prefixed versions of the root quite straightforward. The chief disadvantage to the ONP
wordlist is the possibility for self-selection of word forms, where, if a scholar has investigated a
particular lemma or prefix, it may be disproportionately represented, compared to other similar ones.
Still, the results of ONP were similar to those of the other two databases.
With results from all sources, each adjective’s order in the ranking from most to least common
was averaged, as was their ranking in total number of attestations between the sources. Finally, those
two results were averaged to weight the results slightly towards total number of attestations found.
When this final average was arrived at, any adjectives with identical values were given the same ordinal
11 http://corpus.arnastofnun.is/leit.pl?info=2 contains the full list of specific texts searched.
Laidlaw 52
rank.12 While imperfect, this method at least provides a relatively accurate ballpark-figure for
frequency in Old Icelandic Texts.13
Most Frequent -igr and -ugr Adjectives
The most frequent -igr and -ugr adjectives in the presented corpus and Old Icelandic in general are,
broadly speaking, presented here. Some of them are not present in the corpora studied, namely
mektugr and siðugr, and hǫfugr appears only in the affixed form jafnhǫfugr in DG 11. The adjectives,
ranked from most to least frequent follow in Table 37.
Table 37: The 20 Most Frequent -igr and -ugr Adjectives
1 kunnigr 5 nauðigr 9 liðugr 13 ástúðigr 17 mektugr
2 auðigr 6 blóðugr 9 stǫðugr 14 minnigr 18 siðugr
3 gǫfugr 7 málugr 11 skyldugr 15 hǫfugr 19 burðugr
4 máttigr 8 verðugr 12 saurugr 16 ǫflugr 19 dreyrugr
This is not, by any means, a definitive list, but a useful guideline. Adjectives proximal to each other
in ranking might be transposed, as they occur with broadly speaking similar frequency, but relative
ranking of adjectives distant from each other (i.e. kunnigr vs siðugr) can be safely assumed to be of
very different frequency.
Syncope and Frequency in Old Icelandic Adjectives
In the examination of the corpus, it was found that certain adjectives tended to retain syncope, even
after its decline in other forms, specifically auðigr, blóðugr, gǫfugr, málugr, nauðigr, and ǫflugr, and
these are all, apart from ǫflugr, among the eight most common target adjectives. Other adjectives that
potentially could have been syncopated failed to do so at any point in the dataset or general corpus.
The rank, whether or not syncope was ever attested, approximate date of first attestation, and
approximate date where syncope first failed to surface is shown below in Table 38.
12 There does exist a word frequency list in the form of Íslensk orðtíðnibók (1991), although it covers Modern
rather than Old Icelandic, and this was eschewed in order to obtain a solely medieval word frequency. 13 Full tables of figures are available in Appendix A
Laidlaw 53
Table 38: Overview of Syncope in the 20 Most Frequent -igr and -ugr Adjectives
Rank Lemma Sync.
Attested
First
Attestation
First
Vowel
Ret.
Rank Lemma Sync.
Attested
First
Attestation
First
Vowel
Ret.
1 kunnigr Y c1200 c1270 11 skyldugr N c1280
2 auðigr Y c1175 c1350 12 saurugr Y c1200 c1300
3 gǫfugr Y c1175 c1600 13 ástúðigr Y c1200 c1600
4 máttigr Y c1150 c1270 14 minnigr Y c1200 c1325
5 nauðigr Y c1150 c1350 15 hǫfugr Y c1192
6 blóðugr Y c1250 c1350 16 ǫflugr Y c1200 c1400
7 málugr Y c1200 c1620 17 mektugr N c1350
8 verðugr N c1300
18 siðugr N c1250
9 stǫðugr N c1280
19 burðugr N c1300
9 liðugr N c1290
19 dreyrugr Y c1200 c1280
Date of first attestation, first vowel retention, and whether syncope was attested was taken from ONP,
and cross referenced with the research corpus, with dates from the research corpus taking precedence
over dates from ONP. The same limits to the possible accuracy to information from ONP from §5.2.1
apply, making definitive dating of these appearances somewhat difficult.
