• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Intonation
quickinfo

Dutch grammar has a repertoire of intonational patterns that can be described quite precisely, using boundary tones (%T, T%) and pitch accent tones (T*).

readmore
[+]The autosegmental analysis of intonation

An autosegmental analysis consists of a string of low and high tones (L, H). Different kinds of tone can be distinguished on the basis of the place they take up in the larger phonological structure. Pitch accents are located in the accented syllables, of which every sentence or sentence fragment will have minimally one. The pitch accents of Dutch begin with a starred tone, and may have trailing tones after it. Thus, H* is a monotonal pitch accent, while H*L is a bitonal one. Boundary tones are located on the edges of the intonational phrase (IP). At the left edge, there must be either a %L or a %H, and at the right edge there may be either an L% or an H%. Since the final boundary tone is optional, there are three options at the right edge of the IP. A description will therefore include three elements. First, how is the utterance divided up into IPs; second, where are the pitch accents and, third, what tones are selected (cf. Halliday 1970).

With the exception of the tone grammars of the Limburgisch and Rhenish dialects which have a lexical tone contrast (stoottoon vs sleeptoon, also Tonakzent/Accent 1 vs. Tonakzent/Accent 2) and which are for that reason tone languages, the tone grammars of the West Germanic languages have largely the same structure. In comparison with the tone grammars of other languages without lexical tone, the intonation of West Germanic languages is complex. We therefore introduce the grammar of Dutch in stages, beginning with a less complex tone grammar which covers the great majority of contours. This first stage is given in (1).

Figure 1
[click image to enlarge]

There are four positions in (1). The first is the left-edge boundary tone. The second the prenuclear accent, of which there can be any number, including zero, depending on the number of prenuclear accents in the sentence. The third is the nuclear pitch accent, the obligatory pitch accent of the sentence. The pitch accents in this position will at a later stage appear to be distinct from those in the prenuclear position. Finally, the fourth position is the right-edge boundary tone, which, as said above, is optional. As will be clear, the symbol 'Ø' stands for the absence of a right-edge boundary tone. For a sentence with two accents there will therefore be 2 x 4 x 4 x 3, or 96 contours.

The pronunciation of a tone is known as a target, which has a value on a vertical axis representing F0 and one on a horizontal time axis. The values of a tone's target on the two axes are determined by a large number of factors, like the language (Pierrehumbert's (1980) 'language-specific realization') and the phonological context (Pierrehumbert's (1980) 'context-sensitive realization'), quite apart from the variation due to style. The F0 contour arises through the way one target is connected up with an adjacent one.

extra

The descriptions of British English intonation that were published in the twentieth century had a strong influence on the autosegmental analysis of Dutch. Because of this, the description of Dutch differs from the autosegmental analysis of American English presented by Pierrehumbert (1980), who did not build on the descriptions in the 'British tradition'. The way in which Pierrehumbert (1980) conceived of the phonological representation as distinct from its phonetic realization represented a breakthrough in the research on intonation. Its importance therefore lies not primarily in the details of her analysis of American English, but in the renewed conception of the relation between phonetics and phonology.

[+]Intonation structures and their pronunciation

Contours that start with %L and have H*, L*, H*L or L*H for prenuclear and nuclear accents are among the most neutral contours of the language. Together with the three ways in which contours can end, L%, H% and Ø, they describe 48 two-accent contours, or 96 if the choice between %L and %H for the left boundary of the IP is included (see 1). Figure 2 consists of a single IP with four H*L pitch accents which ends in L%. The targets of tones are temporally aligned with three kinds of edges. These are the beginning of (the sonorant portion of) the rhyme of an accented syllable (for T*), the IP-boundary (for %T and T%) and the left edge of T* (for the trailing tone). In (2), the targets of the trailing tones have not yet been included. Of these three, the rhyme of the accented syllable is the only one that is given by the derivational history of the sentence, as it results from the way the accent rules of the language have provided words with accents. Each of the accented syllables must be associated with a starred tone (T*), as indicated with a vertical association line between the rime and T*. The boundary tones do not care about the syllabic structure, and are positioned (in the sonorant portions) on the edges of the IP.

