2014
The voice onset time of /p t k/ and the realisation of /r/ in English loanwords pronounced by native
speakers of Dutch
SANNE PLOEGSTRA S1957236
Master’s Thesis:
European Linguistics LTX998M20 Revised version
dr. Dicky Gilbers dr. Wolfgang Kehrein
7 July 2014
2
TABLE OF CONTENTS
1. INTRODUCTION ... 4
2. BACKGROUND SECTION ... 5
2.1 English loanwords in the Netherlands ... 5
2.2 Pilot study ... 5
2.3 Influential factors on VOT production ... 7
2.3.1 Non-linguistic factors ... 7
2.3.1.1 Gender ... 7
2.3.1.2 Age ... 7
2.3.1.3 Language background ... 8
2.3.2 Linguistic factors ... 9
2.3.2.1 Context ... 9
2.3.2.2 Stress ... 10
2.3.2.3 Phonetic context ... 10
3. RESEARCH ... 11
3.1 Research question and sub questions ... 11
3.2 Hypotheses and expectations ... 12
3.3 Method ... 13
3.3.1 Participants ... 13
3.3.2 Materials ... 14
3.3.3 Procedure... 19
4. RESULTS ... 20
4.1 VOT ...21
5. DISCUSSION ... 29
5.1 Review of the hypotheses ... 29
5.2 General interpretation of the results ... 29
5.3 Old vs. New ... 30
5.4 Frequency ... 31
5.5 Context ... 34
5.6 Model ... 35
6. CONCLUSION ... 37
WORKS CITED ... 38
APPENDICES ... 40-48
3
ABSTRACT
This paper presents the findings from an empirical study on the realisation of English loanwords by Dutch
natives. Thirty Dutch participants (aged between 18 and 56, and employed in the north of the
Netherlands) were asked to read aloud Dutch and English word lists and Dutch and English sentences in
which loanwords were incorporated. The aim of the study was to measure the VOT of the voiceless
plosives /p t k/ in stressed, initial, prevocalic position, and observe the realisation of initial /r/. Results
showed that in the English context, /r/ was realised as [ɹ] in almost all cases. In the Dutch context, most
words were pronounced with either [R] or [r]. However, some participants pronounced /r/ in the
loanwords as [ɹ]. This concerned loanwords which were either recently incorporated in a Dutch
dictionary, or not that frequent in the Dutch language. In addition, the results showed that Dutch natives
pronounce loanwords with a higher VOT when the loanword is placed in an English context. The analysis
of VOT in loanwords did not support two other hypotheses. Further analysis showed that the phonetic
context of the plosive had an influence on its VOT. Not only the following vowel had an impact, but also
the sonority of the sound after the vowel seemed to have an impact on the VOT of the initial plosive of
that same word. Further research should confirm that sonority has an influence on VOT of the initial
plosive, and to establish whether speaking rate, different placement of a word in a sentence, and syllable
length are also of influence on VOT.
4
1. INTRODUCTION
In September 2013, the Dutch food processing company HAK first aired a television commercial starring Dutch former football player Ronald Koeman. The commercial shows Koeman standing in an open field cooking food. At the end of the scene he says: “Dit is de mooiste [kɔːl] die ik ooit heb gemaakt” (“This is the best [kɔːl] I have ever made”). This sentence is ambiguous because [kɔːl] has two possible interpretations. First of all, [kɔːl] refers to the cabbage that he is cooking in the commercial (‘kool’ in Dutch), which means that he is saying that this is the best cabbage has ever cooked. However, [kɔːl]
could also refer to scoring a point in football, since the realisation of ‘goal’ (/ɡəʊl/) is [kɔːl], which would make him say that this is the best goal he ever scored. The makers of the commercial exploit the result of transfer, which is the influence of one’s first language on one’s second language. In Dutch, [ɡ] is an allophone of /k/ in the words ‘zakdoek’ and ‘bruikbaar’, for example, where /k/ becomes voiced ([ɡ ]) because it is followed by a voiced consonant. While it is an allophone, it is not a phoneme of the Dutch language, and therefore /ɡ/ (voiced velar plosive) in initial position is realised with the closest sound in their native language: [k] (voiceless velar plosive).
This commercial aroused interest in the realisation of English loanwords pronounced by Dutch natives. Is there a difference between the pronunciation of /r/ in the loanword ‘rally’, which was first included in a Dutch dictionary in 1940, and the recently loaned word ‘retail’? How do Dutch natives pronounce the initial voiceless plosives /p t k/ in loanwords such as ‘panty’, ‘tackle’, and ‘cap’ when compared to Dutch words with the same initial sounds? Is the pronunciation of loanwords different in Dutch and English contexts? These three questions were the reason to write this thesis on the voice onset time of /p t k/ and the realisation of /r/ in English loanwords pronounced by native speakers of Dutch.
Section 2.2 of this thesis describes a pilot study concerning the same topic. On the basis of the results and conclusion of this pilot study, the present study was conducted. This thesis starts with a background section (section 2), covering topics such as ‘English loanwords in the Netherlands’ (§2.1) and
‘Influential factors on VOT production’ (§2.3). This section is followed by a description of the empirical study (§3), after which the results are presented (§4) and interpreted in the discussion section (§5).
Finally, a conclusion is drawn (§6).
5
2. BACKGROUND SECTION
2.1 English loanwords in the Netherlands
The Dutch language has been influenced by many other languages in the past. The amount of loanwords is one of the indicators of such an influence. French and Latin have been of great influence on the Dutch language in the past, which resulted in loanwords such as ‘garage’ and ‘trottoir’ from French, and
‘museum and ‘diafragma’ from Latin. However, since the 19
thcentury, the Dutch language is more influenced by the English language. In fact, 1774 English loanwords were incorporated in Dutch dictionaries since 1801. To compare: in that same period, 641 loanwords from Latin origin were added to the Dutch vocabulary. (Van der Sijs, 2008).
The increasing amount of English loanwords is mainly due to technological developments. ICT innovations lead to the creation of a new piece of equipment, the computer. This new device could do all sorts of things that could never be done before. Thus, working with this new piece of equipment required new vocabulary. The first computer functioned in English, and so English ICT terminology quickly entered the vocabulary of many other languages. However, not all speakers decided to use the English terminology. Speakers of Frisian, for example, came up with Frisian terminology to prevent that their language is influenced by English. To download, homepage and backup were replaced with the Frisian words ‘ynlade’, ‘thússide’ and ‘reservekopy’ (Nortier, 2009). Another area in which English is used more often is the transport branch. For example, the name of the airport of Rotterdam was changed into Rotterdam-The Hague Airport, in which the Dutch city ‘Den Haag’ is called The Hague. In addition, the words gate and security don’t seem strange anymore when walking on a Dutch airport. Also, Dutch radio stations, films and television make use of the English language (Noord, 2013). Because of this great amount of exposure to the English language, Schogt (1997) believes that English loanwords adjust to Dutch phonology and morphology very slowly. English films and series are not dubbed in the Netherlands, they broadcast them with the original English audio. Therefore it is likely that Dutch natives imitate the English pronunciation as heard in English spoken films and series, and use this pronunciation in English loanwords.
