1 Summary
This book is divided into five main parts. Part I (pp. 1–53), consisting of three chapters, is entitled ‘The historical and theoretical setting’.Footnote * Chapter 1 is a brief outline of the phonology of the common ancestor of Icelandic and Faroese. Chapter 2 goes more into detail about the phonological changes that occurred between this common ancestor and Icelandic/Faroese. Chapter 3 discusses some theoretical preliminaries for the author's treatment of Icelandic and Faroese phonology, with particular emphasis on the theoretical description and explanation of the widespread diphthongisation processes in these languages. In other words, the theoretical framework for the book as a whole is not outlined first, as one might have expected. A potential problem here is that the first two chapters of the book make several references to concepts from moraic theories, Optimality Theory, Lexical Phonology, Government Phonology and Dependency Phonology, and it could be difficult for some readers to fully appreciate the points that the author is trying to make in these two chapters when the theoretical framework has not yet been given. It is worth pointing out in this connection that most of these theoretical concepts are not explained in Chapter 3 either.
Part II (pp. 55–126) has four chapters, and is entitled ‘The modern sound systems’. Chapter 4 is a relatively short chapter on the Icelandic vowels and diphthongs, highlighting aspects such as derived diphthongs and the relationship between short and long vowels. Chapter 5 is considerably longer, and deals with the Faroese vowels and diphthongs, its length in part driven by the author's decision to include data from many dialects. Chapter 6 deals with the Icelandic consonants, and Chapter 7 with the Faroese consonants. A somewhat unusual aspect of the organisation of this book is that the segments of the languages are not fully presented until 120 pages into the book.
Part III (pp. 127–211), consisting of three chapters, is entitled ‘Systemic relations and syllabic structure’. Chapter 8 raises some theoretical questions concerning the behaviour of vowels and diphthongs with respect to the kind of syllables in which they occur in both Icelandic and Faroese, and is in a sense a recapitulation of Chapters 4 and 5. Chapter 9 deals with the distribution of consonants, with an emphasis on the fact that the set of consonants that can be in the onset of the syllable is not the same as the set of consonants in the syllable coda. Chapter 10 is a more in-depth discussion of the relationship between vowel duration and syllabic structure, in an attempt, as the author puts it, ‘to interpret the relation between the template structure of syllables or suprasegmental constituents and the … properties of the vowels as segments or units in the system’ (p. 198). One tendency to be noted in this third part is how the focus shifts more strongly to the Icelandic language. In Chapter 10, for instance, the first 23 pages are devoted to Icelandic, with no more than two pages on Faroese.
Part IV (pp. 213–267) has only two chapters, and is entitled ‘Segments and syllables on phonological levels’. Chapter 11 deals with pre- and postaspiration in the two languages; the aspiration facts are accounted for within a syllable-based account. Chapter 12 discusses whether the phonological patterns one can detect in various paradigms are still ‘phonologically active’, or whether they have been lexicalised/morphologised. As noted above in connection with the third part of the book, the main focus is on Icelandic.
The fifth and final part (pp. 269–327), with the title ‘Rhythmic structure’, has three chapters. Chapter 13 treats the placement of stress in Icelandic and Faroese words, while Chapter 14 discusses stress and sandhi phenomena at the phrase level. Chapter 15 deals with the intonational properties of phrases in these languages, as well as their semantic and pragmatic interpretations. It is noteworthy that Part V contains no clear discussion of the intonational or pitch properties of lexical stress. Since a whole chapter is devoted to the placement of lexical stress, it would have been beneficial to learn more about its intonational correlates. The book ends with a bibliography and an index of keywords.
2 General assessment
Generally speaking, this book is about the higher-order structure of Icelandic and Faroese, and it deals most extensively with the syllabic structure of these languages. A whole range of phonological phenomena that exist on the phonemic and subphonemic levels are as a result analysed as consequences of the syllabic structure assumed. The book will therefore be of value to readers who are interested in the connection between segmental phonology and syllables, particularly if they already have some knowledge of Icelandic or Faroese.
Árnason's theoretical framework for such syllabic structures is rather unorthodox. A word like austr ‘east’ is syllabified as /au.s.t.r/ (p. 20), and the consonants in this word are referred to variously as ‘extrasyllabic consonants’ (p. 71), ‘extrasyllabic onsets’ (p. 72), ‘extrametrical or forming an onset to an empty final nucleus’ (p. 81), ‘extensions to the Appendix’ (p. 148) and ‘dull syllables’ (p. 201). The main problem with this is not that the syllabification is somewhat unusual, but rather that the syllabification procedure is never properly explained.