Figure 2: Not a single doctor was to be found in the whole town.
[click image to enlarge]

Prenuclear and nuclear
In (3), targets of the trailing Ls have been added, together with the F0 contour. Trailing tones are pronounced differently depending on whether they occur in a prenuclear or a nuclear pitch accent. In prenuclear pitch accents, the trailing tone is pronounced late. Its target is just before the next pitch accent on the right. See Ladd and Schepman (2003) for phonetic measurements of the location of the beginning of the rise to H*. The vertical position varies with the time that elapses before the next accent, which is why the target of the first trailing tone, above the second syllable of hele whole, is higher than that of the second and third, above geen no and te to. The trailing tone of the nuclear pitch accent, by contrast, is located on the left: it comes immediately after the target of  H*. Because in (3) there is very little time between the targets of H* and L%, this synchronization of L with H* is not actually in evidence.

Figure 3: Not a single doctor was to be found in the whole town.
[click image to enlarge]

Continuation
In (4), we see what happens when there is time after a boundary tone or a pitch accent. The pitch is maintained until the next target is reached. It therefore is the morphemic structure of the tone string that determines the variation in the pitch between targets. Within morphemes, i.e. within a pitch accent, there is a linear interpolation, but between morphemes the value of the lefthand target is maintained. Figure 4 has one H*L-pitch accent, on morgen tomorrow, a nuclear accent because it is the last, while it begins with %L and ends with H%. The trailing tone, which is spoken immediately after the target of H*, leaves some distance to be covered till the target of H%, while equally, the %L at the beginning is separated by some distance from the target of H*. Such unspecified stretches of speech (between %L and H* and between L and H%) are pronounced by a continuation of the tone on the left. It means that this tone (L% and L) acquires one target on the left, as usually indicated by means of a bullet, and a second target on the right, indicated by an open bullet. Between those two targets, the pitch is interpolated, giving level pitch.

Figure 4: But shouldn't they better come to Amerongen tomorrow?
[click image to enlarge]

A further example of such ‘double alignment’ is given in (5). The targets of the tones of monotonal pitch accents H* and L* continue up until the next tone specification in the IP or, if no further tones follow, till the end of the IP. The target (or targets, in the case of a longer stretch) of the prenuclear L* is usually somewhat lower than that of the preceding %L.

Figure 5: And when all birds had arrived at the place of the meeting
[click image to enlarge]

A final point concerns the realization of the L in a nuclear pitch accent when no right-edge boundary tone follows (see figure 6). The location of its target is at first sight unexpected, because it is rightmost rather than leftmost. On the basis of the description above, its mid-pitched (or not quite low pitched) target should be located close to that of H*, and be continued till the end of the IP. As it happens, that 'half-completed' falling contour is acoustically close to the vocative chant, so much so that a pronunciation with a mid-level pitch after H* would be interpreted as the vocative chant, a quite different intonation. To keep the half-completed fall distinct from the vocative chant, the gradually descening half-fall is used. There is therefore only one target for the trailing L in the context, as shown in (6).

Figure 6: Why aren't there more of these cushions?
[click image to enlarge]

The left-edge %H
A %H at the start of an IP, already illustrated in (5), describes high pitch before the first pitch accent. Rather than the low pitched beginning as in (3), (4) and (6), the pitch would have the value of H* or be a little higher. Figure 6 forms a minimal pair with (7) in that sense.

Figure 7: Why aren't there more of these cushions?
[click image to enlarge]

There would appear to be a correlation between the choice of the left-edge %T and the first tone of the first pitch accent, T*, such that before L* a choice for %H is more probable (cf. 5), while before H* a choice for %L is preferred (cf. 6). A perception experiment tapping listeners’ intuition about the meaning of the two left-edge tones by Grabe et al. (1998) showed that speakers sound more engaged and friendly when %T and T* have opposite values, i.e. %L H* en %H L*. Their results suggest that (6) will sound friendlier than (7).