2.2 Pilot study
A pilot study concerning the pronunciation of English loanwords by Dutch natives was conducted in 2013
(Ploegstra). In this study, the voice onset time (VOT) of the plosives /p t k/ in initial stressed prevocalic
6 position was measured. The VOT of these sounds (in initial stressed prevocalic position) is different in English and in Dutch. Native speakers of Dutch pronounce /p t k/ in this position unaspirated, whereas native speakers of English do aspirate these sounds. This results in a different voice onset time value.
Lisker & Abramson (1964) found the following VOT values for the voiceless plosives /p t k/ as pronounced by Dutch natives: /p/: 10 ms, /t/: 15, and /k/: 25 ms (Cho & Ladefoged, 1999). Because the sounds are aspirated in English, the VOT values of these sounds in the English language are higher. Yao (2009) claims that the VOT values of /p t k/ pronounced by native speakers of English are approximately between 40 ms and 100 ms, and Lisker & Abramson found the following values for VOT of the English language: /p
h/: 58 ms, /t
h/: 70, en /k
h/: 80 ms (Cho & Ladefoged, 1999). Table 1 below shows the difference between English and Dutch VOT. The relatively high values show that in English, it takes longer before the vocal folds start vibrating.
VOT /p/ /t/ /k/
Dutch 10 15 25
English 58 70 80
Table 1
This pilot study was conducted to determine whether the VOT values of Dutch natives were
different in the pronunciation of English loanwords, when compared to the Dutch VOT values as
mentioned above. Ten participants were asked to read aloud word lists consisting of English loanwords,
and sentences in which English loanwords were incorporated. The participant group consisted of four
women and six men, all native speakers of Dutch. The year of birth of the participants varied from 1962
to 1998, and contact with the English language varied from ‘none’ to ‘daily’. Their pronunciation was
recorded and analysed with PRAAT. However, no definite conclusions could be drawn from the data set
since there were too many factors that could have had an influence on the VOT production of the
participants. Literature should be reviewed thoroughly in order to determine the influential variables,
and to make sure these variables are eliminated in a new study and prevent them from having influence
on the results. Therefore, the section below will discuss several influential factors on one’s VOT
production.
7
2.3 Influential factors on VOT production 2.3.1 Non-linguistic factors
2.3.1.1 Gender
In the pilot study as described above, six of the ten participants were men and the other four were women. Literature review shows that there are a number of differences between men and women when it comes to pronunciation. Smyth & Rogers (2002) state that men’s and women’s speech is different in for example pitch and formant frequencies, but also in VOT. This was discovered by Swartz (1992) who investigated the role of gender in VOT production. Swartz’s study showed that men have shorter VOT’s than women in the production of the consonant stops /t/ and /d/.
These findings were confirmed by research conducted in 1997 by Whiteside & Irving, who tested and measured English plosives in stressed, initial, prevocalic position. Their research also showed that the women had longer VOT’s than the men who participated in their study.
2.3.1.2 Age
The participants of the previously mentioned pilot study were born somewhere between 1962 and 1998.
This means that the oldest participant was either 50 or 51 years old and the youngest participant was
either 14 or 15 years old. It needs to be determined whether this age range could have been the cause of
the unconvincing results of the pilot study. A review of literature showed that several studies have been
conducted to investigate the change in VOT production as one ages. Morris & Brown (1987; 1994)
investigated age-related differences in speech and voice between elderly and younger speakers. Their
results show that the voice onset time of old adults is shorter than the VOT of young adults. See figure 1
(adapted from Morris & Brown 1987, altered by me) for the different VOT values of /p/ and /t/ as
pronounced by older women and younger women.
8 Figure 1
2.3.1.3 Language background
Two of the ten participants of the pilot study were raised in both the Frisian language and the Dutch
language. The other eight participants were native speakers of Dutch. The question that arises here is: is
the VOT of monolinguals similar to the consonant production of bilinguals, or could the perhaps different
VOT values of bilinguals have influenced the results of the pilot study? Research by Fowler et al. (2009)
has shown that simultaneous English-French bilinguals produce VOT values that approach the VOT values
of the monolinguals speaking these language. However, the bilinguals’ VOT is still influenced by their
other native language. Figure 2 (adapted from Fowler et al., 2009) shows the VOT of monolingual
speakers of English and French, and the VOT of the English-French bilingual speakers. One can see that
the VOT values of the bilinguals approach the monolinguals’ VOT values. However, the VOT of the
bilinguals are somewhat shorter than the VOT of the English monolinguals, and somewhat longer than
the French monolinguals. The somewhat shorter VOT of bilinguals in English is caused by their
knowledge of the French language, and the somewhat longer VOT of bilinguals in French is caused by
their knowledge of the English language. This pattern is apparent in the VOT of all three voiceless
plosives /p t k/.
9
/p/ /t/ /k/
Figure 2
These findings support Khattab´s statement that “bilinguals are able to adapt their production mechanisms according to the systems of each language, but that signs of ‘interlanguage interference’
1are inevitable, usually from the strong or dominant language to the weaker” (Khattab, 2000, p.97). The bilingual participants of the pilot study were raised in Dutch and Frisian. With the latter being a minority language, it is expected that Dutch is the dominant language. Thus it is more likely that their Dutch influences their Frisian VOT production, instead of their Frisian interfering with their Dutch VOT production.
2.3.2 Linguistic factors
2.3.2.1 Context
Besides the abovementioned factors gender, age, and language background, several linguistic factors can influence VOT values as well. The empirical research in the abovementioned pilot study consisted of sentences and word lists. However, Lisker & Abramson (1967) report that the context of the word that is tested is also of great importance. They compared the VOT of isolated words with the VOT of words in sentences (of between eight and fifteen syllables each). The results show that the isolated words with initial voiceless plosives have higher VOT values than the words incorporated in ‘running-speech’. In fact, “the difference between mean values for /p t k/ in isolated words as against sentences is about 25
1
A term borrowed from second language acquisition (cf. Flege,1980).
10 msec [...]” (Lisker & Abramson, 1967, p.16). In addition, a previous study by Lisker & Abramson showed that “the mean VOT for word-initial /p/ in English was 30 ms shorter in running speech (fast) than in isolation (slow)” (1964, p.125). On the basis of these findings it can be said that the average VOT of the participants in the pilot study does not reflect the voice onset time properly, since it is the average of the words in sentences and the words in isolation. The loanwords should either have been compared as pronounced in isolation or as pronounced in a sentence, but they should not be combined.
2.3.2.2 Stress
Another factor that influences the VOT of the initial voiceless plosive is whether the syllable in which it is placed is stressed or unstressed. Various studies observed a longer voice onset time in stressed position, compared to unstressed position (Sundberg & Lacerda, 1998). However, Lisker & Abramson (1967) do mention that the difference between stressed and unstressed is rather small and only holds for the voiceless plosives /p t k/ in stressed position. The figure below, adapted from Lisker & Abramson (1967), shows the difference between VOT in stressed and unstressed stops, separating words in isolation and sentences.
