• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Breaking: phonetic aspects
quickinfo

Several phonetic aspects of Breaking are reviewed here. These are:

  • The question as to whether it is feasible to assume one underlying representation for a centring diphthong and its broken counterpart;
  • Breaking as shortening, with respect to both the context in which Breaking shows up and the duration of broken diphthongs.

This topic deals with the above-mentioned phonetic aspects of Breaking in turn.

readmore
[+]One underlying representation for a centring diphthong and its broken counterpart?

The diphthongization of the half open and half close monophthongs was an essential historical 'pre-process' for Breaking (see Breaking: phonological aspects). However, there seems to be evidence that synchronically one and the same phonological form still underlies both the centring diphthongs and the corresponding glide + vowel sequences. This is the position taken by Van der Meer (1985), who assigns a word like stien /stiən/ stone the representation /stiɪn/, which underlies both the singular [stiən] and the plural [stjɪnn̩]. The argument Van der Meer adduces for this analysis is phonetic in nature. Measurements have revealed that centring diphthongs, on cutting off portions from their beginning, are felt to sound like the corresponding glide + vowel sequences. See the following two quotes:

It was found that in several cases the formant pattern of the broken diphthong can be obtained from the formant pattern oft the unbroken diphthong by reducing the duration of the first part of the unbroken diphthong. In a word like slɪət the second part of the diphthong ɪə is represented by formant values which correspond roughly to the vowel ɛ. We were also able to substantiate this hypothesis by auditive methods: when more and more segments of 12.8 msec are cut off from the beginning of a diphthong such as ɪə in slɪət, the observer at a particular point begins to hear the rising diphthong jɛ. This indicates that the phenomenon of Breaking may be related phonetically to shortening, and that in many cases it may be described as a reduction in the duration of the unbroken diphthong De Graaf and Tiersma (1980:118).

[T]he unbroken diphthong iə already contains the sound ɪ in its second part. Owing to masking by the more energetic i at the beginning of the diphthong this part is perceived as ə. When the initial part of the diphthong iə is reduced, the broken diphthong jɪ is heard at a certain moment De Graaf (1985:32).

Furthermore, a scheme in De Graaf and Tiersma (1980:79) shows that in the final stage of a centring diphthong, the values of its first and second formants are not those of the vowel schwa − as suggested by transcriptions like /stiən/(stien, stone) and /soəl/(soal, sole; insole) − but the ones that approximate those of the vowel that is one degree more open than the diphthong's first component.

In speaking and hearing, however, we imagine that we pronounce and hear schwa as the second component of these diphthongs (see also the above quote from De Graaf (1985)). De Graaf and Meinsma (1980:81-82) suggest that awareness of articulation and auditive perception might stand in each other's way here.

However this may be, there seem to be phonetic reasons to represent the alternating pairs [iə]~ [jɪ], [yə]~ [jø], [uə]~ [wo], [ɪə]~ [jɛ], and [oə]~ [wa] as /iɪ/, /yø/, /uo/, /ɪɛ/, and /oɔ/ in underlying representation.

extra

In essence, this analysis is adopted by Hermans (2007). He assigns /iɪ/, /yø/, /uo/, /ɪɛ/, and /oɔ/ the status of diphthongs with an optional second structural position. Since 'long rising diphthongs' are illegitimate, they cannot reach the surface unaltered. Either the optional structural position remains unrealized, in shortening contexts, or it gets the default interpretation schwa, in the remaning contexts.

There are, however, also arguments for the assumption that the alternating sound pairs at hand do not derive from a common underlying representation, but that they, instead, are both present as such in underlying representation.

If Breaking were a genuine, productive phonological process, it would be natural to assume a common underlying representation for both the centring diphthongs and the corresponding glide + vowel sequences. As it has convincingly been shown that Breaking is an opaque phenomenon, the words which take part in it must be learned. The alternation at hand therefore is best accounted for by means of stem allomorphy (or an alternative device).