Upstep
Upstep refers to the higher pronunciation of an H-tone after another H-tone. It commonly applies to an H% occurring after an immediately preceding H-tone. In Dutch, it applies regardless of whether this is H* or trailing H. The concept of upstep was introduced in the description of the intonation of American English by Pierrehumbert (1980), where its application also involves L-tones. Figures 8a and 8b show this for H* and trailing H respectively, the two contexts that occur in Dutch. In (8a)Was je daar helemaal alleen naartoe gegaan? Did you go there all by yourself?, the pitch from -leen is mid or mid-high, and rises from the beginning of the last syllable in the IP. The same pattern can be seen in (8b)Gaan ze nou je dagboek inspecteren? Are they going to inspect your diary?, where the rise starts from the continued target of a trailing H. The meaning difference between (8a) and (8b) would appear to be small. In IP-medial position, it is the pitch in the sonorant portion of the rime that differentiates the contours. That with L*H in (8b) has low pitch in the sonorant part of the rime, here the [ɑ] of dag, but if it had had H* in that position, the vowel would have had mid or high pitch. In (8a), the H* appears on a rime with a longer sonorant portion, where the contour may have an early rise, as opposed to a fully low pronunciation of the vowel in the case of L*H, in which the rise may at best begin during the following /n/. The contrast was postulated for Dutch in Gussenhoven and Rietveld (2000), following Pierrehumbert (1980), who had included it in her description of American English. Gussenhoven and Rietveld (2000) showed that the degree of surprise increases when the target of H* is raised (just as is the case when other H-tones are raised), and increases when the target of L* is lowered. Since this semantic effect cannot be explained if the two contours are two pronunciations of the same phonological contour, both with H* or both with L*, they must have different representations. The contour in (8a) is known as the high rise, that in (8b) as the low rise.

Figure 8
[click image to enlarge]

Summarizing, the trailing tone is pronounced rightmost in prenuclear pitch accents and leftmost in nuclear pitch accents, and linear interpolations are made between the tones of a pitch accent. Exceptionally, the trailing tone of nuclear H*L has its mid target rightmost whenever it occurs without a following T%. Between boundary tones and pitch accents, pitch is governed by the lefthand tone. We here conclude the illustration of the grammar in (1).

[+]Expanding the grammar: Stage 2

In this section, we expand the grammar in (1) in four ways. First, we discuss Downstep, a meaningful lowering of H* after H-tones. Next, we deal with a tritonal prenuclear pitch accent, and explain its position in the grammar as a restructuring of a two-phrase contour. Third, we introduce the leading H, which occurs on the syllable immediately before the accented syllable. L-Prefixation is the fourth addition to the grammar, which consists of a shifting of the pitch contour to the right in order to make room for an L-tone that functions as the starred tone of the pitch accent. This rule leads to more tritonal pitch accents. Finally, we introduce the vocative chant.

Downstep
Every tonal representation for an IP is optionally subject to downstep. This option exists if there is minimally one H-tone that precedes an H*, regardless of any intervening L-tones. Any such representation therefore represents two contours, one with and one without downstep. The target of a downstepped H* is substantially lower that that of a non-downstepped H*. If there are more H*’s after the first downstepped one, these are progressively lowered, creating a descending intonation contour. The context on the left can be the left-edge %H, a trailing H, or an H*. Since downstep is contrastive, it is indicated in the transcription. Downstepped tones are notated as !H*.

In (9a) and (9b) a minimal pair of het winterkoninkje the willow wren is presented. The difference in the vertical target of the two contours (the F0) leads to a horizontal (i.e. temporal) difference for the falling movement, which is later in (9b) than in (9a). In (9a) the start of the fall occurs towards the end of the sonorant rime [ɪn] in win-, while that in (9b) begins before the vowel. In these figures we have included the second target of L% in order to indicate that the difference between the accented syllables will be maintained if there are more unaccented syllables before the accented one.

Figure 9: Downstep
[click image to enlarge]

The meaning of downstep is something like 'No contribution is now expected from the listener'. For this reason, it is a useful contour for the title of a story that is read to a class of school children. A child’s answer to 'Which story shall I read to you today?', asked by a teacher who is about to select a story from a book, might well be said as (9a), with no downstep, because the child expects a resolution of his suggestion. But (9b) would be the better option once the teacher started telling the story. The effect is something like 'And now the story begins!'