Figure 3
Figure 3 shows that in both categories (words in isolation, and sentences) the stressed stops have higher voice onset time values than the unstressed stops. The difference between unstressed and stressed stops in sentences is 6 ms, whereas the difference between unstressed and stressed stops in words is as high as 24 ms.
2.3.2.3 Phonetic context
The abovementioned pilot study contained twelve English loanwords that were tested in several
contexts. The words were chosen on the basis of the date in which they were first incorporated in a
11 Dutch dictionary. The outcome of the study showed a relatively high average VOT for the words team and keeper. These words both have /iː/ as following vowel, and this seemed to cause the high VOT value.
Review of literature shows that the following vowel does have an effect on the VOT of the initial plosive.
Voiceless plosives have higher VOT values when the following vowel is a high closed vowel. When the following vowel is low and open, the VOT is relatively short (Morris et al. 2008, p. 315).
In addition, VOT differs along with place of articulation. “VOT increases when the point of constriction moves from the lips to the velum […]” (Yao, 2009, p.1-2). For example, a plosive produced with the back part of the tongue (velar), has a higher VOT value than sounds that are produced more to the front of the mouth, such as /t/ (alveolar). The alveolar plosive /t/ has, in turn, a higher VOT than the bilabial plosive /p/, with is produced with the lips.
3. RESEARCH
3.1 Research question and sub questions
The main focus of this study is to investigate the realisation of English loanwords by native speakers of Dutch. The pronunciation of English loanwords by native speakers of Dutch will be tested and measured.
The sounds that will be tested in this study are the initial stressed prevocalic consonants: /p t k r/. These consonants are pronounced differently in Dutch and in English. The consonants /p t k/ in initial stressed prevocalic position are aspirated in English ([p
ht
hk
h]) but unaspirated in Dutch ([p t k]). The time that it takes until the vocal folds vibrate to make the following vowel sound is called Voice Onset Time (VOT). The VOT of /p t k/ in Dutch is shorter than the pronunciation of /p t k/ in the same position in English. Lisker & Abramson (1964) found the following VOT values for the voiceless plosives /p t k/ as pronounced by Dutch natives: /p/: 10 ms, /t/: 15, and /k/: 25 ms (Cho & Ladefoged, 1999). Yao (2009) claims that the VOT values of /p t k/ pronounced by native speakers of English are approximately between 40 ms and 100 ms, and Lisker & Abramson found the following values for VOT of the English language: /p
h/: 58 ms, /t
h/: 70, en /k
h/: 80 ms (Cho & Ladefoged, 1999).
Table 1
VOT /p/ /t/ /k/
Dutch 10 15 25
English 58 70 80
12 The /r/ is realised as an alveolar trill [r], alveolar tap [ɾ], uvular trill [ʀ], or a uvular fricative[ʁ] in Dutch, depending on the regiolect that is used (Van Bezooijen, 2005, p. 17) . In English, /r/ in initial stressed prevocalic position is realised as an alveolar approximant [ɹ], or a trill [r], depending on the regiolect: the former is apparent in Standard American English (SAE), the latter in Scottish English. The alveolar approximant [ɹ] is apparent in Dutch, but not in initial stressed prevocalic position. The pronunciation of initial stressed prevocalic /r/ in the loanwords used in this study will be identified and compared to the participants’ pronunciation in Dutch words with /r/ as a stressed initial prevocalic sound.
The loanwords used in this study are selected on the basis of the date they were first included in a Dutch dictionary, and their frequency. In doing so, a division can be made between the pronunciation of recently loaned words and words that were loaned earlier on. For example, tanker was included in a Dutch dictionary between 1926 and 1950, whereas tagged was not included until 2008. The question that arises here is: is there a difference in the pronunciation of /t/ in tagged when compared to the pronunciation of /t/ in tanker?
Similarly, a division is made between frequent loanwords and words that are less frequent. For example, teak is less frequent in the Dutch language than the word team. The question that arises here is: is there a difference in pronunciation of the frequent word team when compared to the pronunciation of the less frequent word teak?
In addition, this study tests the influence of context on the pronunciation of loanwords. The English loanwords are put in a Dutch context and additionally put in an English context. For example, the word tackle is part of both English and Dutch. If this word is incorporated in a Dutch sentence would the pronunciation differ from the pronunciation of the same word in an English sentence?
3.2 Hypotheses and expectations
On the basis of the discussed literature in the literature section hypotheses are formed and discussed
below. The first sub question is: is there a difference in pronunciation between recently loaned words
and older loanwords? (1). It is expected that people are more familiar with the older loanwords and thus
the pronunciation would gradually change into the Dutch pronunciation. People could even be unaware
that some Dutch words originated from the English language. In contrast, recently loaned words are less
familiar to people and it is expected that they pronounce these words with a more native-like
pronunciation.
13 The second sub question is: is there a difference in the pronunciation of frequent when compared to less frequent loanwords (2)? It is expected that frequent loanwords are more evident in the Dutch language and therefore are considered as part of the Dutch language. It could be that participants consider an old, less frequent loanword part of the English language since they do not recognise it, and thus pronounce it more similar to the English language. At the same time they could consider a new, frequent loanword as part of Dutch language, and pronounce it more towards the Dutch pronunciation.
The third sub question focuses on the difference in pronunciation of words in an English context when compared to the pronunciation of the same words in a Dutch context (3). It is expected that the surrounding words in a sentence have an influence on the pronunciation of the loanword. This means that the pronunciation of loanwords in a Dutch context should be more similar to Dutch pronunciation, whereas the pronunciation of loanwords in an English context should be more similar to the English pronunciation.
3.3 Method
The above mentioned hypotheses of the research question and the sub questions will be tested with an empirical study. The sections below will describe the participants, the procedure, and the materials that were used in this study.
3.3.1 Participants
A previous pilot study by Ploegstra (2013) has shown that several factors need to be taken into account to be able to draw conclusions from a data set. Yao (2009) states that besides linguistic factors there are also many non-linguistic factors that are of influence on VOT, which include age, gender, and other physiological characteristics of the speaker (Yao, p. 1). Conducting the study with an homogeneous group eliminates the danger that gender, educational and language background might influence the results.
The participants that were asked to participate in this study are 30 males employed in the north
of the Netherlands. Their native language is Dutch, and they are aged between 18 and 56.
14
3.3.2 Materials
The loanwords were selected from the Dutch online dictionary Van Dale. The year in which the word was first incorporated in a Dutch dictionary was also taken from the Van Dale dictionary. The frequency numbers as mentioned in the table below were provided by Gosse Bouma from the University of Groningen, who compiled a frequency database containing approximately 133 billion words of Dutch web text (<http://www.let.rug.nl/gosse/bin/Web1T5_freq.perl>).
Five loanwords per sound /p t k r/ were selected from the dictionary. The first three listed are used to test the old vs. new loanword hypothesis (1), and the last two listed are used to test the frequent vs. less frequent hypothesis (2). As one can see in table 2, the first three words are dated differently. The pronunciation of e.g. patchwork, panty, and peppadew will show if there is a difference in pronunciation of the older loanword (patchwork) versus the more recent loanword (peppadew). The last two words listed are used to see if frequent words are pronounced differently from less frequent loanwords. For example, popcorn is an older loanword but less frequently used than the newer loanword podcast. The data analysis will show if the pronunciation of the last two words listed under each sound support the
hypothesis as mentioned in 3.2.