This entails that both stems are stored in the lexicon, for which there are two independent pieces of evidence. Firstly, the centring diphthong /øə/ does not alternate with the glide + vowel sequence [jœ]. Since the vowel /œ/ is not a phoneme of Frisian, there cannot be an underlying representation /øœ/ from which both the centring diphthong [øə] and the glide + vowel sequence [jœ] derive. If both alternants of a Breaking pair are stored in the lexicon, as is the case with stem allomorphy in general, there is a principled explanation for why /øə/ occurs, whereas /jœ/ does not.

extra

A word like gleon /ɡløən/ glowing, red-hot thus can neither have the allomorph /ɡljœn/ - nor the disjunctive underlying representation /ɡl{øə/jœ}n/. Fokkema (1940:144) must be credited for noting that the centring diphthong /øə/ does not take part in Breaking (unlike the centring diphthongs which begin with the other close and half-close vowels); he wonders whether this can be ascribed to the fact that /ø/ no longer has a more open (lower) central counterpart, viz. [œ]. The latter only occurs in the Súdwesthoeksk dialect. Besides, it is part of the falling diphthong /œy/. But whereas /œy/ is largely restricted to loanwords, the Breaking relation is firmly entrenched in Frisian morphophonemics.

Secondly, Hoekema (1961:20-21) and Hoekema (1962) point out that there are pairs of words which are only distinguished by the presence of the centring diphthong vs. the corresponding glide + vowel sequence. Examples are listed in (1):

1
Examples of word pairs distinguished by the presence of a centring diphthong vs. the corresponding glide + vowel sequence
a. /iə/ vs. /jɪ/
liere /liərə/ lyre - ljirre /ljɪrə/ smoke-dried beef
liep /liəp/ cunning - ljip /ljɪp/ lapwing
b. /uə/ vs. /wo/
foet /fuət/ foot - fuort /fuot/ gone; immediately
goed /ɡuəd/ good - guod /ɡwod/ stuff, things
c. /ɪə/ vs. /jɛ/
earn /ɪən/ eagle - jern /jɛn/ yarn
weak /vɪək/ soft, pulpy - wjek /vjɛk/ ice-hole
d. /oə/ vs. /wa/
koarts /koəts/ fever - koarts /kwats/ recently
toan /toən/ tone - toarn /twan/ thorn

If the centring diphthongs and the corresponding glide + vowel sequences have the same underlying representation, they must be in complementary distribution. The word pairs in (1) show that this is not the case, hence that both are underlying.

extra

Hoekema (1986) notes that there are pairs of male and female proper nouns which are distinguished in the same way. His list is presented in the table below (ordered as the one in (1) above)):

Table 1
Male proper noun Female proper noun
Nies /niəz/ Njis /njɪz/
Riem /riəm/ Rjim /rjɪm/
Ruerd /ryəd/ Rjurd /rjød/
Doed /duəd/ Duod /dwod/
Oeds /uədz/ Uods /vwodz/
Roel /ruəl/ Ruol /rwol/
Sjoerd /sjuəd/ Sjurd /sjød/
Beart /bɪət/ Bjert /bjɛt/
Geart /ɡɪət/ Gjert /ɡjɛt/

Interestingly, Hoekema notes that the pairs of proper nouns in (ii) are distinguished by /a:/ vs. /a/:

Table 2
Male proper noun Female proper noun
Jaap /ja:p/ Jap /jap/
Klaas /kla:z/ Klas /klas/

The forms in the tables above illustrate the functional equivalence of vowel shortening and Breaking.

There is a third reason for not adopting the abstract underlying representations above. Abstraction is only useful if it results in a simpler lexicon. For instance, the underlying representation /stiɪn/ obviates the need to assume both /iə/, as in stien /sti.ən/ stone, and /jɪ/, as in stiennen /stjɪnən/ stones. There are, however, simplex words with /iə/ without complex counterparts with /jɪ/ and vice versa, like hiem /hiəm/ property, premises and bjirk /bjɪrk/ birch(tree). This means that /iə/ and /jɪ/ belong to the underlying inventory in any case.

