Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
UNIVERSITAT POMPEU FABRA
Màster en Lingüística Teòrica i Aplicada
Treball fi de màster
Loanword Phonology: Semantic Constraints on
Tone Adaptation of English Loanwords in
Mandarin
Keqiao Cheng
Dirigit: Pilar Prieto
Barcelona, July, 2013
Abstract
Previous studies have shown that loanwords stressed syllables usually receive a
predictable high tone in Cantonese (Silverman 1992, Lai et al. 2011, Davis et al.
2012). However, Mandarin as a language which derived from the same origin of
Cantonese, it shows a different pattern and the loanwords stressed syllables do not
always receive the high tone. In this present study, we claim that in the process of
adapting English loanwords in Mandarin, besides the phonological factors, there are
also semantic constraints in the choice of appropriate Chinese characters for
transcription. Study 1 is a dictionary-based qualitative analysis which aimed at finding
out the distribution of different tones that English loanwords stressed syllables receive
in Mandarin. Study 2 is a rating experiment which aimed at testing the Mandarin
native speakers’ intuitive choice of Chinese characters for the transcription of
nonsense loanwords by controlling the target tones and semantic conditions in the
materials. Twenty-four Mandarin native speakers were asked to score the degree of
acceptance of each option. In each question, three homophonic Chinese characters
with different meanings were given for the transcription of the stressed syllable of a
nonsense English word with a made-up definition. The results showed that just like
Cantonese, if no semantic constraint appears, there is a clear preference for the high
tone, and we can identify a preference scale for the adaptation of English loanwords.
When there is no option of the high tone, the hierarchy from high to low is Tone 2,
Tone 4 and Tone 3. However, once the semantic constraints appear, the results do not
show a regular pattern and thus it seems that the tone adaptation of a large amount of
loanwords stressed syllables remains unpredictable in Mandarin.
Table of Contens
1. Introduction ................................................................................................................................... 1
2. Study 1 .......................................................................................................................................... 5
3. Study 2 .......................................................................................................................................... 9
3.1 Goals ................................................................................................................................... 9
3.2 Methods ............................................................................................................................... 9
3.2.1 Participants ............................................................................................................. 10
3.2.2 Materials ................................................................................................................. 10
3.2.3 Procedure ............................................................................................................... 12
3.3 Results ............................................................................................................................... 13
4. Conclusion .................................................................................................................................. 17
References ....................................................................................................................................... 19
Appendix I ...................................................................................................................................... 20
Appendix II ..................................................................................................................................... 22
Appendix III .................................................................................................................................... 30
Appendix IV .................................................................................................................................... 32
1
1. Introduction
Few research papers that dealt with Cantonese (Silverman 1992, Lai et al. 2011,
Davis et al. 2012) concluded that loanwords stressed syllables receive a predictable
high tone when it is adapted in Cantonese. For instance, both primary and secondary
stresses of English words receive a high tone in Hong Kong Cantonese (Davis et al.
2012) as shown in (I):
(I) ˈsaxoˌphone – [sik(55) si:(22) fuŋ(55)]1
Cantonese and Mandarin derived from the same origin, they share a similar tonal
system and a same script system, i.e. the Chinese characters, and thus I started with
the assumption that such conclusion from Cantonese should also work with loanwords
in Mandarin. Yet in Mandarin, from my personal observation, there are a lot of
exceptions to this rule.
(II) Examples of English loanwords in Mandarin
Original
word
Pinyin2
transcription
Tones Original
word
Pinyin
transcription
Tones
´humor you.mo 55.51 ´shock xiu.ke 55.51
´radar lei.da 35.35 ´jeep ji.pu 35.214
´TOEFL tuo.fu 55.35 ´yuppies ya.pi.shi 214.35.51
hys´teria xie.si.di.li 55.55.214.214 ´vitamin wei.ta.ming 35.55.51
´marathon ma.la.song 214.55.55 ´quark kua.ke 55.51
´cool ku 51 ´ounce ang.si 51.55
´poker pu.ke 55.51 gui´tar ji.ta 35.55
1 Traditionally, the numbers from 1 to 5 are used to represent the pitch from low to high in phonology of Chinese
languages. Both in Mandarin and Cantonese, the Tone 1(henceforth T1) is a high tone (55). 2 Pinyin is the official phonological system for transcribing the sound of Chinese characters into Latin script in
China.
