The glottal stop [ʔ] is inserted before vowel-initial syllables within words after /a/ and /ə/ and often also at the beginning of a word, although unlike standard German there is not a requirement to use it.
/ɡ/ is not a native phoneme of Dutch and only occurs in borrowed words, like goal. [ɡ] also occurs as an allophone of /k/ when it undergoes voicing assimilation, like in zakdoek[ˈzɑɡduk].
In the north /ɣ/ often devoices and merges with /x/ to a post-velar [x̠] or uvular [χ] in all environments. In the south the distinction between the two phonemes is generally preserved as velar [xɣ] or post-palatal [x̟ ɣ˖]. Some southern speakers may alternate between the velar and post-palatal articulation, depending on the backness of the preceding or succeeding vowel. Velar, post-velar and uvular variants are called harde g ('hard g'), while the post-palatal variants are called zachte g ('soft g'). There's also a third variant called zwakke harde g ('weak hard g'), in which /ɣ/ is realized as [ɦ] and /x/ is realized as [h]. It is used in Zeeland and West Flanders, which are h-dropping areas, so that /ɦ/ does not merge with glottal variants of /ɣ/ and /x/.
In the Netherlands, /v/ can devoice and merge with /f/. According to Collins & Mees (2003), there are hardly any speakers of Standard Netherlandic Dutch who consistently contrast /v/ with /f/. In Frisian accents of Dutch, word-initial /v/ is always realized as [f].
In low-prestige varieties of Netherlandic Dutch (e.g. the Amsterdam accent) also /z/ can devoice and merge with /s/. In Frisian accents of Dutch, word-initial /z/ is always realized as [s].
Speakers who devoice /v/ and /z/ may also hypercorrectively voice /f/ and /s/: concert[kɔnˈzɛrt] ('concert'), Standard Dutch [kɔnˈsɛrt].
Some speakers may realize /ɦ/ without voice, that is [h]. In Frisian accents of Dutch, word-initial /ɦ/ is always realized as [h]. Some dialects, particularly those from the southwest, exhibit h-dropping.
In the Netherlands, /s/ and /z/ may have only mid-to-low pitched friction, and for many Netherlandic speakers they are retracted. In Belgium those are more similar to English /s z/.
The sequences /sj/ and /zj/ are often assimilated to palatalised [sʲ zʲ], alveolo-palatal[ɕ ʑ], postalveolar[ʃ ʒ], or similar realisations.
/ʃ/ and /ʒ/ are not native phonemes of Dutch, and usually occur in borrowed words, like show and bagage ('baggage'). Depending on the speaker and the position in the word, they may or may not be distinct from the assimilated realisations of the clusters /sj/ and /zj/. If they are not distinct, they will have the same range of realisations noted above.
/m/ and /n/ assimilate their articulation to a following obstruent in many cases:
Both become [m] before /p/ and /b/, and [ɱ] before /f/ and /v/.
/n/ merges into /ŋ/ before velars (/k ɡ x ɣ/). The realisation of /ŋ/ in turn depends on how a following velar fricative is realised. For example, it will be uvular [ɴ] for speakers that realise /x/ and /ɣ/ as uvulars.
/n/ may be realised as [ɲ] before /j/. This also occurs before /ʃ/ or /ʒ/, and before /sj/ and /zj/ under assimilation.
Exact pronunciation of /l/ varies regionally:
In the North, /l/ is 'clear' before vowels; and 'dark' before consonants and pause. Intervocalic /l/ tends to be clear, except after open back vowels /ɑ ɔ/. However, some speakers use the dark variant in all intervocalic contexts. Speakers who use dark /l/ generally also use it before /j/.
Some accents, for example the Amsterdam accent or the Rotterdam one have dark /l/ in all positions. Conversely, some accents in the eastern regions along the German border (for example around Nijmegen), as well as some Belgian speakers, have clear /l/ in all contexts.
The quality of dark /l/ varies; in the North it is pharyngealized[lˤ], but in final position many speakers produce a strongly pharyngealized vocoid with no alveolar contact ([ɤˤ]) instead. In Belgium it is either velarized [lˠ] or post-palatalized [lʲ̠].