Table 2 Previous pilot research has shown that the following vowel influences the VOT of the preceding consonant. For example, /t/ in team has a longer VOT than /t/ in tablet because of the influence of /iː/
on /t/ in team. Therefore almost the same vowels are chosen to follow the initial consonants /p t k r/.
This holds for the first three words listed under /p t k r/. The last two words listed under /p t k r/ do not have the same following vowels as the first three words, but they can still be compared with each other.
The experiment consists of five parts, of which two parts contain Dutch words and sentences, and two parts contain English words and sentences. The last part consists of a questionnaire.
The first section contains a list of 30 Dutch words (see table 3), of which 20 words are words that contain the sounds: stressed initial prevocalic/p t k/, and stressed initial prevocalic /r/. The list also
/p/ /t/ /k/ /r/
word year word year word year word year
patchwork 1950 tanker 1926-1950 cash 1901-1925 racket 1898
panty 1973 tackle 1950 cap 1950 rally 1940
peppadew 2000 taggen 2008 camcorder 1982 ranking 1999
word frequency word frequency word frequency word frequency
popcorn (1950) 219505 teak (1854) 398597 cottage (1901-1925) 275331 remake (1956) 180078
podcast (2006) 1029167 team (1901-1925) 12362582 column (1969) 3383657 retail (2006) 1290710
15 contains 10 words that begin with any other sound so that the participants do not know that /p t k r/ are most important. The following vowels are also the same vowels as in the selected loanwords so that the VOT of the Dutch words with /p t k/ is not altered by different vowels.
maan
zon
pesten
keffen
post
kostte
pen
test
redden
tellen
tien
riempje
gaven
boos
zeker
pels
korte
tieren
mogen
kennen
tergend
rel
riet
gunnen
rest
zomer
kerven
16 Table 3
The second section of the experiment contains a list of 15 sentences in which the previously mentioned loanwords are incorporated in a Dutch context (see table 4 below).
1 Toen bij de ingang van de zaal de tas van Wilma haar camcorder werd gecontroleerd, viel haar popcorn op de grond.
2 Tijdens het feestje serveerde de ober onder andere kaviaar, peppadews, en andere chique hapjes.
3 Bij de kassa werd Peter verteld dat hij alleen cash kon betalen, omdat het pinapparaat stuk was.
4 Nadat de tennisser een rally van dertig slagen verloor, sloeg hij uit frustratie zijn racket kapot op de grond.
5 Lisanne stuurde haar vriendin het volgende berichtje: “Wil jij mij even taggen in die foto? Dan komt hij ook op mijn facebookpagina te staan”.
6 Na een val van zijn paard bleef de ruiter ongedeerd, zijn cap was echter wel wat beschadigd.
7 Het meisje dat een ladder in haar panty had, haastte zich nog snel naar de winkel om een nieuwe te kopen.
8 De voetballer was 3 maanden uit de running als gevolg van een keiharde tackle van iemand van het andere team.
9 Het Nederlandse voetbalteam hoort niet meer bij de best presterende voetballanden van de wereld en zakt steeds verder weg op de FIFA ranking.
10 Lisa vertelde in haar column dat ze een grote sprei met patchwork heeft gekocht die heel mooi past bij haar teak houten meubels.
11 Als gevolg van een lek in een tanker raakten veel vogels besmeurd met olie.
12 Veel Nederlandse radiozenders zijn tegenwoordig ook te beluisteren via een podcast.
13 De kostprijs van de Playstation 4 ligt dichter bij de retail prijs dan eerder is aangegeven.
gist
polsen
binnen
17 14 Dit jaar gaan we op vakantie naar een leuke cottage, gelegen op het platteland.
15 Wist je dat er volgend jaar een remake uitkomt van de film ‘Het Zakmes’?
Table 4
The third section of the experiment contains a list of 16 English sentences in which the same loanwords are mentioned (see table 5). Most sentences are similar to the Dutch sentences. However, some are simplified because the sentences should not be too complicated. The participants (native speakers of Dutch) should be able to understand the sentence and be able to pronounce the sentence in a natural pace.
1 Do you have a camera or a camcorder?
2 Do you like popcorn or do you want another kind of snack?
3 The waiter served caviar and peppadews at the party.
4 Peter had to pay in cash because the PIN was broken.
5 The tennis player smashed his racket on the ground after losing the rally.
6 Ann got the following message from Facebook “Lisa tagged you in a photograph”.
7 The horse rider’s cap broke during the fall.
8 The girl who found a hole in her panty rushed to the store to get a new one.
9 The football player got injured because of a tackle of someone from the other team.
10 The Dutch football team is no longer part of the top ten of the FIFA ranking.
11 Lisa said in her column that she bought a large patchwork quilt at the store, which goes well with her teak wooden furniture.
12 Many birds got black feathers due to a damaged oil tanker.
13 Many Dutch radio programs are available via podcast nowadays.
14 What is the retail price of the newest Playstation?
15 We will spend our holidays in a lovely cottage in the country side.
16 Did you know that there will be a remake of the movie ‘The Godfather’?
Table 5
In the fourth part of the experiment the participants are asked to read aloud a list of 30 English words, of
which 20 contain the same stressed initial prevocalic sound as the loanwords that are tested in this
experiment (see table 6). The other 10 words are randomly chosen English words.
18 palace
magic calculate tanning popping tea dark paradise cancel rapidly tease costs backward merry parent galaxy come reason cabbage rabbit tapping similar reach fire tangled gamble rash dash posh donkey
Table 6
19 The final part of the experiment consists of a questionnaire that is created to get some background information on the participant (see appendix I). The questionnaire consists of questions regarding: age, native language(s), language production/perception, and the amount of contact with the English language.
3.3.3 Procedure
First of all, the participant was asked if he gives permission for being recorded and if he gives permission that the recordings are used for research. if the participant gave his consent, the experiment started. The participant was told that the experiment consists of four parts, of which he is asked to read and speak Dutch in the first two, and read and speak English in the last two parts. In doing so, the participants were prepared for speaking a second language next to their native language. The experiment was constructed in such a way that the participants had some time to read the instruction above the word list, which either said “read the following Dutch words/sentences out loud” or “read the following English words/sentences out loud”, after which the recording was started. After each part, the participant was given the next list of words/sentences and the instructions. After the fourth part, the recording was stopped and the participants were asked to fill in the questionnaire, which was the last part of the experiment.
The words and sentences were recorded with a SONY ICD-PX333M digital voice recorder. The
recorder was placed on the table approximately 20-30 centimetres from the mouth of the speaker. The
recordings were analysed with the computer programmes Audacity 1.3, Adobe Audition, and PRAAT. The
dataset was collected with Microsoft Word and Microsoft Excel.
20
4. RESULTS
The results of the study are shown in appendices 2-5, each representing one part of the experiment. The second appendix (II) contains the results of the 30 participants’ pronunciation of 20 Dutch words in isolation, focusing on voice onset time (VOT). The third appendix (III) includes the VOT per participant of 20 English loanwords incorporated in Dutch sentences. Appendix IV shows the measured VOT per participant of the same 20 English loanword, incorporated in an English context. The fifth (V) appendix contains the VOT of 20 English words in isolation.