Taken together, there seems to be enough synchronic evidence to assume different underlying representations for the alternating centring diphthongs and glide + vowel sequences.

[+]Breaking as shortening: the context

It is an old observation that Breaking and vowel shortening occur in by and large the same morphological environments and thus are 'functionally equivalent'. It is therefore tempting to conceive of Breaking and vowel shortening as one and the same process.

If Breaking is considered an instance of shortening, it is worth the effort to investigate whether it first and foremost manifests itself in phonetic ('natural') contexts of vowel shortening and hardly, or not at all, in those of vowel lengthening.

The literature on Breaking discerns the following contexts as the ones typical for Breaking:

  1. the first syllable of a word which is disyllabic of itself or is disyllabic due to derivation;
  2. the left-hand member of a compound;
  3. monosyllabic words with heavy final consonant clusters (notably those consisting of voiceless consonants), which are heavy of themselves or which are due to derivation.

These contexts deserve some comment:

  • As to context 1: long vowels have the longest duration in monosyllabic words. The duration decreases with the addition of (unstressed) syllables at the right-hand side of the word, until, with three extra syllables, a lower limit is reached Rietveld and van Heuven (1997:284-285).
  • As to context 2: a compound consists of two or more phonological words. Words with a centring diphthong in isolation quite often show up with Breaking when they are the left-hand member of a compound. The fact that this left-hand member is embedded in the compound as a whole seems to make for a configuration of shortening, or in any case for a context which resembles the first context to such an extent that Breaking is possible.

extra

In chapter four of Van der Meer (1985) eight hypotheses about the development of Breaking are put forward. In the eighth one (page 128-130), Van der Meer mentions a context in which heavy stress on the left-hand member of a compound seems to systematically prevent Breaking from occurring:

Hypothesis VIII:
In compounds whose second element is an adjective and whose first element is a noun, there is never breaking of the noun when it is used in a rather heightened emotional way to indicate the degree of the quality referred to by the adjective
This phenomenon is exemplified in (2):

i
deawurch /dɪə#vørɣ/ [ˈdɪəvørəx] [*ˈdjɛvørəx] dead tired
kroandea /kroən#dɪə/ [ˈkroəndɪə] [*ˈkrwandɪə] dead as a doornail
smoardrok /smoər#drok/ [ˈsmoərdrok] [*ˈsmwardrok] extremely busy
spierwyt /spiər#vit/ [ˈspiərvit] [*ˈspjɪrvit] white as a sheet
stienkâld /stiən#kɔ:d/ [ˈstiəŋkɔ:t] [*ˈstjɪŋkɔ:t] freezing (cold)

In general, the final /r/ of smoar and spier in smoardrok and spierwyt does not delete, though /r/-deletion is quite normal in this contect; that deletion does not occur here is also likely to be due to the heightened emotional way in which the (high) degree of the quality referred to by the adjective is expressed.

  • As to context 3: preceding a voiceless consonant, there is a phonetic tendency to interrupt a vowel's realization, which results in a relatively short duration. When a vowel precedes a voiced consonant, the vowel's intensity decreases in a more gradual way, resulting in a longer duration. The total time interval for the vowel and the following consonant is fairly constant, regardless of whether the consonant is voiceless or voiced. The longer duration of the voiceless consonant thus is at the expense of the vowel's duration, whereas the shorter duration of the voiced consonant is compensated for by the somewhat longer duration of the vowel Rietveld and Van Heuven (1997:236-237).

Though the above contexts are the ones typical for Breaking, they present an inherent contradiction. On the one hand, having a decreasing effect on vowel duration, they provide a 'natural' embedding for Breaking. On the other hand, the vowels in question are part of a stressed syllable. Now, it is a common phonetic fact that vowels have a longer duration in stressed than in unstressed syllables. As Rietveld and Van Heuven (1997:256) put it: Except for word-final position, (vowel) duration is a highly reliable correlate of stress, both in accentuated and unaccentuated words [translated from Dutch]. The upshot of this is that in the contexts mentioned above vowels are exposed to phonetic pressures of both shortening and lengthening or, put differently, that there are forces and counterforces here.

extra

This might be the reason that Breaking has never become a fully productive rule in the first place. The forces at hand seem to have reached a certain equilibrium, in that some words have Breaking and others do not.

