Vowel breaking in today’s English*
0 Introduction The term ‘vowel breaking’ is used in diachronic linguistics to describe a process whereby a monophthong changes into a diphthong due to the effects of a following consonant. This term is also used in synchronic linguistics (especially in rule-based theories) in reference to the schwa-ending diphthongs of Modern English. However, since this nomenclature suggests the interaction of a vowel and a consonant, there is a risk that we accept this suggestion as an axiom, and treat the phenomenon accordingly. Though this premise is historically accurate, we shall see that when it comes to describing Modern English, such an analysis imposes restrictions on the theoretical possibilities and leaves certain relevant questions unanswered while posing further challenges. If we accept that what is commonly referred to as ‘breaking’ in English is indeed the interaction of a vowel and a consonant, several challenges must be faced. These include the problem of defining the leftenvironment (i.e. the sounds preceding the schwa), the asymmetry between the behaviour of long and short vowels, and the phonological status of the resulting diphthongs. Furthermore, traditional accounts (e.g. Gussenhoven & Weijer 1990, Krämer 2005, Heselwood 2006, Kijak 2011) often cannot explain why it is the schwa and not another vowel that appears as a result of ‘breaking’. 1 If, however, we treat English ‘breaking’ as the result of the incompatibility of two consonants, the phenomenon can be defined more easily, and the unanswered questions of the traditional accounts can be explained. Although the approach proposed in this paper is not without precedent in the literature (McCarthy 1993, Halle & Idsardi 1997, Orgun 2001), it is included only implicitly, and the argumentation behind the analysis is not made clear. The aim of this paper is therefore twofold. On the one hand,
LingDok Conference in 2015. 1 Although both Gussenhoven & Weijer 1990 and Kijak 2011 argue that the schwa is the result of spreading, their explanation is not satisfactory, since based on their account, we would expect a low vowel to appear. Even if we accept that the schwa is low, (‘The characterization of [ə] as [+low] will be uncontroversial.’ Gussenhoven & Weijer 1990:320), it is unclear why the schwa and not another low vowel appears. In Kijak 2011, it is the ‘A’ prime that spreads, which is also associated with low vowels.
Vowel breaking in today’s English 48
I wish to demonstrate that the term ‘vowel breaking’ itself is inaccurate and misleading from the point of view of a synchronic analysis of English, 2 as well as to show that the chosen transcription system confines the analyst, and therefore often ‘the viewpoint creates the object.’ On the other hand, it is my intention to define the phonotactic basis of the phenomenon, which can serve as a starting point for an improved analysis regardless of the theoretical framework. Therefore, the propositions made in the paper are meant to be applicable and understood within the context of a wide range of phonological theories,3 rather than in terms of a single specific framework.
1 ‘Breaking’ in English There are two kinds of ‘breaking’ in English: pre-L and pre-R breaking. From a diachronic perspective, it means that these consonants changed the preceding vowel into a schwa-ending diphthong. However, not all schwa-ending diphthongs are a result of this historical process. In words like museum /mju’zɪəm/, Ian /ɪən/, messiah /mɪ’saɪə/, etc. there is no R or L in the vicinity of the diphthong. This paper is concerned with all of the Vər and Vəl sequences, and the proposed approach covers even those instances that are traditionally not considered ‘breaking’ (e.g. royal, Bayard, etc.). Therefore, we will use the term ‘breaking’ here in reference to any Və[liquid] sequence. Table 1 shows the environments in which a schwa can appear before R. (1) Pre-R Breaking in English4
_®|| _®# _rV
i: beer cheerful series
e: bear bearing parent
u:/ju: tour touring fury
eɪ layer layering -
ɔɪ Moir moira
aɪ tyre tiring diary
aʊ hour towering dowry
_®C
beard
scarce
gourd
Bayard
-
iron
coward
2 While the primary concern of this paper is SSBE, other accents of English will also be
referred to in order to include those instances of breaking that are unattested in SSBE (see Table 3). 3 Especially the ones that make use of the notion of syllables and syllabic constituents. 4 The symbols in the first row represent the vowel before schwa insertion. In Gimsonian transcription, these words would be transcribed as [bɪə], [bɛə], [tʊə], etc., showing the effects of laxing and compression. The purpose of the simplification is to better illustrate that the environment of pre-L and pre-R breaking is the same (see Tables 3 and 4). The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 49