The realization of /r/ phoneme varies considerably from dialect to dialect and even between speakers in the same dialect area:
The uvular trill[ʀ] is a common alternative, found particularly in the central and southern dialect areas. Syllable-finally, it may be debuccalized to [ɐ], much as in German. This is more common in the (south)eastern areas (Limburg, southeast Brabantian, Overijssel).
The coastal dialects of South Holland produce a voiced uvular fricative[ʁ], which causes it to merge with a uvular /ɣ/ and (if devoicing occurs) with /x/. In these dialects, schijven ("discs") and schrijven ("to write") are homophones.
Word-finally, the alveolar approximant[ɹ] is used by some speakers. In the Leiden dialect it is used everywhere in a word.
The realization of /ʋ/ also varies somewhat by area and speaker:
The main realisation is a labiodental approximant [ʋ], found particularly in the Netherlands.
A number of dialects in Belgium and the southern Netherlands use a bilabial approximant [β̞], which is like [w] but without strong velarization.
In Suriname and among immigrant populations, [w] is usual.
The final 'n' of the ending -en (originally /ən/, with a variety of meanings) is not pronounced in many areas, unless when stressing the word, making those words homophonous with forms without the -n. It is dropped both word-finally and word-internally in compound words. This pronunciation can be morphologically sensitive and can distinguish words, as the -n is dropped only when it is part of the distinct ending -en, but not when the word has a single stem that happens to end in -en. Thus, the word teken "(I) draw" always retains its -n because it is part of an indivisible stem, whereas in teken "ticks (plural)" it is dropped because it is a plural ending. These words are therefore not homophones in dialects which drop -n, despite being spelled identically.
Final -n is retained in the North East (Low Saxon) and the South West (East and West Flemish) where it is the schwa that disappears instead. This creates a syllabic [n] or (after velars) syllabic [ŋ] sound. Examples: laten[ˈlaːtn̩]; maken[ˈmaːkŋ̍]. Some Low Saxon dialects that have uvular pronunciation of /ɣ/ and /x/ (or one of them) also have a syllabic uvular nasal, like in lagen and/or lachen[ˈlaːχɴ̩]
This section requires expansion with: Needs more detail on assimilation, the conditions under which it occurs and the effects.. (November 2014)
Dutch devoices all obstruents at the ends of words. This is partly reflected in the spelling, the voiced "z" in plural huizen [ˈɦœy̆zə] becomes huis [ɦœy̆s] ('house') in singular. Also, duiven [ˈdœy̆və] becomes duif [dœy̆f] ('dove'). The other cases are always written with the voiced consonant, but a devoiced one is actually pronounced: the "d" in plural baarden [ˈbaːrdə] is retained in singular spelling baard ('beard'), but the pronunciation of the latter is [baːrt], and plural ribben [ˈrɪbə] has singular rib ('rib'), pronounced [rɪp].
Because of assimilation, often the initial /v z ɣ/ of the next word are usually also devoiced: het vee ('the cattle') is /(ɦ)ət feː/. The opposite may be true for other consonants: ik ben ('I am') [ɪg bɛn].
Dutch has an extensive vowel inventory consisting of 13 plain vowels and four diphthongs. Vowels can be grouped as back rounded, front unrounded and front rounded. They are also traditionally distinguished by length or tenseness. The vowels /eː øː oː/ are included in the diphthong chart to the right because many dialects realize them as diphthongs, though they behave phonologically like the other long monophthongs.
Dutch allophones of unrounded monophthongs, from Collins & Mees (2003:92, 130, 132 and 134). Black vowels occur before /r/ in Standard Netherlandic Dutch and Randstad, and the red vowel occurs before the dark /l/.
Dutch allophones of rounded monophthongs, from Collins & Mees (2003:92, 130, 132 and 134). Black vowels occur before /r/ in Standard Netherlandic Dutch and Randstad, and the blue vowel occurs before /ŋ/.