All four appendices have the same layout, which shows the participant numbers (P1-P30) horizontally and the words that were tested in the first column vertically. The light grey coloured rows starting with ‘AVG’ show the average voice onset time per participant of the words starting with the same initial sounds. For example, participant one (P1) has an average VOT of 16 ms when words starting with /pɛ/ are pronounced. The last couple of rows show the pronunciation of initial /r/. The results are marked either R, r, or ɹ, which stand for uvular trill/fricative (R), alveolar trill/tap (r), and alveolar approximant (ɹ). A transcription was added in parentheses in cases where the actual pronunciation was different from the proper pronunciation. For example, in appendices III and IV with the results of English loanwords incorporated in sentences, the word teak (/tiːk/) was occasionally pronounced as [teɪk]. The pronunciation of <ea> as [eɪ] influences the VOT of the preceding plosive. Some VOT numbers are followed by ‘(w)’, which means that this word has a weak first syllable and thus the initial plosive was unstressed. For example, the words retail and cottage were once pronounced as [ri’teɪl] and [kɒ’taːƷә].
The change of stress placement may have affected the VOT of the initial plosive too as it does affect the duration of the vowel following the initial plosive.
The participants’ answers from the questionnaire, the fifth part of the research, are shown in appendix VI. The first three columns list their participant number, their age, and their native language.
The next column shows where the participants have lived and for how long. The last three columns show the answers to the questions ‘Do you have knowledge of other languages, and how well do you speak/read/write/understand these languages?’, ‘Did you have English education in primary/secondary school?’, and ‘Do you have contact with English natives at work/outside of work?’.
In total, 30 participants took part in the research. The youngest participant was 18 years old at
the time of the recordings, and the oldest participant was 56 years old. Five of the 30 participants have
lived abroad for some time. Six participants were bilingually raised, of which two were raised
21
‘Dutch/German’, three were raised ‘Dutch/Frisian’, and one participant was raised in both the Dutch and the English language.
Section 4.1 shows the results in graphs. The four graphs below show the average VOT in ms of all participants. The x-axis shows the duration in ms, and the y-axis shows the sounds that were focused on:
the initial, stressed, prevocalic plosives /p/, /t/, /k/. The y-axis not only shows the initial plosive, but it also shows the following vowel sound. The four graphs represent the average VOT in the Dutch word list, the loanwords in Dutch sentences, the loanwords in the English word list, and the loanwords in English sentences.
4.1 VOT
G raph 1
0 5 10 15 20 25 30 35 40
/ko/
/ke/
/ti/
/te/
/po/
/pe/
/ko/ /ke/ /ti/ /te/ /po/ /pe/
Average VOT (ms) of all participants
in Dutch Word list 36 33 38 26 17 14
Average VOT (ms) of all participants in Dutch Word list
22 Graph 2
Graph 3
0 5 10 15 20 25 30 35
/ko/
/ke/
/ti/
/te/
/po/
/pe/
/ko/ /ke/ /ti/ /te/ /po/ /pe/
Average VOT (ms) of all participants
in Dutch sentences 30 29 33 23 15 12
Average VOT (ms) of all participants in the pronunciation of the loanwords in Dutch sentences
0 10 20 30 40 50 60
/ko/
/ke/
/ti/
/te/
/po/
/pe/
/ko/ /ke/ /ti/ /te/ /po/ /pe/
Average VOT (ms) of all
participants in English Word list 41 37 49 31 16 16
Average VOT (ms) of all participants in English Word list
23 G raph 4
The following graph shows the difference between the participants’ pronunciation of Dutch and English words starting with /p t k/ and the different following vowel sounds.
Graph 5
0 5 10 15 20 25 30 35 40 45 50
/ko/
/ke/
/ti/
/te/
/po/
/pe/
/ko/ /ke/ /ti/ /te/ /po/ /pe/
Average VOT (ms) of all participants
in English sentences 37 34 46 30 14 14
Average VOT (ms) of all participants in the pronunciation of the loanwords in English sentences
/pe/ /po/ /te/ /ti/ /ke/ /ko/
Dutch words 14 17 26 38 33 36
English words 16 16 31 49 37 41
0 10 20 30 40 50 60
A sti te l
Average VOT (ms) of all participants in Dutch and English word list
24 The graph below illustrates the difference in the pronunciation of English loanwords and Dutch in the same words in English context.
Graph 6
Graph 7 contains the average VOT of all participants in all four parts of the experiment: the pronunciation of Dutch and English words, and the pronunciation of /p t k/ in the loanwords in Dutch and English context.
Graph 7
/pe/ /po/ /te/ /ti/ /ke/ /ko/
Dutch sentences 12 15 23 33 29 30
English sentences 14 14 30 47 34 37
0 5 10 15 20 25 30 35 40 45 50
A sti te l
Average VOT (ms) of all participants in Dutch and English sentences
/pe/ /po/ /te/ /ti/ /ke/ /ko/
Dutch words 14 17 26 38 33 36
English words 16 16 31 49 37 41
Dutch sentences 12 15 23 33 29 30
English sentences 14 14 30 47 34 37
0 10 20 30 40 50 60
VOT (m s)
Average VOT (ms) of all participants in Dutch and English word
list, and Dutch and English sentences
25 The graph below shows the average VOT of the participants’ pronunciation of the loanwords arranged by date.
Graph 8
The following graph depicts the difference in pronunciation between frequent loanwords and less frequent loanwords.
0 5 10 15 20 25 30 35 40
camcorder (1982) cap (1950) cash (1901-1925) taggen (2008) tackle (1950) tanker (1926-1950) peppadew (2000) panty (1973) patchwork (1950)
camcorde
r (1982) cap (1950) cash (1901- 1925)
taggen (2008)
tackle (1950)
tanker (1926-
1950)
peppade w (2000)
panty (1973)
patchwor k (1950)
English sentences 34 31 37 30 28 34 16 14 12
Dutch sentences 29 28 32 23 24 21 14 12 12
Average VOT (ms) per word in Dutch and English sentences
26 Graph 9
0 10 20 30 40 50 60
cottage (less frequent) column (frequent) teak (less frequent) team (frequent) popcorn (less frequent) podcast (frequent)
cottage (less frequent)
column (frequent)
teak (less frequent)
team (frequent)
popcorn (less frequent)
podcast (frequent)
English sentences 37 38 45 50 14 15
Dutch sentences 29 32 32 34 14 16
Average VOT (ms) per word in Dutch and English sentences
27 T he two tables below show the results per participant of their pronunciation of /r/ in initial prevocalic position in Dutch words and in the loanwords in Dutch context. The results are marked either R, r, or ɹ, which stand for uvular trill/fricative (R), alveolar trill/tap (r), and alveolar approximant (ɹ).
The participant numbers are shown vertically, and the words that were tested are shown horizontally.
Table 7 Table 8
The two tables below show the results per participant of their pronunciation of /r/ in initial prevocalic position in English words and in the loanwords in English context. The participant numbers are shown vertically, and the words that were tested are shown horizontally.