Phonological analyses of segmental properties ‘make use of privative features or melodic elements of the type used in Dependency Phonology’ (p. 41). Although this latter approach is unconventional, it is of little consequence in this book, as the framework in this context only provides an alternative picture of what the core elements of the phonological segments are. As an example, the palatal stop /cʰ/ is analysed as {_, I, ʔ, h, H} ({@, I, ʔ, h, H}) (p. 100).
It should be mentioned that it is not common in today's phonological research to rely so heavily on the notion of syllables and syllable boundaries in order to account for phonotactics and segmental phonology. The main argument provided for the reality of syllables and syllable boundaries is that vowels are predictably lengthened at the boundary of a stressed syllable (pp. 146–149). There is an apparent circularity in this argument, however, since the location of such syllable boundaries in turn is determined by whether such vowel lengthening occurs or not (p. 146). Given how heavily the analyses in this book rest on assumptions of syllabic structure and its central role in Icelandic and Faroese phonology, it would have been helpful if more arguments supporting this approach had been provided.
As a first in the Oxford series The Phonology of the World's Languages, this book treats the phonology of two languages at the same time. The author acknowledges the unusual approach taken here, and admits that ‘there is no such thing as the phonology of two languages’ (p. xiii). One question left unaddressed is why this approach was chosen in the first place. There can be little doubt, though, that this book is strongly skewed in favour of Icelandic, especially in the second half of the book. The phonological phenomena of Icelandic are always treated first, followed by an assessment of the equivalent phenomena in Faroese phonology, and to what extent Faroese exhibits the same properties as Icelandic or differs from it. One could say, then, that this book to some degree is a contrastive grammar of these two languages, seen from the perspective of Icelandic.
The main impression I am left with after having read this book, though, is determined by two of its major characteristics: its discursive style makes the book a hard read, and it contains a substantial number of factual errors and typos. These and other critical remarks are treated in the following section.
3 Critical remarks
3.1 Style
The discursive style includes the use of many subordinate clauses and parenthetical comments. It is not unusual to see sentences with as many as four parenthetical comments (for example on pages 33 and 195). A typical example of the heavy use of (preposed) subordinate clauses is the sentence: ‘Because of its history and as reflected in the spelling, resulting from palatalization and cluster simplification, the palato-alveolar /ʃ/ does not …’ (p. 176). The complexity of the sentences in this book often requires the reader to re-read many passages, and even then many sentences may remain enigmatic. An example is given below.
From the point of view of the framework applied here, the main question is whether we should assume distinctions in the input for the weak syllables, motivated by morphological oppositions, which then may be neutralized in some environments, or whether some other principles govern the distribution of the variants or some input restructuring has occurred which can account for the variation. (p. 93)
Linguistic data in this book is quite commonly presented and interspersed in running text. This strategy adds to the length and complexity of the sentences, and it can also make it hard to locate and retrieve the data on a later occasion, especially as the book does not have a word index.
The analyses, terminology and examples often change from one part to the next, sometimes even within a paragraph or a single sentence. An example of changes to the analyses can be found on page 15, where it is claimed that a final postconsonantal /r/ in Old Norse (the conventional name for the common ancestor of Icelandic and Faroese) was syllabic (i.e. /-C
/), whereas on the very last page of the book (p. 327) it is said that this /r/ was not syllabic. An example of terminological inconsistency is the use within two pages (pp. 4–5) of five different names for what appears to be the same (the parent language of Icelandic and Faroese): ‘Common West Nordic’, ‘Old Icelandic’, ‘Old Norse’, ‘Old West Nordic’ and ‘Proto-West Nordic’. Inconsistencies in the linguistic examples are especially frequent, but I will limit this to three illustrative examples here.
(i) On page 4 the author aims to illustrate the Icelandic construction with possessive pronouns (‘my hand’) as opposed to one using prepositions (‘the hand on me’), and writes hönd mín vs. hendin á mér. But this example also changes the definiteness of the noun (hönd is indefinite, hendin is definite) and the stem variant of the noun (hönd vs. hendi), both of which have nothing to do with the phenomenon this section aims to illustrate. A better example would be simply höndin mín vs. höndin á mér.