While there is no clear correlation between frequency and syncope, a rough connection can
be established. The seven most frequent of the adjectives all display syncope regularly, suggesting
that Seven of the twenty most common adjectives never displayed syncope, though they are unevenly
distributed throughout the data. The average first date of attestation of the adjectives that did not
display syncope was approximately 1283, whereas the average date syncope-exhibiting adjectives
attestation was 1200. While not suggesting that the vowel-retained adjectives were novel words at the
end of the thirteenth century, it is in the period that syncope was observed to be in decline in the
dataset overall, and their attestations may both be influenced by, and contribute towards, the trend at
that time toward vowel retention. It is quite likely that the later attestation date for adjectives that do
not exhibit syncope is another indication of their relative rarity within the corpus, and by extension,
their relative low ranking amongst the most frequent adjectives in general. Given how infrequently
they are exposed to the appropriate environments, it is not particularly surprising that they failed to
adopt the syncope rule. This, of course, cannot be a complete picture of syncope in the adjectives, as
Laidlaw 54
some numerically less frequent adjectives did display syncope at least some of the time. The adjectives
which failed to display syncope do share some phonological features, discussed in §5.3.
Bynames contributing to Retention of Syncopated Forms
One factor that has some bearing on the retention of vowel syncope in common adjectives is the
formation of bynames in Old Norse. Bynames often were given based on characteristics personal to
the named, such as their physical or mental characteristics, and such names could be handed down
from one generation to another. In this manner, these often-adjectival formations became somewhat
nominal, which by virtue of the slow-changing nature of proper names, may have tended to retain
older linguistic features. Bynames, generally, took the form of a proper name followed by the definite
article and an adjective, for example Ólafr inn helgi (Olaf the holy) or Guðríðr in auðga (Gudrid the
wealthy). Sometimes the byname appears without the definite article, with forms such as Ólafr helgi,
though the byname adjective still surfaces in a weak form as though it had been preceded by the
definite article. When disyllabic-stemmed adjectives are used in bynames syncope is universally
expected, as all weak adjectival endings are vowel-initial and would be expected to trigger it.
Only a few of the most common target adjectives are used as bynames, with examples found
only of auðigr, kunnigr, nauðigr, and ǫflugr. Of these, only auðigr and ǫflugr are used in their un-
affixed form (though fjǫlauðigr is attested as well), while kunnigr and nauðigr surface as fjǫlkunnigr and
ánauðigr. Heilagr also appears very frequently as a byname (Lind 1921). Heilagr and auðigr were by far
the most common of the disyllabic adjectives that appeared as bynames, with the others appearing
only a few times thoughout the corpus. This nominalization of the syncopated form auðgi may have
delayed syncope decline in the lemma, as it was a not infrequent byname. The case of heilagr is
somewhat more complicated due to the stem allomorphy between heilag- and helg-, and the existence
of the proper names Helgi and Helga. That said, given the fact that it is only auðigr and heilagr that
appear frequently as bynames, it seems a strech to assume the nominalization had a large effect on the
presence or absense of syncope in the paradigm overall. At most, the influence of bynames was
restricted to auðigr and heilagr.
Laidlaw 55
On heilagr
Phonologically similar to the -igr and -ugr adjectives is heilagr (‘holy’), briefly explained in §2.2.