Vowel length is not always considered a distinctive feature in Dutch phonology, because it normally co-occurs with changes in vowel quality. One feature or the other may be considered redundant, and some phonemic analyses prefer to treat it as an opposition of tenseness. However, even if not considered part of the phonemic opposition, the long/tense vowels are still realised as phonetically longer than their short counterparts. The changes in vowel quality are also not always the same in all dialects, and in some there may be little difference at all, with length remaining the primary distinguishing feature. And while it is true that older words always pair vowel length with a change in vowel quality, new loanwords have reintroduced phonemic oppositions of length. Compare zonne(n)/ˈzɔnə(n)/ ("suns") versus zone/ˈzɔːnə/ ("zone") versus zonen/ˈzoːnə(n)/ ("sons"), or kroes/krus/ ("mug") versus cruise/kruːs/ ("cruise").
The distinction between short /i y u/ and long /iː yː uː/ is only slight, and may be considered allophonic for most purposes. In most environments except before /r/, most dialects realise them as short vowels. However, some recent loanwords have introduced distinctively long /iː yː uː/, making the length distinction marginally phonemic.
Notes about the long close-mid vowels /eː øː oː/:
They are often realized as narrow diphthongs [eɪ̯ øʏ̯ oʊ̯]. Dialects which use monophthongal pronunciations [eː øː oː] in all or most positions include some eastern and southern Netherlandic ones, Standard Belgian, and also many other Belgian accents.
Some speakers from Leiden use a very wide pronunciation of /eː/, namely [æɪ̯]. It is not used before /r/ (see below), and it does not merge with /ɛi̯/, which, for these speakers, is realized as [ɛː] instead.
Certain dialects of southern Holland have a more central starting position of these diphthongs, tending towards [əɪ̯ əʏ̯ əʊ̯]. The dialect of Antwerp realizes /eː/ as [ɛi̯~ɛə̯~eə̯] instead.
The long close and close-mid vowels are often pronounced with different vowel height or as centering diphthongs before an /r/ in the syllable coda:
The long close vowels become [iə̯ yə̯ uə̯], or are allophonically lengthened to /iː yː uː/.
The long close-mid vowels are commonly raised to near-close [ɪː ʏː ʊː], or converted to centering diphthongs [eə̯ øə̯ oə̯] or [ɪə̯ ʏə̯ ʊə̯], depending on the dialect. Some speakers may not have such allophones at all, and pronounce [eː øː oː] in every position. Some, particularly Netherlandic speakers have similar allophones before coda /l/, while others may maintain a long monophthong or even the usual closing diphthongs before /l/. Certain Netherlandic speakers may also neutralize some of the tense-lax vowel contrasts before coda /l/, similarly to neutralizations found in Cockney. Perhaps the most common neutralization is /ɛl/ - /eːl/: heel[ɦɛɫ], [ɦɛɤˤ] ('complete', 'full', 'whole', 'very'), Standard Dutch [ɦeːɫ].
The long open-mid vowels /ɛː œː ɔː/ only occur in a handful of loanwords, mostly from French.
The open vowels /ɑ/ and /aː/ are generally realised so that the latter is more fronted than the former, although the exact realisations may differ by dialect.
In many southern areas, the distinction is less clear as /aː/ has a more backed realisation, closer to [ɑː]. This is especially in Brabantian and Limburgish. In some subdialects, especially the city dialect of Antwerp, this backing is particularly strong, and /ɑ/ is fronted towards [a] as well, so that the two sounds effectively switch articulation.
This swapping of realisations may also happen in some dialects of North Holland, including notably that of Amsterdam. Fronted short [a] only occurs before some consonants, e.g. man[man].
In the Low Saxon-speaking areas, the local realisation of /aː/ may be particularly fronted, and tend towards [æː].
Several dialects have retained the distinction between the so-called "sharp-long" and "soft-long" e and o, a distinction that dates to early Middle Dutch. The sharp-long varieties originate from the Old Dutch long ē and ō (Proto-Germanic ai and au), while the soft-long varieties arose from short i/e and u/o that were lengthened in open syllables in early Middle Dutch. This distinction is not preserved in most modern standard Dutch pronunciations and is not recognised in educational materials, but it is still present in many local dialects, such as Antwerpian, West Flemish and Zealandic. In these dialects, the sharp-long vowels are often opening diphthongs such as [ɪə̯] and [ʊə̯], while the soft-long vowels are either plain monophthongs [eː] and [oː] or slightly closing [eɪ̯] and [oʊ̯].