A well-known contour with downstep is one that became known as the flat hat (also known as the platte hoed). It is illustrated in (10a)Eerlijk duurt het langst Honesty is the best policy, which was described in Hart et al. (1990) as a contour consisting of an accent-lending rise followed by an accent-lending fall. In (10a), the context for !H* is the preceding H*. The non-downstepped variant of (10a), which is not given here, has a later fall, just like the fall in (9b). Figure 10b - Al die ingewikkelde regelingen zijn afgeschaft All those complicated regulations have been abolished - shows that downstep is iterative. That is, the target of every H* is lower than the one before. As a result, the H* pitch accents form a contour resembling the terracing of sloping terrains used for agriculture. Instead of H*, (10c) - We wachten tot Jan z'n jas aan gaat trekken We will wait till Jan will put his coat on - has H*L’s in pre-nuclear positon. The pronunciation of the trailing L-tones is observed as dips before the targets of H*, here on tot till en z'n his. Contour (10d) - Maar daarvoor hoef je 't niet te doen You don't need to do it for that - shows that the trailing H can provide the context for downstep. The target of H* is lowered, occurring after the end of the slow rise towards a point just before the accented syllable, on te.

Figure 10
[click image to enlarge]

An interesting contrast is shown in (11a, b)een ogenblikje geduld alstublieft one moment please, where the downstepped H* before H% in (11a) contrasts with L* in (11b). Bruce Hayes was the first to notice that the description in Pierrehumbert (1980) did not account for the contrast between !H* and L*. In her description, a downstepped !H* was analyzed as H+L*. A realization rule treated L* as if it was !H*. Later analyses have generally replaced H+L* by  H+!H* (Grice 1995: fn 7). Figure 11a is a request, but (11b) is an echo question. Their meanings are obviously different.

Figure 11
[click image to enlarge]

Prenuclear H*LH
An example of a prenuclear H*LH is shown in (12). Here, a prenuclear pitch accent has a steep fall, followed by a slow rise to the next pitch accent.

Figure 12
[click image to enlarge]

We will go to the theatre tonight

The contour would appear to contradict the regularity outlined above that prenuclear trailing tones are pronounced late, on the assumption that the first pitch accent is H*L. An analysis with H*L on vanavond tonight predicts a slow fall to the rising movement on schouwburg theatre. The steep fall would be expected if there was an IP-boundary after vanavond tonight, but there is no indication that there is any such prosodic break. Taking a different route, we could assume that after the trailing L of the first pitch accent there is another trailing tone H, which has a rightmost target, as is regularly the case for the last trailing tone. This analysis correctly describes the contour, on condition that a tritonal prenuclear pitch accent is introduced, H*LH, as shown in (13).

Figure 13
[click image to enlarge]

There are two arguments that can be presented in favour of the analysis in (13). The first is that Hart et al. (1990) defined a rise `5´ which only occurs after the slow rise that follows a prenuclear accent. This `5' has been indicated in the contour in (13), and evidently suggests the presence of two H-targets over de schouw-. It is important to see that the assumption of this additional rise, an acceleration of the slow rise described as `4´, was independent of and predated any tonal theory of Dutch intonation, and therefore represents independent confirmation of the tonal analysis in (7). The second argument is based on the meaning of (13). Its intonation suggests it is an answer to a question about the listener’s plans for the evening. That is, vanavond evening is the topic of the conversation, just as is we we, and constitutes the known information to which additional information is provided in (13): naar de schouwburg to the theatre. The same meaning is expressed in a more explicit way when the topic and the new information are spoken as different IPs. In that case, the topic has a pitch accent H*L followed by H%, as shown in (14). This suggests that (13) has developed from (14) by way of a faster realization, whereby the IP-boundary was lost and H% reinterpreted as a third trailing tone of the prenuclear pitch accent.

Figure 14
[click image to enlarge]

The introduction of H*LH as a prenuclear pitch accent in the grammar predicts that it can combine with other nuclear contours than H*L. This prediction is correct, as shown in (15), for instance, where it occurs before L*H H%. It conveys a high degree of surprise over the timing of the visit to the theatre, announced in response to a preceding statement by the listener about their visit to the theatre that evening.

Figure 15
[click image to enlarge]

Are you going to the theatre tonight?

Contour (13) is known in other West Germanic languages, too. It has however, only been described for Dutch so far. Examples of this contour are attested in the recordings of O’Connor and Arnold (1973), where they are depicted as (14).