Outside of the scope of this investigation, the behaviour of heilagr nevertheless bears some relation
to the behaviour of the target adjective forms. Given the stem allomorphy between heilag- and helg-,
and the eventual emergence of the competing paradigms heilagur and helgur, heilagr shows the
influences of contrasting forces on -Vgr terminal adjectives. Heilagr was excluded from the ranking
devised for ranking the adjectives, though when it was run through, it was by far the most common
of these adjectives.14
Heilagr was regularly syncopated in the target environments, with forms such as helgi and
helga showing up widely throughout the corpus. Syncope produced a predictable monophthongization
in the stem, alternating between heilag- and helg-. That said, heilagr began to show signs of inconstant
application of syncope, with forms like heilagar, heilaga, heilagare, and heilagu15 appearing in the
thirteenth century. With the increased prevalence of u-epenthesis in the NSM form toward the end
of the thirteenth century (Haraldur Bernharðsson 2014, 190), this led to competing paradigms, one
featuring syncope and monophthongization, and one featuring vowel retention. The monosyllabic
stem was further strengthened by the existence of the proper names Helgi and Helga, clearly related
to the same root and created from the weak form of the adjective. On the other hand, the
disappearance of syncope in phonologically similar -igr and -ugr adjectives could have strengthened
the disyllabic stem. Once the NSM form displayed a vowel-initial ending without exhibiting syncope,
the other vowel-initial endings could be formed without triggering stem syncope.
5.3 Phonological Environment
The unifying feature of the target adjectives is their similar phonological environment. Each adjective
has a -Cig or -Cug terminal stem, each stem is disyllabic, and, with few exceptions, the final consonant
is non-liquid, unless the final vowel is -u-. Given the irregular application of syncope throughout the
target adjectives, some closer examination of the consonant clusters created and avoided may be in
14 See appendix A 15 Contained in texts AM 645 4°, Holm perg 17 4°, ibid, and AM 232 fol., respectively. Holm perg 17 4° and
AM 232 fol. are not part of the dataset, but information regarding forms is taken from ONP.
Laidlaw 56
order. This will focus primarily on the twenty most common, though examples from outside this
range may be consulted.
Consonant Clusters Created and Avoided
In the twenty most common target adjectives, the consonant clusters created or avoided through
syncope or retention and their preceding vowel(s) follow below in
Table 39. Due to the phonological similarities, the monophthongized stem of heilagr, helgr, is
included. Consonant clusters that would be created in adjectives that never undergo syncope are in
boldface.
Table 39: Consonant Clusters in Target Adjectives
Lemma Form w/Syncope
& NSMw Ending
Consonant Cluster
w/Leading Vowel Lemma
Form w/Syncope
& NSMw Ending
Consonant Cluster
w/Leading Vowel
heilagr helga elg skyldugr *skyldgi yldg
kunnigr kunngi unng saurugr saurgi aurg
auðigr auðgi auðg ástúðigr ástúðgi úðg
gǫfugr gǫfgi ǫfg minnigr minngi inng
máttigr máttgi áttg hǫfugr hǫfgi ǫfg
nauðigr nauðgi auðg ǫflugr ǫflgi ǫflg
blóðugr bóðgi óðg mektugr *mektgi ektg
málugr málgi álg siðugr *siðgi iðg
verðugr *verðgi erðg burðugr *burðgi urðg
stǫðugr *stǫðgi ǫðg dreyrugr dreyrgi eyrg
liðugr *liðgi iðg
Overall, the initial vowels of the created or avoided clusters have similar features, and overlap in
distribution, though “a”, “o”, and “œy” only occur in syncope-exhibiting clusters. Consonants appear
to have similar distribution, however all of the clusters avoided among the most common target
adjectives feature a dental consonant, either a stop or fricative, before the final velar stop. While
dentals also occur in the same position in syncope-creted clusters, their preceding environment is
slightly different. Syncope is present when the dental is preceded by a long vowel or diphthong, or a
short vowel followed by a non-liquid consonant, and vowel retention is exhibited when the dental is
voiced, and is preceded by a short vowel, or a liquid consonant in turn preceded by a short vowel.