DUTCH WORDS redden rel rest riempje riet
P1 R R R R R
P2 R R R R R
P3 R R R R R
P4 r r r r r
P5 R R R R R
P6 R R R R R
P7 r r r r r
P8 r r r r r
P9 r r r r r
P10 r r r r r
P11 r r r r r
P12 r r r r r
P13 r r r r r
P14 R R R R R
P15 r r r r r
P16 r r r r r
P17 r r r r r
P18 r r r r r
P19 R R R R R
P20 r r r r r
P21 r r r r r
P22 r r r r r
P23 R R R R R
P24 R R R R R
P25 r r r r r
P26 r r r r r
P27 R R R R R
P28 r r r r r
P29 r r r r r
P30 r r r r r
DUTCH SENTENCES rally racket ranking retail remake
P1 R R R R R
P2 ɹ R ɹ R R
P3 R R R R R
P4 r r ɹ ɹ ɹ
P5 R R R ɹ ɹ
P6 R R R R R
P7 r r r r r
P8 r r r r r
P9 r r r r r
P10 r r r r r
P11 r r r r r
P12 r r r r r
P13 r r ɹ ɹ ɹ
P14 R R R R R
P15 r R R R ɹ
P16 r r r r r
P17 r r r r r
P18 r r r r r
P19 R R R R R
P20 r r ɹ ɹ ɹ
P21 r r r r r
P22 r r ɹ r r
P23 R R R R R
P24 R R R R R
P25 r r r r r
P26 r r r ɹ ɹ
P27 R R R R R
P28 r r r r r
P29 r r r r ɹ
P30 r r r r r
28
Table 9 Table 10
The following table shows the pronunciation of initial, stressed, prevocalic /r/ in older loanwords and recent loanwords:
Table 11
The following table shows the pronunciation of initial, stressed, prevocalic /r/ in frequent loanwords and less frequent loanwords:
Table 12 DUTCH CONTEXT ENGLISH CONTEXT racket (1898) 0 x [ɹ], 11 x [R], 19 x [r] all [ɹ]
rally (1940) 1 x [ɹ], 9 x [R], 20 x [r] all [ɹ]
ranking (1999) 5 x [ɹ], 10 x [R], 15 x [r] all [ɹ]
DUTCH CONTEXT ENGLISH CONTEXT retail (frequent) 5 x [ɹ], 10 x [R], 15 x [r] 29 x [ɹ], 1 x [r]
remake (less frequent) 7 x [ɹ], 9 x [R], 14 x [r] all [ɹ]
ENGLISH SENTENCES rally racket ranking retail remake
P1 ɹ ɹ ɹ ɹ ɹ
P2 ɹ ɹ ɹ ɹ ɹ
P3 ɹ ɹ ɹ ɹ ɹ
P4 ɹ ɹ ɹ ɹ ɹ
P5 ɹ ɹ ɹ ɹ ɹ
P6 ɹ ɹ ɹ ɹ ɹ
P7 ɹ ɹ ɹ ɹ ɹ
P8 ɹ ɹ ɹ ɹ ɹ
P9 ɹ ɹ ɹ ɹ ɹ
P10 ɹ ɹ ɹ
rɹ
P11 ɹ ɹ ɹ ɹ ɹ
P12 ɹ ɹ ɹ ɹ ɹ
P13 ɹ ɹ ɹ ɹ ɹ
P14 ɹ ɹ ɹ ɹ ɹ
P15 ɹ ɹ ɹ ɹ ɹ
P16 ɹ ɹ ɹ ɹ ɹ
P17 ɹ ɹ ɹ ɹ ɹ
P18 ɹ ɹ ɹ ɹ ɹ
P19 ɹ ɹ ɹ ɹ ɹ
P20 ɹ ɹ ɹ ɹ ɹ
P21 ɹ ɹ ɹ ɹ ɹ
P22 ɹ ɹ ɹ ɹ ɹ
P23 ɹ ɹ ɹ ɹ ɹ
P24 ɹ ɹ ɹ ɹ ɹ
P25 ɹ ɹ ɹ ɹ ɹ
P26 ɹ ɹ ɹ ɹ ɹ
P27 ɹ ɹ ɹ ɹ ɹ
P28 ɹ ɹ ɹ ɹ ɹ
P29 ɹ ɹ ɹ ɹ ɹ
P30 ɹ ɹ ɹ ɹ ɹ
ENGLISH WORDS rapidly rabbit rash reason reach
P1 ɹ ɹ ɹ ɹ ɹ
P2 ɹ ɹ ɹ ɹ ɹ
P3 ɹ ɹ ɹ ɹ ɹ
P4 ɹ ɹ ɹ ɹ ɹ
P5 ɹ ɹ ɹ ɹ ɹ
P6 ɹ ɹ ɹ ɹ ɹ
P7 ɹ ɹ ɹ ɹ ɹ
P8 ɹ ɹ ɹ ɹ ɹ
P9 ɹ ɹ ɹ ɹ ɹ
P10 ɹ
r rɹ ɹ
P11 ɹ
rɹ ɹ ɹ
P12 ɹ ɹ ɹ ɹ ɹ
P13 ɹ ɹ ɹ ɹ ɹ
P14 ɹ ɹ ɹ ɹ ɹ
P15 ɹ ɹ ɹ ɹ ɹ
P16 ɹ ɹ ɹ ɹ ɹ
P17 ɹ ɹ ɹ ɹ ɹ
P18 ɹ ɹ ɹ ɹ ɹ
P19 ɹ ɹ ɹ ɹ ɹ
P20 ɹ ɹ ɹ ɹ ɹ
P21 ɹ ɹ ɹ ɹ ɹ
P22 ɹ ɹ ɹ ɹ ɹ
P23 ɹ ɹ ɹ ɹ ɹ
P24 ɹ ɹ ɹ ɹ ɹ
P25 ɹ ɹ ɹ ɹ ɹ
P26 ɹ ɹ ɹ ɹ ɹ
P27 ɹ ɹ ɹ ɹ ɹ
P28
rɹ ɹ ɹ ɹ
P29 ɹ ɹ ɹ ɹ ɹ
P30 ɹ ɹ ɹ ɹ ɹ
29
5. DISCUSSION
5.1 Review of the hypotheses
This study focuses on the pronunciation of /p t k r/ in initial, stressed, prevocalic position, and the pronunciation of these sounds in English loanwords. The following hypotheses were formed.
First of all, it is expected that the pronunciation of older loanwords differs from the pronunciation of words that were recently loaned from the English language. The participants probably articulate recently loaned words in their English pronunciation, while older loanwords are pronounced more towards the pronunciation of native Dutch words. The second hypothesis is that there is also a difference in the pronunciation of /p t k r/ between frequent and less frequent loanwords. Because of the exposure to frequent loanwords, it is expected that these loanwords are quickly adapted to the Dutch pronunciation, whereas it is expected that the less frequent loanwords still appear foreign to most Dutch natives and thus they articulate these words in their English pronunciation. In addition, it is expected that the participants adapt their pronunciation of loanwords to the language of the text they are reading. In order to answer the sub questions and confirm the abovementioned hypotheses, the results need to be reviewed.