/ɛi̯ œy̯ ɑu̯/ are the most common diphthongs and commonly the only ones considered "true" phonemes in Dutch. /ɑi̯/ and /ɔi̯/ are rare and occur only in a few words.
The "long/tense" diphthongs are generally analysed phonemically as a long/tense vowel followed by a glide /j/ or /ʋ/, where the latter has the allophone [w] between vowels, and [u̯] between a vowel and a consonant. Thus, what is underlyingly /eːʋ/ is realised on the surface as [eːw] or [eːu̯], and likewise for the others.
The first element of /ɛi̯/ is pronounced more open than the short vowel /ɛ/ by many speakers, and therefore may be more accurately transcribed as [ɛ̞i̯]. For some Netherlandic speakers it may be as open as [æi̯], or even [ai̯]. On the other hand, the dialect of Antwerp realizes /ɛi̯/ as [aə̯]. Many Belgians tend to monophthongize it [ɛ̞ː].
In most of the northern areas, /œy̯/ is pronounced with an unrounded onset ([ʌ̈y̯]). Some people in Belgium pronounce a monophthong [œː] instead. Before a vowel, and sometimes also word-finally, a glide is added as final element: [œy̯j] or [œj].
The first element of /ɑu̯/ is often raised by many speakers. In Belgium, a common pronunciation is [ɔu̯], although the first element is more open: [ɔ̞u̯]. Many Belgians tend to monophthongize it to [ɔ̞ː]. In the north, it is generally unrounded, giving [ʌu̯]. In some areas, particularly Holland, the first element may be fronted [au̯].
The first element of the long mid diphthongs /eːu̯ oːi̯/ may be raised or lowered: [ɪːu̯] and [ɔːi̯] respectively. However, not all speakers do this. Speakers that diphthongise /eː/ and /oː/ will sometimes also do this to the diphthong, resulting in a falling triphthong with two closing elements: [eɪ̯u̯]/[oʊ̯i̯] or the more centralised [əɪ̯u̯]/[əʊ̯i̯].
Most native Germanic words (the bulk of the core vocabulary) are stressed on the root syllable, which is usually the first syllable of the word. Germanic words may also be stressed on the second or later syllable if certain unstressed prefixes are added (particularly in verbs). Non-root stress is common in loanwords, as such words are generally borrowed with the stress placement unchanged. In polysyllabic words, secondary stress may also be present. Certain prefixes and suffixes will receive secondary stress: vòòr.kó.mən, wéér.lòòs. The stressed syllable of a word will receive secondary stress within a compound word: bóm.mèl.ding, ál.co.hol.per.cen.tà.gə.
While stress is phonemic, minimal pairs are rare, and marking the stress in written Dutch is optional, never obligatory, but sometimes recommended to distinguish homographs that differ only in stress. The most common practice is to distinguish een (indefinite article, which, as a clitic, bears no stress) from één (the cardinal number one). It is also written to distinguish some verbs, as stress placement on prefixes also carries a grammatical distinction, such as in vóórkomen (to occur) and voorkómen (to prevent). In vóórkomen and other verbs with a stressed prefix, the prefix is separable and separates as kom voor in the first-person singular present, with the past participle vóórgekomen. On the other hand, verbs with an unstressed prefix are not separable: voorkómen becomes voorkóm in the first-person singular present, and voorkómen in the past participle, without the past participle prefix ge-.