Laidlaw 57
Syncope Present : auðigr auðig-i > auðgi – VVD{i,u}g- > VVDg-
blóðugr blóðug-i > blóðgi – V:D{i,u}g- > V:Dg-
lostugr lostug-i > lostgi – VC[-liq]{i,u}g- > VC[-liq]g-
Syncope Absent : liðugr liðug-i > liðugi – VD{i,u}g-
verðugr verðug-i > verðugi – VC[+liq]D{i,u}g-
þurftugr þurftug-i > þurftugi – VC[+liq]CD{i,u}g-
Not all target adjectives fall into this pattern, however, as sálugr and mektugr, follow the rules of the
syncopated adjectives, never exhibit syncope. Likewise, neither does vanmáttigr, or skammminnigr,
despite the prevalence of syncopated forms of their root adjectives. Perhaps timing is of some
importance here, as mektugr, sálugr, vanmáttigr, and skammminnigr, are all first attested (in ONP) in
the early to middle fourteenth century, once vowel syncope was observed to be on the decline. While
one might expect forms such as vanmáttigr and skammminnigr to behave in the same manner as their
roots, it is curious that both fail to display syncope while their roots do. If these were novel forms to
the time made through creative affixation, or were being recorded for the first time in literary form,
their attestation in the newer, unsyncopated form may be expected.
The aversion to a Vowel-Liquid-Dental-Velar-Vowel cluster seems readily explicable by
means of simple aversion to complex consonant clusters. A combination of sounds as would be
produced by a form such as *verðgi seems highly difficult to pronounce, and even more complex ones
like Vowel-Liquid-Labial-Alveolar-Velar-Vowel such as *þurftgi would be next to impossible given
Icelandic phonology. That said, complex consonant clusters have been produced through syncope in
other configurations, most notably in forms like ǫflgi, which produces a similarly complex Vowel-
Labial-Liquid-Velar-Vowel cluster, with no apparent problem in pronunciation. Given the normal
patterns of Icelandic syllable structure, there is not a clear reason for one cluster to be avoided, and
another created, with the exception of the aversion to that of *þurftgi.
Syllabic Patterns Created ǫfl.gi, auð.gi, gǫf.gi,
blóð.gi
VCC.CV, VC.CV,
CVC.CV, CCVC.CV
Syllabic Patterns
Avoided
*verð.gi, *lið.gi,
*þurft.gi
CVCC.CV, CVC.CV,
CVCCC.CV
Laidlaw 58
When the medial vowel is retained, however, the following syllabic structures emerge:
Syllabic Patterns Created
via Retention
au.ði.gi, ver.ðu.gi,
þurf.tu.gi
VCC.CV, VC.CV, CVC.CV,
CCVC.CV
Syllabic Patterns Created
via U-Epenthesis
au.ði.gur, ver.ðu.gur,
þurf.tu.gur
V.CV.CVC, CVC.CV.CVC,
CVCC.CV.CVC
The syllabic patterns created and avoided appear similar, without clear preference for one type
of syllable pattern. The only form which appears to be entirely avoided is the cluster of four consonants
that would be present in forms like *þurftgi. There is a slight tendency towards open syllables, and a
preference to avoid complex codas, which are both fairly common linguistic preferences.
Given the similarity between the phonetic and syllabic environments, it does not appear that
syncope and vowel retention were particularly governed by the phonology of the adjectives. While
obviously following a general phonological principle, it is clear that vowel syncope was never a
particularly strong rule, and the contrast between the phonetic environments that it surfaced in and
the ones that it avoided was too small to maintain the distinction. The phonetic difference between -
igr and -ugr adjectives and the syncope-retaining disyllabic adjectives like gamall is considerably
larger, and partially accounts for how it would disappear in the former, while remaining in the latter,
as do the phonological rules surrounding u-epenthesis.
5.4 Factors Contributing to Vowel Retention
u-Epenthesis
As established earlier, vowel retention in -igr and -ugr disyllabic adjectives occurred around the same
time as when an epenthetic -u- was added to the NSM ending. This changed the syllabic structure at
the end of the “default” inflectional form, the nominative singular masculine, and may have had
analogical effects on the rest of the paradigm.