5.2 General interpretation of the results
The graphs 1-9 in the results section confirm the literature findings in section 2.3.2.3. In general, the VOT values increase when the place of articulation moves from the front of the mouth (lips) to the back of the mouth (velum). The words with /p/ as initial plosives have lower VOT values than words starting with /k/.
In addition, the graphs 1-7 & 9 show a clear peak in the VOT with words starting with /tiː/. These findings support Klatt’s study as described in section 2.3.2.3. Klatt found higher VOT values when the following vowel is a high closed vowel (such as /iː/), and lower VOT values when the following vowel is a low open vowel (such as /ɑː/).
Looking at graphs 1-4 of the results section, it stands out that words starting with /k/ followed by
the more open vowel /ɔ/ have consistently longer VOT values than /k/ followed by /ɛ/. The same holds
for the bilabial plosive /p/: the average VOT is higher when followed by /ɔ/, and relatively lower when
followed by /ɛ/. These examples seem to indicate a pattern: /pɔ/ and /pɛ/ both start with the same
plosive /p/, and in both cases the following vowels are produced in the middle of the mouth. However,
30 /ɔ/ is produced in the back of the mouth, whereas /ɛ/ is produced in the front of the mouth. Therefore it can be concluded that when the initial plosives are the same, the place of articulation of the following vowel has an influence on the VOT of the initial plosive.
Graph 7 shows the average VOT of all participants in Dutch and English words, and in Dutch and English context. This graph shows a constant difference between the VOT of Dutch or English words and the VOT of words incorporated in either a Dutch or an English context. Dutch words in isolation have always higher VOT values than Dutch words in sentences. The same can be said for the English language:
English words in isolation have higher VOT values than the words incorporated in an English context.
These results support the findings of Lisker & Abramson (1964;1967) as reviewed in the literature of the background section 2.3.2.1.
5.3 Old vs. New
In order to confirm the first hypothesis, the VOT of /p t k/ in the older loanwords should be relatively low, and the VOT of /p t k/ in recent loanwords should be relatively high. In addition, the data would support the hypothesis when the older words with initial /r/, which were pronounced [ɹ] in English before the word was borrowed, are now pronounced either [r] or [R] in the Dutch word list.
Looking at table 11 of the results section, which shows the pronunciation of initial /r/ in Dutch and English context, one could say that the data supports the hypothesis. For the reason that in Dutch the newest loanword ranking is still pronounced with [ɹ] by five people, the somewhat older loanword rally is only pronounced with [ɹ] once, and the oldest loanword of the three is pronounced with either [r]
or [R]. These findings show that the English pronunciation of /r/ gradually changes into a Dutch pronunciation of /r/. The results of the pronunciation of /r/ in English context shows that all participants recognise the words as part of the English language when it is placed in an English context.
Graph 8 of the results section illustrates the average VOT of /p t k/ in the participants’
pronunciation of the older and newer loanwords in both Dutch and English context. As one can see, the
graph shows a diverse pattern: the expected pattern is only apparent in the pronunciation of loanwords
with initial /p/ in Dutch and English context (patchwork-panty-peppadew). Loanwords with initial /t/ or
/k/ show a different pattern. In Dutch context tackle seems to have a higher VOT, whereas in English
context tanker has a higher VOT value. The results of initial /k/ shows that cash, the oldest loanword of
the three, has the highest VOT in both Dutch and English context, whereas cap has the lowest VOT. This
pattern does not support the hypothesis as mentioned in section 3.2.
31 There are several possible explanations for this irregular pattern. In the case of cash and cap, both initial plosives and following vowels are the same. The only difference is the final consonant sound;
/ʃ/ in cash and /p/ in cap. It appears as if the following vowels could not have caused a different VOT for both words, since the following vowels are the same. However, research has shown that the duration of the vowel can also be influenced by the final consonant. The vowels before fricatives tend to be longer than vowels before stops (Naeser, 1970; Hewlett et al. 1999). Moreover, Port & Rotunno (1979) found a relation between VOT and vowel duration, namely, VOT tends to be longer before long vowels. Thus, it could be possible that the final consonants have influenced the vowel duration and that the duration of the vowel influenced the VOT of the initial plosive, which results in a higher VOT for cash than for cap. As for tackle and tanker, the initial plosive and the following vowel are the same. However, the sound in coda position in tanker, [ŋ], is a nasal sound, and /tɛ/ in tackle is an open syllable. Other literature needs to be reviewed to determine whether a nasal sound or an open syllable affect the VOT of the initial sound.
Another possible explanation for the irregular pattern could be that the participants had to read out sentences they had never seen before. New sentences could lead to problems with reading or speaking because it is the first time that they see that particular sentence. This could have influenced the pronunciation of the participants, and thus could have influenced the VOT value. In addition, the sentences that were used were not similar to one another. The loanwords that were tested did not have the same place in a sentence. For example, in the English context the word tanker was placed at the end of the sentence, whereas tackle was incorporated in the middle of a sentence. Further research should point out whether these factors could have had an influence on VOT, and if this could explain the different VOT values.
Perhaps the frequency of these loanwords could have had an influence on the results as well. As can be seen in graph 8, tanker is a slightly older loanword than tackle, and cash is older than cap. It could be the case that the VOT production doesn´t correlate with the year in which the word was first incorporated, but with the frequency of the words. If frequency is of influence on VOT will be discussed in the next section.
5.4 Frequency
Table 12 of the results section shows the pronunciation of initial /r/ in the frequently used loanword
retail, and the less frequently used word remake. The less frequently used word is always pronounced
with [ɹ] in English context, which seems to indicate that the participants are less familiar with this word.
32 The frequently used word retail, was pronounced with [ɹ], except for one time when it was pronounced [r]. Looking at the frequently used word in Dutch context, one can see that it was pronounced with [ɹ]
five times, whereas the less frequent word was pronounced with [ɹ] seven times. In other words, the frequent word, which is more often used in the Dutch language, was more often pronounced in a Dutch pronunciation. Moreover, the less frequent word was more often pronounced with the English pronunciation which suggests that people are less familiar with the word and thus more often apply the English pronunciation.
Graph 9 shows the difference in pronunciation of a frequent loanword and a less frequent loanword. One would expect the frequent loanwords to adapt to the Dutch pronunciation quite quickly, and the less frequent loanwords to remain ’foreign’ to Dutch natives . However, the graphs show a different pattern than was expected. In both Dutch and English contexts, the frequent words have a higher VOT value than the less frequent loanwords. This pattern is apparent in all three initial plosive sounds. The loanword podcast has a higher VOT than popcorn, team has a higher VOT than teak, and column has a higher VOT value than cottage. As mentioned above, research by Naeser (1970) and Hewlett et al. (1999) has shown that the duration of the vowel (and thus the VOT of the initial plosive) can also be influenced by the final consonant of a CVC syllable. Looking at the final consonants of the first syllables of these words, one can see that /pɔd/, with final voiced plosive /d/ results in a higher VOT than /pɔp/ with final voiceless plosive /p/. Also, word initial /t/ has higher VOT in /tiːm/ than in /tiːk/, and /kɔl/ has higher VOT for initial /k/ than /kɔt/. Interestingly, voiceless plosives following the vowel in this first syllable seem to induce lower VOT values than nasals and liquids. It seems as if the sonority of the consonant following the vowel affects the VOT of the initial consonant. The sonority scale as shown below (Giegerich, 1992, p. 133) runs from stops to low vowels, with stops being the least sonorous sounds, and low vowels the most sonorous.