If Breaking is an instance of shortening, it is expected to occur in contexts of vowel shortening and hardly, or not at all, in those of vowel lengthening. That is why it is less likely to show up in the following contexts:

  1. a stressed word-final syllable;
  2. monosyllabic words with a simple coda, in particular a voiced consonant.

These contexts deserve some comment as well:

  • As to the first context: according to Rietveld and Van Heuven (1997:256) the vowels of final syllables have a somewhat longer duration than those of syllables in non-final position, which they connect (on page 288-289) to the phenomenon of the 'progressive deceleration' of the speech rate towards the end of a sentence.
  • The second context is the 'natural' counterpart of the third context typical for Breaking: a voiceless consonant has a longer duration than a voiced one, while the total time interval for the vowel and the following consonant is fairly constant (irrespective of the consonant's voicing quality). The shorter duration of a voiced consonant therefore is compensated for by a somewhat longer duration of the vowel Rietveld and Van Heuven (1997:236-237).

Van der Meer (1985:21-29) gives an overview of the contexts in which Breaking occurs. There are hardly any instances of Breaking in the final position of monosyllabic words. Some well-known cases are found in North-East Fryslân − a dialect area known as "De Dongeradielen". See the following overview:

2
Instances of Breaking in the final position of monosyllabic words (found in the dialect area known as De Dongeradielen (in the North-east of Fryslân))
a. Verbal forms
dji /djɪ/ did (elsewhere: die /diə/ )
hji /jɪ/ had (elsewhere: hie /hiə/ )
stji /stjɪ/ stood (elsewhere: stie /stiə/ )
wji /vjɪ/ was (elsewhere: wie /viə/ )
kuo /kwo/ could (elsewhere: koe /kuə/ )
suo /swo/ should (elsewhere: soe /suə/ )
b. Adverbs
duo /dwo/ then (elsewhere: doe /du/ )
huo /vwo/ how (elsewhere: hoe /hu/ )

The instances of Breaking in (3a) are likely to have arisen in the context of cliticization, see the examples in (4):

3
Examples of the context of cliticization which may have given rise to the broken forms in (3a)
a. hy die der /diə dər/ [diədr̩] [djɪdr̩] wat by he did there something close (by) he had a side-line
b. it stie der /stiə dər/ [stiədr̩] [stjɪdr̩] noch it stood there still it still stood there
c. ik soe der /suə dər/ [suədr̩] [swodr̩] krekt hinne I would there just to I just wanted to go there

The verb and the following adverb constitute one phonological word here, a configuration which equals the first context where breaking can occur.

The adverbs in (3b) are followed by /t/ or /tst/ in case they function as subordinating conjunctions: doe't /du+t/ [dut] when, doe'tst /du+t+st/ [dust] when you, hoe't /hu+t/ [hut] how, and hoe'tst /hu+t+st/ [hust] how you, a configuration equalling the third context where breaking can ocur.

extra

As indicated, the verbs of (3a) have a centring diphthong outside the dialect area of "De Dongeradielen". They are also realized with only the (short) full vowel (see Sipma (1913:32, §124, point 8), so die /diə/ [diə] did, stie /stiə/ [stiə] stood, koe /kuə/ [kuə] could, and woe /vuə/ [vuə] wanted, for instance, may show up as [di], [sti], [ku], and [vu]. This is also noticed by Tiersma (1985:20) and Tiersma (1999:17), who call this 'allegro shortening', which occurs in rapid conversational speech styles.