The NSM ending -r was clearly in the earliest stages of the language non-syllabic, most clearly
demonstrated through its appearance in the skaldic poetic form dróttkvætt. Dróttkvætt requires six
syllables per line, and instances of NSM endings are plentiful, a notable example from Þórsdrápa, a
10th century poem, reads:
Laidlaw 59
drjúgr var Loptr at ljúga Þórsdrápa 1:3
driugr var loptr at liu.ga
Which, given the usual division of syllables, renders ljúga as the only two-syllable word, based on the
constraint that all lines of dróttkvætt contain six syllables. While this clearly was not a problem for
speakers of the language at that time, this had changed by the end of the thirteenth ceuntry, when the
epnethetic -u- began to slowly appear. Scribes had a tendency to write sounds as they heard them,
which suggests that by the end of the thirteenth century the vocal expression of the NSM ending had
either become syllabic, thus necessitating some sort of vowel representation, and follow from that had
acquired a genuine vowel at the onset of the ending (Ari Páll Kristinsonn 1992, 15-33). The
underlying form of the ending was likely still /-r/ at this point, otherwise one would expect alveolar
assimilation to no longer occur in adjectives such as gamall, and given the continuation of syncope in
vowel-initial declensions of such disyllabic adjectives, one might expect to see syncope in the NSM
form.
/gamal/ + /-r/ > gamall
/gamal/ + /-ur/ > *gǫmlur
The lack of these forms underscores the process of syllabification of the terminal /-r/ (as well as
medial /r/ in words such as fegrð) which contributed to redefining the shape of the end of disyllabic
adjectives.
The effects of the syllabification of the ending differ between prose and poetry. Poetry, already
very linguistically conservative, retained many archaic features of spelling and inflection, and the
appearance of an epenthetic vowel in everyday speech did not deter the retention of an non-
epenthesized ending. Prose, on the other hand, given its stronger connection to everyday speech,
clearly prefered to reflect the realities of speech as closely as possible, which has direct impact on
adjectives which would have exhibited syncope in environments with vowel-initial syllabic endings.
Adjectives such as auðigr, were realized no longer as disyllabic in the common NSM form,
but trisyllabic, dramatically changing the syllabic rhythm. U-epenthesis made the syncope rule
opaque, creating an environment which would conceivably trigger vowel syncope, yet failed to do so.
Laidlaw 60
Below, the contrast in syllabification between NSM and ASM forms is presented, first featuring
syncope without u-epenthesis, and then including u-epenthesis but without syncope.
NSM au.ðigr > au.ði.gur
vs ASM auð.gan > au.ði.gan
The change in root form could easily have been generalized across the paradigm, with other inflected
forms being influenced by the lack of syncope in the NSM. The benefits to a speaker or child learning
the language at time of adopting a trisyllabic and non syncopated form can be summed up in a few
points. First, particularly troublesome consonant clusters are avoided; while there is overlap in the
structures of clusters created and avoided, the trend towards simpler consonant clusters is clear.
Generally, the limits on consonant clusters allow no more than three together, though when there are
three consonants in sequence, they are split between the coda of one syllable and the onset of the
next. Second, there is paradigmatic levelling, making generation of forms somewhat more
straightforward. Instead of grappling with stem allomorphy between a disyllabic and monosyllabic
stem, the paradigm is simplified and generalized as a disyllabic stem. Finally, any conflict with the
productive /-ligr/ morpheme is eliminated. /-ligr/ as a productive adjective making morpheme fits
the same phonological rules for syncopation, yet never did. By avoiding syncope in this environment,
the common -igr and -ugr endings may have taken on a near morphological status, reinforcing their
form surfacing with a retained vowel. This simplifies the morphophonology for new speakers of the
language, by not having to distinguish between the -ligr and -igr/-ugr endings for the purposes of
syncopation.