It must be noted that the data in the table is subject to variation both in the pronunciation of the schwa and in the pronunciation of the R. In non-rhotic accents (as e.g. in SSBE), 5 the status of the R is disputable in the words of the first and second rows, since the R is only pronounced if it is followed by a vowel. The first row shows ‘breaking’ before a pause (||). As regards the realization of the liquid, this environment is equivalent to that of the forms in the second row. Therefore, there is no R in cheerful,6 but there is in bearing or in the phrase tour is. Even more problematic are the words in the last row, where the R is not recoverable. This paper is not concerned with the status of R in these words. Nevertheless, the approach put forward here can explain the schwa in these cases too (if an R is hypothesized);7 therefore, these examples will be included. The other variable in the data is the pronunciation of the schwa. If we examine SSBE as spoken today, we must conclude that the majority of speakers (especially the younger generation) does not pronounce a schwa in the words of the first three columns. Therefore the word beer (formerly /bɪə/) is pronounced /bɪ:/ today, bear (that used to be /bɛə/) is /bɛ:/, and tour (/tʊə/), is /to:/ now. To get a clearer picture of the pronunciation of today’s SSBE, let us consider Table 2, in which the phoneme symbols of Table 1 are compared to those of the more up-to-date CuBE system (Lindsey and Szigetvári 2013). (2)
Pre-R Breaking and today’s pronunciation
Trad. CuBE _®|| _®# _rV _®C
i: ɪ: beer cheerful series beard
e: u:/ju: ɛ: o:/jɵː bear tour bearing touring parent fury scarce gourd no schwa today
eɪ ɛj layer layering Bayard
ɔɪ aɪ aʊ oj aj aw Moir tyre hour tiring towering moira diary dowry iron coward schwa possible today8
5 The abbreviation SSBE stands for Standard Southern British English (as spoken today). It is
synonymous with CuBE (Current British English), but I will treat them differently in this paper, since CuBE will be reserved for the name of a transcription syste m, while SSBE will be used in reference to an accent. 6 When the whole paradigm is considered, R is present in related forms: cheering, cheery, etc. 7 Otherwise, the schwa must be considered lexical. 8 The presence of the schwa is subject to variation. In the case of words like moira, where smoothing is not possible, the schwa is less common than in the structurally similar diary, which is sometimes pronounced [‘daəri]. See 3.1. and Table 6 for further comments on the different behaviours of these words. The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 50
Since there is no schwa in the beer-bear-tour types of words today (as shown in Table 2), there is no reason to include them in our discussion of ‘breaking’ – at least not in connection with the present state of the language. Therefore, we will confine our investigation to those forms in which a schwa can still be pronounced today, but we will shortly revisit the problem of beer, bear, and tour in 3.2. Let us consider pre-L breaking now in Tables 3 and 4. (3)
_l|| _l# _lC
Pre-L Breaking in English i: eɪ ɔɪ feel fail foil feeling failing foiling field frailty spoilt
aɪ file filing child
u: fool fooling coolth
əʊ foal bowling cold
aʊ fowl howling -
As can be seen, the environment of the pre-L schwa-insertion is the same as that of the dark (velarized) L. It means that in contrast with pre-R breaking, there is no prevocalic position here. It also means that the appearance of the schwa depends on the darkness of the L in the forms of the second row. That is to say, if a speaker pronounces a velar L in feeling, then the schwa can appear; if however, the L is clear, there is no epenthesis. The fool-foal-fowl types of words in the last three columns may or may not be pronounced with a schwa depending on the dialect. For instance, the schwa in these forms is optional in General American (Wells 2008:103), depends on the sociolect in Scottish English (Wells 1982:412), whereas in other accents (e.g. in SSBE), the schwa is not pronounced. Let us compare the transcriptions of these vowels with the corresponding symbols of CuBE. (Table 4). (4)
Pre-L Breaking and today’s pronunciation
Gimson i: eɪ ɔɪ aɪ u: əʊ aʊ CuBE ij ɛj oj aj ʉw ow aw _l|| feel fail foil file fool foal fowl _l# feeling failing foiling filing fooling bowling howling _lC field frailty spoilt child coolth cold The symbols used in the CuBE system show clearly that the former high monophthongs ([i:], [u:]) have diphthongized. This is an important difference, since it can now be seen that the schwa-insertion is only possible after a diphthong. Table 5 sums up the relevant data from the previous tables as well as the instances that are to be explained. The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 51
(5)
Və[liquid] sequences in today’s English
1 2 3 4
Gimson eɪ ɔɪ aɪ aʊ CuBE ɛj oj aj aw _®|| Moir tyre hour _®# layering tiring towering _rV moira diary dowry _®C Bayard iron coward
Gimson i: eɪ ɔɪ aɪ u: əʊ aʊ CuBE ij ɛj oj aj ʉw əw aw 5 _l|| feel fail foil file fool foal fowl 6 _l# feeling failing foiling filing fooling bowling howling 7 _lC field frailty spoilt child coolth cold -