Some words ending in a centring diphthong show Breaking when they are the left-hand, and stressed, member of a compound (see Van der Meer (1985:140-143)). Some examples are provided in (5):

4
Examples of left-hand members of compounds with Breaking
(De) Headammen /jɛ#damən/ place name (lit. hay dams)
reafonk /rjɛ#foŋk/ scarlet fever
sniebal /snjɪ#bɔl/ snowball
striebult /strjɪ#bølt/ straw pile

It should be noted that Breaking is losing ground here, to the extent that it is becoming obsolete. The average Frisian speaker realizes a centring diphthong in these and comparable compounds.

Though the second context where Breaking can occur is one of lengthening, broken forms do occur, which is exemplified in (6):

5
Examples of Breaking in monosyllabic words with a simple coda (in particular a voiced consonant)
a. With a voiced consonant as a coda
gjin /ɡjɪn/ not a, not any, no; none
guon /ɡwon/ some
skjin /skjɪn/ clean
spjir /spjɪr/ spruce
guod /gwod/ stuff, things
skuon /skwon/ shoes
hjir /jɪr/ here
b. With a voiceless consonant as a coda
djip /djɪp/ deep
ljip /ljɪp/ lapwing
kuos /kwos/ (name for a) pig
giet /ɡjɪt/ goes (3rd ps. sg. present tense of gean to go )
stiet /stjɪt/ stands (3rd ps. sg. present tense of stean to stand )
moas /mwas/ moss
extra

The forms giet /ɡjɪt/ goes, stiet /stjɪt/ stands and moas /mwas/ moss occur alongside forms with the corresponding centring diphthong; the distribution between both forms is a dialectal matter.

Though words with Breaking are less likely to show up in monosyllabic words with a simple coda (particularly a voiced one), there is

"a hard core of monosyllabic words that are genuine examples of Breaking"
Van der Meer (1985:22).

extra

There are also some morphonological tendencies as to the occurrence of Breaking. One is that when the plural form of a noun shows Breaking, the diminutive form will do so as well. Another is that broken diphthongs are hardly found preceding a voiced fricative (not at all preceding /ɣ/, hardly ever preceding /z/ and /v/). If the latter does occur, this is more often in the diminutive than in the plural form, counter to the general trend. De Graaf and Tiersma (1980:119) ascribe the latter to the fact that the fricative is voiced in the plural and voiceless in the diminutive, hence in the context given the preceding diphthong is likely not to show Breaking in the plural, whereas the opposite holds for the diminutive.

[+]Breaking as shortening: the duration of broken diphthongs

De Graaf and Tiersma (1980:113) note that there seems to be little "phonetic similarity" between Breaking and vowel shortening:

"It has been observed by several authors (...) that breaking and shortening occur in the same environment and are somehow related to one another. This holds notwithstanding the fact that there is, at least superficially, little phonetic similarity between the alternation of a falling diphthong such as ɪə with the rising diphthong jɛ on the one hand and the alternation of long and short vowels like i: and i on the other"
.

However, on page 119 they admit that

"[B]reaking clearly involves a reduction in length of the broken diphthong, and in this sense it is functionally equivalent to the process of shortening"
.

The reduction in length of the broken diphthong is much bigger than might be expected to result from merely adding a syllable to the words in question. This is addressed in the following quotation (on page 117):

"[T]here is a definite shortening in the length of a diphthong in a monosyllabic stem when another syllable is added by some morphological process. The length reduction without breaking [exemplified with lietenliət+ənsongs (plural form of lietsong), goedheidɡuəd+hidgoodness (derived from goedgood), boatenboət+ənboats (plural form of boatboat), roazenroəz+ənroses (plural form of roasrose), roaskeroəz+kəsmall rose (diminutive form of roasrose)] varies in our study (...) with ratios between the values 73 and 87%, while the length reduction with breaking [exemplified with sleattensljɛt+ənditches (plural form of sleatslɪətditch), weak(je)vjɛkto soak (converted form of the adjective weakvɪəksoft), guodlikgwod+ləkgentle; meek (derived from goedɡuədkind(-hearted)), huodsjevwod+sjəsmall hat (diminutive form of hoedhuədhat), doarrendwar+əndoors (plural form of doardoərdoor), doarkedwar+kəsmall door (diminutive form of doardoərdoor)] is significantly greater: the ratios fluctuate between 58 and 76%. (...) As a rough result, one may conclude that for the cases with breaking some 20% more shortening occurs than for the cases without breaking."