Analogical Forces in -igr and -ugr Adjectives
Since u-epenthesis may have had some analogical effects on the paradigm, a more general look at
analogical forces at play in -igr and -ugr adjectives prior to u-epenthesis may be in order. As previously
established, vowel syncope was not uniform across -igr and -ugr adjectives, with some variation based
on overall frequency and presence of a dental prior to the final vowel. There are also the adjectives
formed by the very phonologically similar adjective-making -ligr morpheme, which never exhibited
vowel syncope. Given the minor phonological difference between adjectives that display and avoid
Laidlaw 61
vowel syncope and the resistance to syncope in -ligr terminal adjectives, it is conceivable that syncope
was avoided in order to bring the syncopated paradigm in line to other paradigms which did not
exhibit syncope. This would also help distinguish disyllabic adjectives which continued to display
syncope, such as mikill and gamall, from the -igr and -ugr adjectives which, while disyllabic, were
inconsistent regarding the application of syncope. Analogical levelling would also bring disyllabic
stemmed adjectives in line with similar monosyllabic stemmed adjectives, for example digr, which
sometimes became disyllabic through prefixation and predictably did not exhibit syncope in the stem.
With a sizable number of examples of similar adjectives which did not syncopate, it seems likely that
analogical force had some bearing on the disappearance of syncope from the paradigm. Furthermore,
analogical levelling has been already observed in the phonetically similar heilagr, where the
monophthongized stem helg- became generalized and extended into its full paradigm. This illustrates
the ability for similar disyllabic adjectives to undergo the same sort of process.
6. Conclusions and Areas for Further Study
The first research question posed was to discover when syncope broke down in the target adjectives.
From an overview of the dataset formed, it seems clear that syncope began to disappear toward the
end of the thirteenth century, and was all but extinct in the language by the end of the sixteenth
century. The relatively slow decline of the phenomenon may be partly explained by scribal
conservatism, as some more frequently adjectives tended to retain syncope after it had fallen out of
use in less frequent adjectives. Frequency indeed may have played a role, as speakers acquiring the
language would have had fewer opportunities to apply the syncope rule to the less frequent adjectives,
making them more prone to paradigmatic levelling, and never picking up the syncope rule. The
decline of vowel syncope appears to happen around the same time as the emergence of u-epenthesis
in the language, though the relationship between the two is not completely clear.
While there is no proverbial smoking gun clearly explaining the disappearance of vowel
syncope from disyllabic -igr and -ugr adjectives, there are numerous clues which point towards the
reasons for its disappearance. As stated previously, vowel syncope declined around the same time as
the rise of u-epenthesis, which, given that they both affect the syllable structure at the end of words,
Laidlaw 62
is not surprising. U-epenthesis, however, does not entirely account for the disappearance, most
notably because of the appearance of -igr and -ugr adjectives which never exhibited syncope at all.
There are some phonological differences between the -igr/-ugr adjectives which display and
avoid syncope, particularly a slight tendency in adjectives with a dental prior to the final vowel to
avoid syncope. The differences appear to be quite minor, however, and illustrate there being no clear
phonological distinction between the two paradigms. The pattern of disyllabic adjectives containing a
dental prior to the final vowel avoiding syncope is, however, a small but novel discovery in the field.
Analogical levelling is likely to be another contributing force to the disappearance of syncope.
Seeing that there are multiple phonologically similar adjectives which never undergo syncope, and
that syncope is strongly associated with different disyllabic adjectives (gamall type) it makes sense that
the -igr and -ugr adjectives would tend to favour vowel retention. This is supported by the lack of
syncope in less frequently or later attested -igr and -ugr adjectives, suggesting that the older,
syncopated ones, were slowly brought in line with the unsyncopated types.
There are numerous avenues for further research into the topic of vowel syncope which could
yet yield interesting results. There is opportunity for a deeper examination of texts from the late
fourteenth century onward, as not as many of them were included in the data set. Particular attention
ought to be paid to manuscripts that feature initial attestations of particular sagas, as they may be
assumed to be least constrained by scribal conservatism. In addition, an in-depth phonological analysis
of disyllabic adjectives that never exhibited syncope may be in order, due to the higher concentration
of dental consonants present in them.