2 Environment 2.1 The left environment If we accept Gimson’s phonemes as the starting point of our analysis, we must use at least two features to define the left environment: [non-low] and [long]. If however, we consider the sounds represented by the CuBE symbols, it can be seen that the schwa is always preceded by a glide. It means that the left environment can be defined by making reference to only one feature: [glide]. 2.2 The right environment The right environment of ‘breaking’ can also be defined in a non-disjunctive fashion. Let us notice that the liquid is followed by a morphological boundary (strong morpheme boundary or word boundary) in rows 1–2 and 5–6 of Table 5. It means that there is reason to believe that the liquid is in the coda in these cases. We have also seen in Table 3 that the environment of pre-L schwa-insertion coincides with the environment of the dark L. In terms of syllabic structure, it means that the L is in the rhyme. Based on these observations, it is not unfounded to hypothesize that the schwa can appear if the following liquid is tautosyllabic. The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 52
2.3 A simplified definition of the environment The environment of the schwa-insertion can therefore be defined as follows: glideliquid]σ. The epenthetic schwa between the two consonants is due to the phonotactic constraint blocking tautosyllabic glide+liquid clusters. Such clusters of incompatible consonants are resolved by the insertion of a vowel. English ‘breaking’ is therefore the resolution of the phonotactic constraint
instance, if an accent has no schwa in rule, goal, and foal, the constraint is
3 Problems 3.1 Syllable boundaries If we accept that the schwa appears before a tautosyllabic liquid, one might ask whether the R of words like diary is in the coda (/dajr.i/). However, it does not follow from what has been argued earlier. Epenthesis is not the only source of the schwa, it might as well be lexical. The phonotactic constraint only excludes the possibility of an epenthetic schwa in e.g. joyride, where the R is clearly in the next syllable. Syllabic structure can also explain the variation between speakers. Let us consider the word tiring. It is pronounced by some as /tajərɪŋ/, while others have no schwa in this word: /tajrɪŋ/. Based on what we have said so far, we can either say that the word contains a lexical schwa in the case of the former speaker, or argue that the word is morphologically complex, and the R is therefore in the coda. Hence, the blocked cluster is resolved by the schwa. For the other speaker, who pronounces /tajrɪŋ/, the word has been lexicalized (become monomorphemic), and based on the Onset Maximization Principle, the R is in the onset of the following syllable. There is no blocked cluster in this case, and therefore there is no need to insert a schwa. Table 6 summarizes the types of words with prevocalic R.
9 However, the schwa of sour then remains unexplained.
The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 53
(6)
Words with prevocalic R
complex today Smoothing Smoothing possible
hiring
historically complex 1 Smoothing possible but not common r#V
prioress diary
schwa
always
always
motivation for schwa
R in coda
R in coda
example
historically monomorphemic complex 2 Smoothing Smoothing not obligatory possible (or vanishingly rare) mayor, virus prayer always never old-fashioned lexical
-
lexical if present
These categories are not deterministic, and words from each group can enter any other group. In the example mentioned above, tiring has left the category of morphologically complex words and entered the group of virus-types (in which there is no motivation for the schwa, and is therefore omitted.) 3.2 The beer-bear-tour types or words As mentioned earlier, most speakers of SSBE have no schwa in words like beer, bear, and tour. However, the question remains: based on the phonotactic constraint described above, can we explain the schwa if a speaker does pronounce it in these words? Although the theoretical possibility is given, it must be noted that while the phonetic reality is not far from the representations /‘bijə/ (beer) and /‘tuwə/ (tour), it would be difficult to argue that bear also has a glide: /bɛjə/. Note that the chosen analysis has no bearing on the question of R-liaison: the simple statement ‘an R can appear after long monophthongs or a schwa when followed by a vowel’ applies regardless of whether we choose to analyze these words as discussed above or as /bɪ:/ /bɛ:/ /to:/. 3.3 The problem of the diphthongs We have seen in 2.3 that the schwa appears between two consonants (and this is also what the consonant-letters suggest in the CuBE transcriptions). This raises the question of whether the two parts of a diphthong can be analysed differently. In other words, can we say that the CV skeleton of fight is CVCC and not CVVC if we have accepted that the third segment of file is a consonant? The discussion of this question is beyond the scope of this paper; The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 54
however, we must note that the simple fact that epenthesis is expected between two consonants does not prove that there are no diphthongs in English. We can conceive of such a theoretical framework in which diphthongs are made up of VC and not VV segments. In such a framework, the status of the English diphthongs is not wavered by our analysis. Szigetvári (2015) writes extensively about this problem.