It appears to be the case that the context in which Breaking preferably displays itself, viz. in derived forms, is the one in which ('automatic', 'natural') phonetic shortening is most likely to occur. The other way around Breaking is hardly found where phonetic lengthening is a strong tendency, e.g. in word-final position. Be that as it may, the shortening involved in Breaking exceeds phonetic shortening by far. Breaking has thus been reinterpreted as a categorical, phonological phenomenon.

References
  • Fokkema, Klaas1940Over de Friese klinkersBundel opstellen van oud-leerlingen aangeboden aan Prof. Dr. C.G.N. de VooysGroningen/BataviaJ.B. Wolters Uitgevers-Maatschappij N.V.140-145
  • Graaf, Tseard de1985Phonetic aspects of the Frisian vowel systemNowele523-40
  • Graaf, Tseard de1985Phonetic aspects of the Frisian vowel systemNowele523-40
  • Graaf, Tseard de & Meinsma, Gerrit L1980De brekking fan sintralisearjende twalûden yn it FryskUs Wurk2977-82
  • Graaf, Tseard de & Tiersma, Peter1980Some phonetic aspects of breaking in West FrisianPhonetica37109-120
  • Graaf, Tseard de & Tiersma, Peter1980Some phonetic aspects of breaking in West FrisianPhonetica37109-120
  • Graaf, Tseard de & Tiersma, Peter1980Some phonetic aspects of breaking in West FrisianPhonetica37109-120
  • Graaf, Tseard de & Tiersma, Peter1980Some phonetic aspects of breaking in West FrisianPhonetica37109-120
  • Hermans, Ben2007Friese stijgende diftongen vormen een complex segmentPhilologia Frisica anno 2005 : lezingen en neipetearen fan it ... Fryske filologekongres1733-62
  • Hoekema, Teake1961In pear fonologyske opmerkingenUs Wurk1019-21
  • Hoekema, Teake1962Fonologyske wurdpearen mei op- en delgeande twielûdenUs Wurk111-3
  • Hoekema, Teake1986Brekking en lûdferkoarting yn FrouljusnammenUs Wurk3555-56
  • Meer, Geert van der1985Frisian breaking: aspects of the origin and development of a sound changeEstrikkenGrins/GroningenStifting FFYRUG
  • Meer, Geert van der1985Frisian breaking: aspects of the origin and development of a sound changeEstrikkenGrins/GroningenStifting FFYRUG
  • Meer, Geert van der1985Frisian breaking: aspects of the origin and development of a sound changeEstrikkenGrins/GroningenStifting FFYRUG
  • Meer, Geert van der1985Frisian breaking: aspects of the origin and development of a sound changeEstrikkenGrins/GroningenStifting FFYRUG
  • Meer, Geert van der1985Frisian breaking: aspects of the origin and development of a sound changeEstrikkenGrins/GroningenStifting FFYRUG
  • Rietveld, Antonius C.M. & Heuven, Vincent J. van1997Algemene FonetiekUitgeverij Coutinho
  • Rietveld, Antonius C.M. & Heuven, Vincent J. van1997Algemene FonetiekUitgeverij Coutinho
  • Rietveld, Antonius C.M. & Heuven, Vincent J. van1997Algemene FonetiekUitgeverij Coutinho
  • Rietveld, Antonius C.M. & Heuven, Vincent J. van1997Algemene FonetiekUitgeverij Coutinho
  • Rietveld, Antonius C.M. & Heuven, Vincent J. van1997Algemene FonetiekUitgeverij Coutinho
  • Sipma, Pieter1913Phonology and Grammar of Modern West FrisianLondon, New YorkOxford University Press
  • Tiersma, Pieter M1985Frisian reference grammarDordrechtForis Publications
  • Tiersma, Pieter M1999Frisian Reference GrammarAfûk, Ljouwert
printreport errorcite