The reason behind disappearance of syncope appears to be twofold; the rise of u-epenthesis
influencing word-terminal phonology, and the force of analogical levelling resolving the inconsistent
application of syncope across phonologically similar words. The study of the phenomenon is
incomplete, but I believe that I have provided a broad picture of the state of syncope over a significant
portion of the Classical Old Icelandic period, from which more detailed examinations with more
precise conclusions can be drawn.
Laidlaw 63
Appendix A – The Most Common -igr and -ugr Adjectives
Total Attestations Comparative Ranking
Snara SÁM ONP SUM ONP SÁM Snara Average
ástúðigr 14 7 42 63 16 15 8 13
auðigr 151 37 245 433 3 6 1 3.33
blóðugr 34 62 82 178 14 3 5 7.33
burðugr 3 3 26 32 21 19 11 17
dreyrugr 2 8 11 21 22 10 14 15.33
gǫfugr 102 122 132 356 8 2 3 4.33
he(i)l(a)gr 52 596 515 1163 1 1 4 2
hǫfugr 1 8 35 44 19 10 16 15
kunnigr 150 41 341 532 2 5 2 3
liðugr 2 3 140 145 6 19 14 13
lostigr 0 3 35 38 19 19 21 19.67
málugr 26 11 138 175 7 9 6 7.33
máttigr 8 57 178 243 4 4 9 5.67
mektugr 0 3 84 87 13 19 21 17.67
minnigr 1 2 86 89 12 23 16 17
móðugr 1 5 11 17 22 18 16 18.67
nauðigr 24 16 161 201 5 8 7 6.67
ǫflugr 3 6 39 48 18 17 11 15.33
saurugr 5 8 59 72 15 10 10 11.67
siðugr 0 7 42 49 16 15 21 17.33
skyldugr 1 8 90 99 11 10 16 12.33
stǫðugr 1 8 110 119 10 10 16 12
verðugr 3 20 120 143 9 7 11 9
Laidlaw 64
Appendix A – continued
Total
Attestation
Rank
Comparative
Ranking
Average
Total/Comparative
Average
ástúðigr 16 13 14.5
auðigr 3 3 3
blóðugr 7 7 7
burðugr 21 18 19.5
dreyrugr 22 16 19
gǫfugr 4 4 4
he(i)l(a)gr 1 1 1
hǫfugr 19 15 17
kunnigr 2 2 2
liðugr 9 14 11.5
lostigr 20 23 21.5
málugr 8 8 8
máttigr 5 5 5
mektugr 14 21 17.5
minnigr 13 19 16
móðugr 23 22 22.5
nauðigr 6 6 6
ǫflugr 18 17 17.5
saurugr 15 10 12.5
siðugr 17 20 18.5
skyldugr 12 12 12
stǫðugr 11 11 11
verðugr 10 9 9.5
Laidlaw 65
Bibliography
Ari Páll Kristinsson. 1992. “U-innskot í íslensku.” Íslenskt mál 14: 15‒33.
Bandle, Oskar. 1956. Die Sprache der Guðbrandsbiblía. Bibliotheca Arnamagnæana 17. Kopenhagen.
Beck, Heinreich. 1993. Wortschatz der altisländischen Grágás (Konungsbók). Göttingen: Vandenhoeck
& Ruprecht.
Einar Ól. Sveinsson. 1934. Formáli. Laxdæla saga. Íslenzk Fornrit 5. Reykjavík: Hið íslenzka
fornritafélag.
Einar Ól. Sveinsson. 1953. Studies in the Manuscript Tradition of Njálssaga. Studia Islandica 13.
Reykjavík: H.F. Leiftur.
Fix, Hans. 1984. Wortschatz der Jónsbók. Frankfurt am Main: Lang.
Gordon, E.V. 1957. An Introduction to Old Norse. Second Edition. Oxford: Oxford University Press.
Grape, Anders, Gottfrid Kallstenius, Olof Thorell. 1962-77. Snorre Sturlasson's Edda: Uppsala-
Handskriften DG 11, II. Uppsala: Almqvist & Wiksell.