(ii) In the discussion of Icelandic and Faroese vowels in Chapter 5, Árnason considers some derived nominal forms: the definite form huginn [hy;ːjin] ‘mind.sg.def’ in Icelandic (p. 65) and the definite dative form seyðinum [seiːjinʊn] ‘sheep.dat.sg.def’ in Faroese (p. 82). Later on the same pages, however, these examples change, without comment, to the citation forms hugi [hyːji] ‘mind.sg.indef’ and seyður [seiːjʊɹ] ‘sheep.nom.sg.indef’. These new forms are neither glossed nor explained.
(iii) Example (9.32) on page 161, illustrating initial fricatives in Icelandic, has hjörð [çœrð] ‘herd’, but a few lines later we find: ‘we note that … the historical cluster /hj/ has supplied a voiceless palatal fricative [ç] in hjarta’, as if the example had hjarta rather than hjörð (the transcription [çœrð] in the example is erroneous; it should be [çœrθ]).
Another typical aspect of the discursive style in this book is the frequent use of ‘red herrings’. Many sections initially lay out some data and arguments in favour of theory α, and then present conflicting data and arguments in favour of a different analysis, β. At this point the reader will assume that the discussion will culminate in a conclusion in favour of β, but more often than not the discussion will return to α by presenting other types of arguments against analysis β. This procedure is so frequent in this book that I needed to withhold any judgment until I had finished the entire chapter, as on many occasions I had been led down the wrong path.
3.2 Errors
There is a substantial number of typographical errors in the English text in this book. More important is that there is a high rate of occurrence of errors in linguistic transcriptions. It is, of course, difficult for readers, like me, who do not speak Icelandic and Faroese to notice such errors, but I nevertheless found more than 30 errors in transcription within the first two parts of the book.
Given the number of errors, it is not easy to determine whether differences in transcriptions are due to typographical errors or inconsistencies. On pages 158–160, for example, the Faroese word bonkinum ‘bench.dat.sg.def’ (which is intermittently spelled beinkinum and benkinun, without explanation) is transcribed both as [pɔɲʧinʊn] and [pɔiɲʧinʊn], and the word handilsmaður ‘merchant’ as [hant
smεaːvʊɹ] and [hantilsmεaːυʊɹ]. It is not possible to determine without a solid understanding of Faroese whether these differences are due to errors or whether they reflect speaker or dialect variation.
Most of the mistakes (in the first two parts of the book) seem to affect the Faroese transcriptions. Sometimes both the transcription and the spelling of Faroese words change without comment. The Faroese word for ‘long’, for instance, is on some pages given as langur, with the transcription [lεŋkʊɹ] (pp. 25, 26, 176), but on other pages as longur, with the transcription [lɔŋkʊɹ] (pp. 115, 124, 297). This leaves the reader bewildered as to what the Faroese word actually is. According to dictionaries, there is no Faroese word longur meaning ‘long’. The word langur is the adjective ‘long’, whereas longur is the comparative of the adverb leingi ‘for a long time’ (Poulsen et al. Reference Poulsen, Simonsen, Jacobsen, Johansen and Hansen1998: 671, 684).
The most striking transcription errors concern the Faroese consonant /r/. This consonant is according to Árnason normally realised as [ɹ] (pp. 114–115), yet the transcriptions vary somewhat randomly between [r] and [ɹ], even within transcriptions of the same word. Although [ɹ] is more common, I found more than 50 Faroese forms transcribed with [r] within the first 100 pages of the book. It is difficult to say whether the transcriptions with [r] are mere typos (which would bring the number of typos within the first two parts of the book close to 100), or just a recurrent inconsistency.
I only collected typos from the first two parts of the book, but given that there are a few dozen in the beginning of the book, it is fair to estimate that there are at least as many as 100 in total. This constitutes a major problem for this book as a reference work for Icelandic and Faroese phonology, since readers cannot use or cite a form from this book and at the same time be somewhat confident that the form is correct.
3.3 Phonological explanations
The phonological explanations in this book are generally couched in an explicit teleological framework, in which phonological settings ‘want’ or ‘aim’ to change from their original configuration to a more ‘natural’ one (e.g. pp. 24, 47, 50, 152), although it is never explained what is meant by ‘natural’. In practice, however, many of these explanations are little more than descriptions or reformulations of the data. There are many examples of this in the book, but I will note only two here.