Figure 4
33 If the VOT of an initial plosive was indeed (indirectly) affected by the sonority of the following consonant when the vowel sounds following the initial consonants are similar, this would mean that, for example /pɛ/ followed by /l/ (a liquid) would result in a higher VOT for initial /p/ than when /pɛ/ is followed by /s/. This was tested with the word lists that were recorded in this study. The table below shows the tested word lists in Dutch and English, with the words ordered according to sonority category of the consonant following the initial consonant and vowel, and shows the corresponding average VOT values.
Table 13 As one can see in the table, the words with the most sonorous consonants following the first two sounds (initial plosive, and following vowel), have the highest VOT values when compared to words with less sonorous consonants in that same position. In Dutch words starting with /tɛ/, one can see a great difference in VOT between tellen and tergend. One would expect these words to have somewhat similar VOT values because /l/ and /r/ are both liquids. However, it seems as if /x/ in tergend, which is a fricative, levels the characteristics of /r/ to that of the fricative /x/. This results in an average VOT for tergend which approaches the average VOT for test with /s/ also being a fricative, and not approaching the VOT for tellen with /l/ being a liquid. In the English word list, tea has a higher VOT than tease which shows that no consonant, and thus an open syllable is more sonorous than the fricative /s/ in tease.
Dutch Word List Sonority Category AVG VOT of all participants (ms) English Word List Sonority Category AVG VOT of all participants (ms)
pels liquid 14,3 palace liquid 16,2
pen nasal 13,7 paradise liquid 15,9
pesten fricative 12,9 parent liquid 16,6
polsen liquid 17,7 posh fricative 17,4
post fricative 16,3 popping stop 14,0
tellen liquid 29,9 tanning nasal 32,5
tergend liquid+fricative 25,0 tangled nasal 31,5
test fricative 24,5 tapping stop 30,1
tieren liquid 40,3 tea_ open syllable 50,5
tien nasal 36,2 tease fricative 47,7
kerven liquid 34,7 calculate liquid+stop 37,1
kennen nasal 33,8 cancel nasal+fricative 38,1
keffen fricative 30,0 cabbage stop 34,7
korte liquid 37,6 come nasal 41,4
kostte fricative 34,6 costs fricative 39,7
34 However, literature should be reviewed to determine whether a consonant cluster following the vowel affects the VOT of the initial consonant, since this is the case in many of the words in table 13.
5.5 Context
The third question that this study tries to answer is whether participants pronounce a word in a different way when it is put in a different context. Therefore, English loanwords were selected and put in both Dutch and English context. Tables 7-12 of the results section show the data retrieved concerning the pronunciation of initial /r/ in loanwords. It was expected that the context would indeed have an influence on the pronunciation of initial /r/ in loanwords. In other words, it is expected that initial /r/ is pronounced as either [r] or [R] in the loanwords in Dutch context, and initial /r/ in loanwords in English context is realised as [ɹ].
The results as shown in table 11 confirm the expectation, since in an English context the loanwords racket, rally, ranking, are pronounced with [ɹ]. Table 12 shows that all but one participant pronounce the loanwords retail, and remake with initial [ɹ]. However, the loanwords in a Dutch context have a more varied pronunciation. Of 150 recordings (30 participants each pronouncing the words racket, rally, ranking, retail, and remake), 18 occurrences of initial [ɹ] were evident. These data show that the participants almost unanimously use [ɹ] in English context, and that there are nine participants who use [ɹ] in Dutch context. A possible explanation for the use of [ɹ] in Dutch context by Dutch participants is that they were not familiar with the word and thus the word seemed English-like to them.
This study also looked into the difference in VOT between the loanwords in Dutch and English context. Graphs 5 and 6 show a difference in the VOT production of loanwords in Dutch and in English.
The average VOT of words in English is higher than the VOT of words pronounced in Dutch. This supports
the hypothesis that speakers adapt their pronunciation to the context the word is in. However, one
remarkable result can be seen in graphs 5 and 6: the VOT of Dutch words starting with /pɔ/ is higher than
VOT of English words, whereas the other initial sounds show a different pattern. The same pattern holds
for the loanwords with /pɔ/: the loanwords have shorter VOT in English sentences, than in Dutch
sentences. This could not have been caused by the length of the word, since both podcast and popcorn
are bisyllabic. In addition, both words were tested in the middle of a sentence, so it is not the position of
the word that has had an influence on the VOT of the initial sound. A possible explanation for the
difference of VOT in isolated words starting with /pɔ/ could be that the participants were unfamiliar with
the words that were chosen for this list: posh and popping both seem difficult words for non-native
speakers of English. Their hesitance during the reading of the sentence could have resulted in a shorter
35 VOT, when compared to the Dutch words . However, the same pattern is apparent in the pronunciation of the loanwords popcorn and podcast in Dutch and English context. If these words would have been influenced by the hesitance of the speaker, then the pronunciation of the loanword in the Dutch sentence should have had a shorter VOT value too. For the reason that the participants first saw and read the loanword in Dutch context, and by the time they read the same word in an English context, they would have been more familiar with how to pronounce it.
This phenomenon can be explained by means of the abovementioned theory regarding the influence of the sonority of the consonant following /pɔ/. Looking at the final consonants of the first syllables of the words popcorn and podcast, one can see that /pɔd/, with final voiced plosive /d/ results in a higher VOT than /pɔp/ (graph 9). In the abovementioned sonority scale by Giegerich, voiceless plosives are least sonorous, whereas voiced plosives are slightly more sonorous. The voiced characteristic of /d/ as opposed to /p/ lengthens the VOT of initial /p/, whether it is at once, or /d/
influences /ɔ/ which in turn influences the VOT of /p/.
5.6 Model
On the basis of the findings in this study, the following model was created for comparing words with initial /p t k/
2. The literature review has shown that voice onset time depends on the place of articulation, the height of the following vowel, and the place of the following vowel: if words have the same initial voiceless plosive, then the following vowel shows which word has the higher VOT. Level three of the model is created on the basis of the findings in this study. Besides place of articulation of the initial plosive, and vowel height of the following vowel, sonority of the postvocalic consonant seemed to be a third factor in the phonetic environment that has an influence on VOT of the initial plosive. The word lists were analysed and the words were ordered according to the sonority of the consonant following the initial plosive and the vowel sound. As a result, the words with the least sonorous consonants in the position after the initial plosive and following vowel corresponded with the relatively lower VOT value, and the most sonorous consonants in this position corresponded with the relatively
2
Note that this model is created on the basis of the findings in this study, which focused on /p t k/,
therefore other initial plosives are not included.
36 higher VOT value
3. Further research is needed to determine what factors influence VOT of the initial plosive when the first three sounds are similar in a CVC syllable.
Figure 5
3