Leading H
Before the T* of the pitch accent, there may be a leading H, as in H+L* or H+H*. The syllable preceding the accented syllable will be high-pitched as a result. The pronunciation of L* or H* then follows, together with that of any trailing tones. Within the pitch accent, downstep is an obligatory feature, so that H+L* and H+H* contrast as do !H* and L* in (11a, b). In (16b, c) - Vind je het behang mooi? Do you like the wallpaper? - both the mid target of H* and the low target of L are preceded by high pitch on the preceding syllable. This contrast was described for English by Grice (1995). Figure 16aKrijg je dan ook de uitslag te horen? Will you get the results then? is a question that does not imply any expectation of an answer. Figure 16b might suggest that the speaker finds the colour or design of the wallpaper ugly and expects that the listener will agree. Leading H is only realized on the immediately preceding syllable. If a string of initial unaccented syllables is high, this is to be attributed to a %H tone. If it precedes an H*, this H* may or may not be downstepped, since the context %H is not part of the pitch accent, and the usual optionality applies.

Figure 16
[click image to enlarge]

L-prefixation
H* can be preceded by an L-tone, which takes on the function of the starred tone, causing H*L to be L*HL. In effect, this moves the peak one syllable over to the right or, if the accented syllable is the last in the IP, to a later point in that syllable. In the latter case, the syllable will be lengthend to accommodate the extra tone. In the literature on English, this contour type is known as a ‘scooped’ contour (Ladd 1980), later as a contour with a ‘delayed peak’ (Ladd 1983). Gussenhoven (1983), writing on English, described it as a morpheme ‘Delay’, a modification which could also affect L*, causing the rise after L* to be later than in the contour without prefix-L. In Dutch, L-prefixation is rare, but has been reported for the Randstad, particularly Amsterdam (Peters et al. 2014). In (17), we give an example for the sentence in (16a). On final syllables, L*HL L%  will be 'truncated', meaning that most of the final falling part is left unpronounced. When the final segment is a voiceless obstruent, the pitch contour may look like a rise, with a barely noticeable falling section before the final voiceless part of the signal. This is shown graphically in (17b) - Die gaan er allemaal uit! They are going to leave!, where the truncated section of the contour line has been dotted and the unpronounced targets shaded.

Figure 17
[click image to enlarge]

The vocative chant
The vocative chant (also known as call contour) occupies a special position in the intonation system of Dutch for three reasons. First, as observed by Hayes and Lahiri (1991) for English, the pitch accent induces a bimoraic vowel in the syllable associated with the starred tone, neutralizing the quantity distinction between long and short vowels. Second, the trailing tone associates with stressed syllables after the nuclear pitch accent, as observed for English by Liberman (1975). Third, the trailing H-tone may be multiplied so as to appear on syllables with secondary stress within the word and on every post-nuclear main stressed syllable if there are more words, as observed by Gussenhoven (1993). The representation of the pitch accent is H*H. Because of the obligatory character of downstep within the pitch accent, its pronunciation is [H*!H]. The first of the two special properties are illustrated in (18a,b), where both tones are pronounced on the same syllable, an IP-final accented syllable, which has an initial high-pitched and a final mid-pitched part. Figure 18a has a short (lax) vowel [ɑ], in the proper name Han. The duration of the vowel is identical to that of the long (tense) vowel [aː] in haan cock, rooster in (18b), despite the fact that when pronounced with any other pitch accent, like a declarative H*L L%, the vowels do differ in duration. Second, the trailing H associates to syllables after the accented syllable, if there are any. It goes to the IP-final syllable if there are no syllables with word stress after the accented syllable, as shown in (18c), which has an unstressed medial syllable ne- which is high-pitched, like the first (accented) syllable, and a step down to the final -ke. Figure 18d has a syllable with secondary stress before the final syllable. It attracts the trailing H, while the final syllable obtains a copy of that tone, which is downstepped relative to the preceding tone on -ham-. Finally, the high pitch continues on post-accentual unstressed syllables in following words, here is and ge-, but ends at the primary stress of a post-accentual word.