(i) The loss of /i/ and /y/ in a variety of Icelandic is called paradoxical, because it occurs only in open syllables and not in closed syllables. But ‘the paradox is only apparent’, because this ‘natural change, caused by some tendency in the system, [is] checked in closed syllables’ (p. 134). Calling the change ‘checked’ is a mere reformulation of the fact that the change does not occur, so it is not clear exactly how this analysis resolves the paradox.
(ii) The standard dialect of Icelandic distinguishes between aspirated and unaspirated stops both word-initially and postvocalically, as in [tʰaːla] ‘talk.inf’ vs. [taːla] ‘valley.acc.pl’. After a voiceless consonant, on the other hand, only unaspirated stops exist, as in [spaːði] ‘spade’ and [hεsty
] ‘horse’, rather than the phonotactically illicit forms *[spʰaːði] and *[hεstʰy]. These stops are nevertheless analysed as being underlyingly aspirated (p. 217ff). The question then is why they surface as unaspirated. Three explanations are offered to account for this (cf. the comment in §3.1 that explanations tend to change during the course of the book): (a) they do not ‘need’ to be aspirated (p. 217), (b) the aspiration ‘is simply implausible for phonetic reasons (presumably to do with articulation)’ (p. 218) and (c) the glottal feature causing aspiration is ‘suppressed’ in ‘foot-internal position’ (p. 226). But since the phonetic reasons for the missing aspiration are not given (other than stating that they presumably have to do with articulation), the basis for concluding that it is phonetically implausible to realise the forms with aspiration is unclear. And calling a missing feature ‘suppressed’ does not explain why the feature is missing – it simply restates the fact that it is missing.
3.4 Phonetics
As mentioned above, a number of phonological phenomena in Icelandic and Faroese are analysed as caused by syllabic structures and syllable boundaries. One good reason to question the status of syllable boundaries as linguistic primitives, however, is the fact that they have no direct acoustic correlates. More details are therefore needed to back up the claim made in this book that it is possible to directly ‘observe’ the syllable boundaries in Faroese words (p. 71).
The Faroese sections of this book contain many discussions of phonetic detail, but since no phonetic measurements are reported, it seems as if these details are impressionistically determined. On pages 94–95, for example, a Faroese informant is claimed to have a wide range of vowel allophones in unstressed syllables: [i ʊ ε ɵ ə ɜ ʌ ɑ a]. One might wonder, however, whether it is really possible to accurately perceive the distinction between vowels such as [ε ɵ ə ɜ ʌ] in unstressed syllables in a language variety of which one is not a native speaker (as is the case of the author).
A few phonetic descriptions in this book are somewhat puzzling, such as: ‘the alveolar stops have a purely dental articulation’ (p. 99). If the stops are ‘purely’ dental, then they cannot be alveolar. In Faroese, the ‘dental stops … may be slightly ‘less dental’ (or more alveolar or apical) than the corresponding Icelandic ones' (p. 115), but whether or not the active articulator is apical bears no direct relation to whether the passive articulator is dental or alveolar. Finally, in a few places (e.g. pp. 115, 124, 126) things like ‘[lj] or [l
]’, ‘[jɔuːs] or [ɔuːs]’ are written. But phonetically speaking, it is not clear to me what the difference between [j] and [] is.
3.5 Historical phonology
As mentioned in §1, the book has an entire chapter on the phonology of Old Norse. There are, however, a few erroneous claims about Old Norse phonology in that chapter. For one, it is claimed that voiced and voiceless stops are neutralised as voiceless stops in postvocalic position (p. 6), but this is not the case, cf. ON stabbi ‘stump’ vs. stappi ‘mash’, greddir ‘feeder’ vs. grettir ‘frowner’, and baggi ‘bag’ vs. bakki ‘hill’.
It is also claimed (pp. 6–7) that Old Norse nasals were required to agree in place with a following stop (a claim based on the vocabulary of Icelandic and Faroese rather than the Old Norse vocabulary), but this is also not the case; e.g. ON jamtr ‘men from Jämtland’, klymta ‘murmur’, kanpr ‘moustache’, etc. A footnote on page 294 admits that such clusters exist in Icelandic as well, as in [scε
ta] ‘to entertain’.
Some pre-Old Norse consonant assimilations are discussed in Chapter 12, where the sequence -tt- in Icelandic words like sótti ‘sought’ is said to stem from an ‘assimilation of /kt/ to /tt/ which took place in Proto-Nordic’ (p. 250). But the consonant sequence */kt/ did not exist in Proto-Nordic or even in Proto-Germanic, due to the workings of Grimm's Law, according to which the Proto-Indo-European sequence */kt/ changed to Proto-Germanic */ht/. In other words, the word sótti has -tt- from Proto-Nordic */ht/.