4 Advantages 4.1 Unmarkedness A universal argument in favour of the approach presented in this paper is that it defines the process as the avoidance of a marked structure (CC). The sequence after schwa insertion (CVC) (7a) is unmarked, whereas the sequence hypothesized in traditional approaches (VVC) (7b) is a marked structure, and therefore markedness has to be explained, which complicates the theory. (7)
Unmarked vs. marked skeleton a)
b)
If we suppose that the schwa appears between consonants, the motivation of the process is obvious: the underlying marked CV skeleton is ‘repaired’ by the insertion of a V segment between the two consonants, making the structure unmarked. 4.2 The lack of schwa insertion after short vowels We have seen in 2.1. that ‘breaking’ is traditionally thought to affect long vowels only. 10 However, these accounts provide no explanation as to why there is no schwa insertion in words like fill, fell, full, etc. (i.e. after short 10 Short vowels before R are either unaffected (in words with V́
(lax)rV sequences, a.k.a. carrot-
words) or broadened, but never broken. The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 55
vowels). In other words, why do these theories have to make reference to a second feature, the [+long] feature? If we treat ‘breaking’ as described in this paper, the problem can be avoided. It falls out of this approach that short vowels are not affected. To be more exact, no vowels are affected by the process. It is the presence of a glide that triggers ‘breaking’, and there are no glides in fill, fell, and full that could combine into a blocked cluster with the following liquid. Therefore, the epenthesis is not motivated.
4.3 The quality of the inserted vowel In traditional accounts of ‘breaking’, the quality of the schwa is often unexplained. It might as well be the case that the inserted vowel assimilates to the height of the preceding vowel, as happened in Old English (Lass 1994:48). In Modern English, the quality of the inserted vowel is unaffected by the preceding vowel, which also suggests that the vowel has nothing to do with this process. On the other hand, the following consonant is also thought to affect the quality of the epenthetic vowel by some authors (as mentioned in connection with Gussenhover & Weijer 1990 and Kijak 2011), and in these cases, the vowel should ‘take over’ certain characteristics of the consonant. However, this does not seem to be the case. If we adopt the present approach, there is no need to explain the quality of the schwa, since it is the default vowel of English, and as such, it is the vowel that resolves blocked clusters.
5 Conclusion In this paper, I have argued that ‘breaking’ in today’s English should be considered the epenthetic resolution of a blocked consonant cluster. Therefore, the term ‘vowel breaking’ is inaccurate in reference to the present state of the language: the phenomenon is expected in a C_C environment, not in a V_C environment. The approach put forward here enables the analyst to answer the questions that are usually left unanswered in traditional accounts of ‘breaking’ regardless of the theoretical framework.
The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy
Vowel breaking in today’s English 56
References Gussenhoven, Carlos and Jeroen van de Weijer. 1990. On V-place spreading vs. feature spreading in English historical phonology. The Linguistic Review 7(4):311–332. Halle, Morris and Idsardi, William. 1997. R-hypercorrection and the elsewhere condition. In: Derivations and Constraints in Phonology. Oxford: Clarendon Press. 331–348. Heselwood, Barry. 2006. Final schwa and r-sandhi in RP English. Leeds working papers in linguistics 11:78–95. Kijak, Artur. 2011. Intrusive consonants: The internal structure of English liquids. Poznań Studies in Contemporary Linguistics 46(4): 407–427. Krämer, Martin 2005. English schwa insertion before liquids and phonological opacity. http://blogg.uit.no/mkr001/files/2013/03/mk-cls41.pdf. Lass, Robert. 1994. Old English: A Historical Linguistic Companion. Cambridge: Cambridge University Press. Lindsey, Geoff – Szigetvári, Péter 2013: Current British English - searchable transcriptions. http://seas3.elte.hu/cube/. McCarthy, John. J. 1993. A case of surface constraint violation. Constraint-based theories in multilinear phonology. Canadian Journal of Linguistics, special issue 169–195. Orgun, Cemil. O. 2001. English r-insertion in optimality theory. Natural Language & Linguistic Theory 19(4):737–749. Szigetvári, Péter 2015. No diphthong, no problem http://seas3.elte.hu/phono/notes/nodiphthong.html Wells, John. C. 1982. Accents of English [1–3]. Cambridge: Cambridge University Press. Wells, John. C. 2008. Longman Pronunciation Dictionary [3rd ed.]. Harlow: Pearson
The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Márton Jánosy