Figure 18
[click image to enlarge]

[+]Transcription of texts

In the following section an example of an intonation transcription is provided. The willow wren was translated from the original Low Saxon De Tunkrüper taken from Wisser (1922). With thanks to Jörg Peters, transcriptions are provided with stylized pitch contours. Each black bullet is the target of a tone, white bullets represent a second target of a tone that has a continued pronunciation. The sound files were obtained from various sources comprising different variants and dialects. In table 1 a complete transcription of all the recordings is given as to provide the reader with the story in one piece, while in table 2 each contour is given separately with its corresponding sound file.

Table 1: Complete transcription and recording of 'Het winterkoninkje'
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]

Table 2: Separate transcriptions with stylized contours and corresponding recordings of 'Het Winterkoninkje'
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]
[click image to enlarge]

References
  • Wisser, Wilhelm (ed.)1922Plattdeutsche Märchen. Ausgabe für ErwachseneJenaDietrichs
  • Grabe, Esther, Gussenhoven, Carlos, Haan, Judith, Post, Brechtje & Marsi, Erwin1998Pre-accentual pitch and speaker attitudes in DutchLanguage and Speech4163-85
  • Grice, Martine1995Leading tones and downstep in EnglishPhonology12183-233
  • Grice, Martine1995Leading tones and downstep in EnglishPhonology12183-233
  • Gussenhoven, Carlos1983A semantic analysis of the nuclear tones of EnglishBloomington. IULC
  • Gussenhoven, Carlos1993The Dutch foot and the chanted callJournal of Linguistics2937-63
  • Gussenhoven, Carlos & Rietveld, Toni2000The behavior of H* and L* under variations in pitch range in Dutch rising contoursLanguage and Speech43183-203
  • Gussenhoven, Carlos & Rietveld, Toni2000The behavior of H* and L* under variations in pitch range in Dutch rising contoursLanguage and Speech43183-203
  • Halliday, Michael1970A Course in Spoken English: IntonationOxford University Press
  • Hart, Johan `t, Collier, René & Cohen, Antonie1990A perceptual study of intonation: an experimental-phonetic approach to speech melodyCambridgeCambridge University Press
  • Hart, Johan `t, Collier, René & Cohen, Antonie1990A perceptual study of intonation: an experimental-phonetic approach to speech melodyCambridgeCambridge University Press
  • Hayes, Bruce & Lahiri, Aditi1991Durationally specified intonation in English and BengaliSundberg, Johan, Nord, Lennart & Carlson, Rolf (eds.)Music, language, speech and brain: Proceedings of an international symposium at the Wenner-Gren Center, Stockholm, 5–8 September 1990Basingstoke, UKMacmillan78–91
  • Ladd, D. Robert1980The Structure of Intonational Meaning. Evidence from EnglishBloomington. IULC
  • Ladd, D. Robert1983Phonological features of intonational peaksLanguage59721-759
  • Ladd, D. Robert & Schepman, Astrid2003“Sagging transition” between high pitch accents in English: Experimental evidenceJournal of Phonetics3181–112
  • Liberman, Mark1975The Intonational System of EnglishPhD dissertation. Cambridge, MA. MITThesis
  • O’Connor, J.D. & Arnold, G.F1973Intonation of Colloquial EnglishLondonLongman
  • Peters, Jörg, Hanssen, Judith & Gussenhoven Carlos2014The phonetic realization of focus in West Frisian, Low Saxon, High German, and three varieties of DutchJournal of Phonetics46185-209
  • Pierrehumbert, Janet B1980The Phonology and Phonetics of English IntonationPhD dissertation. Cambridge, MA. MITThesis
  • Pierrehumbert, Janet B1980The Phonology and Phonetics of English IntonationPhD dissertation. Cambridge, MA. MITThesis
  • Pierrehumbert, Janet B1980The Phonology and Phonetics of English IntonationPhD dissertation. Cambridge, MA. MITThesis
  • Pierrehumbert, Janet B1980The Phonology and Phonetics of English IntonationPhD dissertation. Cambridge, MA. MITThesis
  • Pierrehumbert, Janet B1980The Phonology and Phonetics of English IntonationPhD dissertation. Cambridge, MA. MITThesis
  • Pierrehumbert, Janet B1980The Phonology and Phonetics of English IntonationPhD dissertation. Cambridge, MA. MITThesis
  • Pierrehumbert, Janet B1980The Phonology and Phonetics of English IntonationPhD dissertation. Cambridge, MA. MITThesis
printreport errorcite