Icelandic, Faroese and Norwegian are sister languages, all descended from Old Norse, which itself is very well attested. When diachronic developments of Faroese are discussed in this book, however, references are not made to Old Norse, but to Icelandic (e.g. pp. 122, 174). On other occasions Icelandic is explicitly treated as the origin of Faroese (p. 124). The choice of Icelandic as the starting point of Faroese in this book has led to what I consider to be a clear misanalysis of Faroese diachronic phonology, as will be illustrated in the following.
The ON initial clusters kn- and hn- have merged in Icelandic as hn- (Þórólfsson Reference Þórólfsson1925: xxxii, Hægstad Reference Hægstad1942: 109f), as seen in (1). In Faroese and Norwegian, on the other hand, the two clusters remain separate, in that kn- is retained whereas hn- ends up as n- (Hægstad Reference Hægstad1917: 126, Indrebø Reference Indrebø, Hovda and Thorson1951: 85). This is seen in the two rightmost columns in (1).Footnote 1
(1)
Well-established principles of the comparative method guide us in cases like this. If one language (Icelandic) has x where its sister languages (Faroese and Norwegian) have x and y, and there is no phonological condition that can explain how x and y could have split from *x, then *x and *y must have merged as x in the first language. As the table above shows, Icelandic has two words hnoða, and there is no phonological condition which can explain why the Faroese and Norwegian cognates have n- in noða but kn- in knoða. The conclusion must therefore be that the original onsets in these two words have merged in Icelandic, but remained distinct in Faroese and Norwegian. The parent language of Icelandic, Faroese and Norwegian is, however, so well attested that there is no real need to invoke any comparative method at all – we can simply look at Old Norse, where there is ample evidence to demonstrate that kn- and hn- merged in Icelandic as hn-.
By virtue of taking Icelandic as the origin of Faroese (p. 124), on the other hand, Árnason is forced to conclude that the initial h- in Icelandic words like hnoða, hnot and hnota has either been lost in Faroese (pp. 124, 174) or developed into a stop (p. 124), without any attempt to explain when we find one or the other. As should be clear from the exposition above, a sound change of the type hn->kn- did not occur in the history of Faroese.
3.6 Typesetting concerns
The typesetting shows a number of undesirable properties, despite the book being published by the prestigious Oxford University Press. The spacing between characters is often too wide, or sometimes so narrow that two characters overlap, and some characters are wrongly italicised. Diacritics are commonly misaligned, sometimes to the extent that the main character and the diacritic are written completely apart when they should be combined, as in the representation for what should be [
] (p. 97). On other occasions the wrong IPA diacritic is used altogether, as when the diacritic for non-syllabicity [
] is used for the diacritic for dental articulation [
] (p. 115), and when [
] (which is not an IPA diacritic) is used for [
] to indicate a lowered pronunciation, as on page 169. It is not uncommon to find the wrong IPA character, as when the bilabial fricative [Φ] is used instead of the vowel [ø] (p. 23), an italicised lateral [l] for an approximant [ɹ] (p. 81), a voiceless velar lateral [
] instead of a voiceless approximant [
] (p. 95), or a palatal stop [c] for an open vowel [ɔ] (p. 242). The phonetic symbol [i] is throughout the book typed with the Greek letter [ι].
Word endings are often typed with an en dash instead of a hyphen, e.g. –ir for the correct -ir. The slashes / / used in phonemic transcriptions are often separated from the actual transcriptions across line breaks, and a great many transcribed words are also split in the middle across line breaks. Phonological rewrite rules are written without the arrow, as on page 236, where we find ‘something like a - œ / _y’. The book also uses ligatures inside IPA transcriptions, which can be misleading. The ligature [fi], for instance, is easily mistaken for a transcription [fi] rather than the intended transcription [fi]. It is overall regrettable to find so many typesetting issues in phonological transcriptions in a reference work.
3.7 Conclusion
Although this book remains quite interesting because of its strong stance on the importance of syllabic structure for the phonotactic properties of languages, it is marred by an unnecessarily complicated discursive style with many inconsistencies, and by its large number of typographical and factual errors and substandard typesetting. Most of these problems might be resolved in a careful preparation of a new edition, and it would be of great benefit to the international phonological community if a more reliable source of Icelandic and Faroese could be made available by a careful and rigorous reworking of this book.
