• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Stress retraction

Stress retraction refers to instances where compound stress is realized further to the left than when a specific word is pronounced in isolation / predicative position. Unlike iambic reversal, which is optional, stress retraction is obligatory and applies to complex words that are stressed on their right-hand constituent, with the exception of nominal compounds with irregular stress on the right-hand member of the compound (e.g. stadhuis [stɑd.ˈhœys] city hall).

[+]General information

Word groups that are subject to stress retraction are adjectival compounds of the type weak-strong, prepositional compounds and complex first names. For the first two word groups, the stress shift is visible in attributive position (most significantly adjectival compounds), in comparison to the pronunciation of that word in predicative position / isolation. The members of the last group, complex first names, change their stress patterns when followed by a surname.

[+]Adjectival compounds with compound stress on the right constituent in predicative position (type weak-strong)

A classic example of stress retraction is found in the different realizations of the complex word doodziek very ill, a compound of the adjectives ziek ill and dood very, lit. dead. In predicative position (1a), stress is on the right-hand constituent; in attributive position (1b), it is on the left-hand constituent.

a. hij is doodziek [dotˈsik] he is very ill
b. een doodziek kind [ˈdotsik] a very ill child

Stress on other syllables occurs only in the context of a contrastive reading (see (2)) or, alternatively, they can occur as a result of metalinguistic communication.

a. hij is niet ziek maar doodziek [ˈdod.zik] he is not ill but very ill
b. geen doodongelukkig maar een doodziek kind [dod.ˈzik] not a very unhappy but a very ill child

The stress shift in doodziek equals that of adjectives that are modified by a degree adverb, as is shown in (3):

a. hij is erg ziek [ɛrx.ˈzik] he is very ill
b. een erg ziek kind [ˈɛrx.zik] a very ill child

As indicated in example (1) in the topic on iambic reversal, iambic reversal is not only obligatory in adjectives modified by a degree adverb; it occurs in all adjectival compounds that have stress on their right-hand constituent in predicative position, except for nominal compounds. Consider the example in (4) of the adjectival compounddiervriendelijk animal-friendly. As can be deduced from (4b) and (4c), compound stress is realized on the second constituent in predicative position but on the first one in attributive position:

a. diervriendelijk [[dier][vriendelijk]] animal-friendly
b. Dit hotel is diervriendelijk. [dir.ˈvrin.də.lək] This hotel is animal-friendly.
c. een diervriendelijk hotel [ˈdir.vrin.də.lək] an animal-friendly hotel

Adjectival compounds of the type strong-weak are not subject to stress shift; that is, their stress pattern will always be strong-weak, independent of the context. Consider the example in (5) for the word zeeziek seasick:

a. Hij is zeeziek. [ˈze.zik] He is seasick.
b. een zeeziek kind [ˈze.zik] a seasick child
[+]Prepositional compounds

In predicative position and in isolation, prepositional compounds have compound stress on the right-hand constituent, in attributive position, they are stressed on the left-hand one. Consider the examples in (6), taken from Booij (1995):

attributive position:
a. onderaan [ɔn.dər.ˈan] at the bottom
      onderaan de berg [ˈɔn.dər.an] at the foot of the mountain
attributive position:
b. bovenop [bo.və(n).ˈɔp] over, above, on top
      bovenop de schuur [ˈbo.və(n).ɔp] on top of the shed
attributive position:
c. achterin [ɑx.tər.ˈɪn] in the back
      achterin de tuin [ˈɑx.tər.ɪn] in the back of the garden
[+]Complex first names (when combined with a surname)

Complex first names are subject to the stress shift when they are combined with a surname. This is shown for the name Jan-Pieter Man in (7). (7a) indicates the pronunciation of the complex first name Jan-Pieter in isolation, that is, without the surname. In this case, the right-hand constituent receives compound stress. In (7b), a surname Man is added. The surname receives nuclear stress, while the compound stress of the complex first name is shifted. That this shift is obligatory is indicated by (7c), which in terms of stress forms a minimal pair with (7b): in (7c), nuclear stress on Pie clearly indicates that the relevant syllable is part of the surname while in (7b), where Pie does not receive nuclear stress, it is part of the first name.