Guðbrandur Vigfusson, Unger, Carl Rikard eds. 1860-68. Flateyjarbok : En samling af norske konge-
sagaer. Christiania (Oslo): Det Mallingske Boktrykkeri.
Haraldur Bernharðsson. 2014. Icelandic. A historical linguistic companion. Reykjavík. [Manuscript]
Jón Helgason. 1929. Málið á Nýja Testamenti Odds Gottskálkssonar. Safn Fræðafélagsins um Ísland og
Íslendinga 7. Kaupmannahöfn: Hið íslenzka fræðaféla.
Jón Helgason. 1955. The Saga manuscript 2845, 4° in the Old Royal Collection in the Royal Library of
Copenhagen. Manuscripta Islandica 2. Copenhagen: Ejnar Munksgaard.
Jón Helgason. 1962. Njáls saga : The Arna-Magnæan Manuscript 468, 4to (Reykjabók). Manuscripta
Islandica 6. Copenhagen: Ejnar Munksgaard.
Kålund, Kristian. 1888-1894. Katalog over Den Arnamagnæanske Håndskriftsamling. Copenhagen:
Gyldendal.
Kålund, Kristian. 1900. Katalog over de oldnorsk-islandske Håndskrifter i Det store kongelige Bibliotek
og i Universitetsbiblioteket. Copenhagen: Gyldendal.
Kjeldsen, Alex Speed. 2013. Filologiske studier i kongesagahåndskriftet Morkinskinna. Bibliotheca
Arnamagnæana. Supplementum 8. Copenhagen: Museum Tusculanum.
Laidlaw 66
Larsson, Ludvig. 1891. Ordförrådet i de älsta islänska handskrifterna. Lund : Lindstedt.
Larsson, Ludvig. 1956. Glossar till Codex AM 291, 4: to (Jómsvíkinga saga): utg. av Sture Hast. Lund:
Gleerup.
de Leeuw van Weenen, Andrea. 2000. A Grammar of Mǫðruvallabók. Leiden: CNWS.
de Leeuw van Weenen, Andrea. 2009. Alexanders Saga: AM 519a 4° in the Arnamagnæan Collection,
Copenhagen. Copenhagen: Museum Tusculanum.
de Leeuw van Weenen, Andrea. 2004. Lemmatized Index to the Icelandic Homily Book. Reykjavík:
Stófnun Árna Magnússonar.
de Leeuw van Weenen, Andrea. 1987. Mǫðruvallabók, AM 132 fol. Leiden: Brill.
de Leeuw van Weenen, Andrea. 1993. The Icelandic Homily Book: Perg. 15o in the Royal Library,
Stockholm. Reykjavík: Stofnun Árna Magnússonar
Lind, Erik Henrik. 1921. Norsk-isländska personbinamn från medeltiden. Uppsala: Lundequistska.
McKinnell, John. 1993. “Vatnshyrna.” In Medieval Scandinavia: An Encyclopedia, Edited by Phillip
Pulsiano, 689-690. New York: Garland.
“Netútgafan” [Electronic Saga Texts] Netútgafan. http://www.snerpa.is/net/index.html
Noreen, Adolf. 1923. Altisländische und Altnorwegische Grammatik. Fourth Edition. Halle: Niemeyer.
“Ordbog over det norrøne prosasprog – Dictionary of Old Norse Prose.” [Web Database] Københavns
Universitet. http://onp.ku.dk
Peterson, Paul. 2012. Old Norse Nicknames. Unpublished master’s thesis, University of Iceland.
Available at http://skemman.is/
“Snara” [Web Database] Snara ehf. http://snara.is/
“Textasafn SÁM.” [Web Database] Stofnun Árna Magnússonar í íslenskum fræðum.
http://corpus.arnastofnun.is/
Vilhjálmur Finsen. 1852. Grágás: Islændernes lovbog i fristatens tid. Copenhagen: Brødrene Berlings.