Dutch has a strong stress accent, like other Germanic languages, and uses stress timing due to its relatively complex syllable structure. It has a preference for trochaic rhythm, with relatively stronger and weaker stress alternating between syllables in such a way that syllables with stronger stress are produced at a more or less constant pace. Generally, every alternate syllable before and after the primary stress will receive relative stress, as far secondary stress placements allow: Wá.gə.nì.ngən. Relative stress will preferably not fall on /ə/, so syllables containing /ə/ may disrupt the trochaic rhythm. To restore the pattern, vowels are often syncopated in speech: kín.də.rən > /ˈkɪn.drə(n)/, há.ri.ngən > /ˈɦaːr.ŋə(n)/, vər.gə.líj.king > /vər.ˈɣlɛi.kɪŋ/. In words where secondary stress is imposed lexically onto the syllable immediately following the stressed syllable, a short pause is often inserted after the stressed syllable to maintain the rhythm, to ensure that the stressed syllable has more or less equal length to the trochaic unit following it: bóm..mèl.ding, wéér..lò.zə.
Historically, the stress accent has reduced most vowels in unstressed syllables to [ə], as in most other Germanic languages. This process is still somewhat productive, and it is common to reduce vowels to [ə] in syllables carrying neither primary nor secondary stress, particularly in syllables that are relatively weakly stressed due to the trochaic rhythm. Weakly stressed long vowels may also be shortened without any significant reduction in vowel quality. For example, politie (phonemically /poːˈlitsi/) may be pronounced [poˈli(t)si], [pəˈli(t)si] or even [ˈpli(t)si].
The syllable structure of Dutch is (C)(C)(C)V(C)(C)(C)(C). Many words, as in English, begin with three consonants, e.g. straat (street). Words that end in four consonants are mostly superlative adjectives.
/-k-/ > /-x-/: German machen vs. Dutch maken (help·info), English make
/-p-/ > /-f-/: German Schaf vs. Dutch schaap (help·info), English sheep
/-t-/ > /-s-/: German Wasser vs. Dutch water (help·info), English water
/-θ-/ > /-d-/: German das, Dutch dat (help·info) vs. English that
Dutch generalised the fricative variety of Proto-Germanic */ɡ/ as [ɣ] or [ʝ], in contrast with German, which generalised the stop [ɡ], and English, which lost the fricative variety through regular sound changes.
Dutch underwent a few changes of its own. For example:
/ft/ changed to /xt/ (North) or /çt/ (South), spelled ⟨cht⟩, but this was later reverted in many words by analogy with other forms. Compare English loft, German Luft, Dutch lucht (pronounced [lʏxt] (help·info) or [lʏçt] (help·info)).
Proto-Germanic */uː/ turned into /yː/ through palatalization, which, in turn, became the diphthong /œy/ (help·info), spelled ⟨ui⟩. Long */iː/ also diphthongized to /ɛi/ (help·info), spelled ⟨ij⟩.
De noordenwind en de zon hadden een discussie over de vraag wie van hun tweeën de sterkste was, toen er juist iemand voorbijkwam die een dikke, warme jas aanhad. Ze spraken af dat wie de voorbijganger ertoe zou krijgen zijn jas uit te trekken de sterkste zou zijn. De noordenwind begon uit alle macht te blazen, maar hoe harder hij blies, des te dichter de voorbijganger zijn jas om zich heen trok. Tenslotte gaf de noordenwind het maar op. Vervolgens begon de zon krachtig te stralen, en onmiddellijk daarop trok de voorbijganger zijn jas uit. De noordenwind kon toen slechts beamen dat de zon de sterkste was.
The phonetic transcription illustrates the speech of "a highly educated 45-year-old male who speaks Belgian Dutch with a very slight regional Limburg accent." Sentence stress is not transcribed.
De noordenwind en de zon waren ruzie aan het maken over wie het sterkste was toen er een reiziger voorbij kwam met een warme jas aan. Ze spraken af dat degene die erin zou slagen om de reiziger zijn jas te doen uittrekken de sterkste was. De noordenwind blies zo hard hij kon, maar hoe harder hij blies hoe warmer de reiziger zich induffelde. Uitendelijk gaf hij zijn poging op. Dan begon de zon hard te schijnen en de reiziger deed onmiddellijk zijn jas uit. De noordenwind moest toegeven dat de zon het sterkste was.
Gussenhoven, Carlos (1999), "Dutch", Handbook of the International Phonetic Association: A guide to the use of the International Phonetic Alphabet, Cambridge: Cambridge University Press, pp. 74–77, ISBN0-521-65236-7