a. Jan-Pieter [jɑn.ˈpi.tər]
b. Jan-Pieter Man [ˈjɑn.pi.tər.ˈmɑn]
c. Jan Pieterman [jɑn.ˈpi.tər.mɑn]

Stress retraction is often considered to be the resolution of a stress clash (Gussenhoven 1984, Visch 1989, Backhuys 1989, Booij 1995). According to these analyses, the default stress pattern of adjectives like doodziek and APs like erg ziek is weak-strong and surfaces in predicative position. In attributive position, however, this would result in a so-called stress clash with stress on the following noun, and this is the reason for stress retraction. The classic examples are those repeated in (8):

a. doodziek   → doodziek kind
b. erg ziek   → erg ziek kind

The notion stress clash entails that stress will be shifted in those cases in which two prominent stresses occur too close to each other; in terms of metrical grids, this has been formulated as a prohibition of adjacency at two consecutive grid levels. In a less technical way, we can say that retraction is predicted to apply in situations where the most prominent syllables of two adjacent words in a phrase are not separated by a stressed syllable of a lesser degree of prominence.

This gives two scenarios under which stress retraction should occur. First of all, stress shifts should occur when two primary stresses are immediately adjacent, as in (8): e.g. in predicative position, ziek has compound stress in doodziek, thus it is more prominent than the initial constituent. If now, in attributive position, another word is added, as e.g. kind, the two prominent positions are ‘too strong’ to remain adjacent. To resolve this, compound stress shifts from ziek to dood, thereby changing the stress pattern of the compound. As a result, the two prominent syllables are now separated by a less prominent syllable; as this syllable, however, forms a prosodic word on its own, it has word stress.

For the analysis, the prominence of the second syllable, albeit weaker, is crucial, as the shift also occurs when there is an intervening entirely stressless syllable. Consider the example in (9) where the stressless syllable contains a schwa, which can never be stressed:

doodziek   → een doodzieke meid a very ill girl

The traditional analysis explains this by stating that the decline in stress must not be too strong between two adjacent syllables. Thus, the intervening syllable is prosodically too weak to avoid the stress shift. There are, however, a few problems with this standard analysis, as is pointed out in Köhnlein and Van Oostendorp (2012). First, it is not clear why intensifiers would have this different 'default' stress pattern. Trommelen and Zonneveld (1986) claim that the reason is that they are phrasal, for which the independent evidence is that they cannot be combined with degree adverbs:

? een erg doodziek kind

It is, however, difficult to determine whether the ungrammaticality here is truly syntactic or rather semantic in nature, if real at all. As there is no further independent evidence for a different constituent structure for intensifiers, the claim is thus not very strong.

Furthermore, we want to shed doubt on the idea that stress clash is responsible for the shift. In (11), the default stress would be five syllables away from the stress on the noun, which is hardly a clash; on the other hand, the resulting pattern has two adjacent word accents. The shift thus causes a clash rather than solving one. Yet it is absolutely obligatory:

een dik, doodziek paleontoloogje a fat, very sick little paleontologist
  • Backhuys, Kees-Jan1989Adjectival compounds in DutchBennis, H. & Kemenade, A. van (eds.)Linguistics in the NetherlandsDordrecht1-10
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Booij, Geert1995The phonology of DutchOxfordOxford University Press
  • Gussenhoven, Carlos1984On the grammar and semantics of sentence accentsDordrechtForis
  • Köhnlein, Björn & Oostendorp, Marc van2012Dutch stress retraction is an interface phenomenon
  • Trommelen, Mieke & Zonneveld, Wim1986Dutch morphology: evidence for the right-hand head ruleLinguistic Inquiry17147-170
  • Visch, Ellis1989The rhythm rule in English and DutchUtrecht UniversityThesis
printreport errorcite