2
´gene ji.yin 55.55 ´microphone mai.ke.feng 51.51.55
´disco di.si.ke 35.55.55 bi´kini bi.ji.ni 214.55.35
´hacker hei.ke 55.51 ´golf gao.er.fu 55.214.55
Twenty loanwords were randomly chosen from Xiàndài Hànyǔ Cídiǎn (Modern
Chinese Dictionary) as showing in (II), and we can see that different from Cantonese,
about half of the stressed syllables do not receive T1 in Mandarin, which does not
correspond to the initial prediction of this study.
Importantly, the adaptation of a foreign word in Chinese languages (no matter
Mandarin or Cantonese) can never be a pure phonological process since Chinese has a
totally distinct ideographic script system, i.e. the Chinese characters. That is to say,
every Chinese character itself carries a semantic meaning by itself, and there is no
direct relation between the pronunciation of a word and its corresponding character(s).
When we adapt an English word into Mandarin, we have to use the characters as
phonological symbols to note the original pronunciation syllable by syllable.
For instance, in (III) there is the brand-name of a popular laundry detergent with
its Chinese transcription 汰渍 (Tai.zi), which is known as a very successful
transcription of commercial brands in China. The meanings of the two characters are
“eliminate” and “stain” respectively, and thus besides its original meaning, the visual
makeup of this transcription will give the customers attractive information of
“eliminate stain” no matter what the real meaning of the original word is.
(III) The brand-name Tide and its Chinese transcription
3
As another illustration, the word cool was phonologically adapted as [ku] in
Mandarin, and in order to adapt this pronunciation, we had to look for an appropriate
character whose pronunciation is [ku] for the transcription. In this case, the Chinese
translators always face the problem that there is a clash of the semantic meaning of
the character itself and the meaning of the original English word, and they must
choose a best ideogram for phonological transcription. The loanword [ku] from cool
has a narrower meaning in Chinese, and it is only used to describe a person or
something that is “fashionably attractive or impressive”. The possible characters for
the transcription are shown in Table 1:
Original
word
Pinyin
transcription
Corresponding
character
Original meaning of the
character
cool ku [55] 枯 adj. withered
ku [55] 哭 v. cry
ku [35] None
ku [214] 苦 adj. bitter
ku [51] 库 n. storehouse
ku [51] 裤 n. pants
ku [51] 酷 adv. extremely adj.
oppressive
Table 1 Possible characters for the transcription of cool
According to Chao (1968), the cognitive schema of Chinese characters is that
each monosyllabic morpheme corresponds to a character, and such phenomenon
easily leads the Chinese native speakers to treat every syllable as a meaningful unit,
even if it is a meaningless syllable. Such cognitive tendency is embedded in the
Chinese languages. Li (2007), in his research of transliteration and phonological
translation, claims that in Chinese each identical syllable corresponds to a large
4
selection of Chinese characters, each with different meanings embedded in the visual
makeup of the character, whether or not a meaning is desirable, then it plays a role in
the choice of transliteration strategy. That is to say, when we adapt an English word
into Chinese, we always have to consider whether the meaning of the character itself
and the meaning of the English word match or not.
Thus, in the case of the loanword cool, the final official transcription has been
the last candidate 酷(ku, T4) which is a polysemic adjective ad adverb, with a
relatively vague meaning. If we had adapted it to T1, it would be totally strange for
Chinese speakers to relate the character’s meaning, either 枯(withered) or 哭(cry),
with the real meaning of the word cool.
The main goal of this study was to investigate the tone adaptation of English
loanwords in Mandarin. Because of the peculiarities of Chinese ideographic script
system, there is always interference between the optimal choice from a phonological
point of view and the meanings of the Chinese characters when assigning them to
transcribe a stressed syllable. For this reason, I proposed two research questions for
this study: (1) Is the tone adaptation of stressed syllables of English loanwords
predictable when there is no semantic constraint of the ideograms (in this case the
optimal choice should be T1)? (2) If a stressed English syllable does not receive T1 in
Mandarin, then there must be semantic constraints involved, and is there any rule or
pattern to choose from the other three tones for adaptation?
In order to shed light on these research questions, two small-scale studies were
designed. Study 1 aimed at testing whether the tone adaptation of English loanwords
stressed syllables in Mandarin shows the same pattern as in Cantonese, i.e. whether
they also receive T1 in Mandarin. Study 2 aimed at finding out the pattern of tone
adaptation under different semantic conditions by testing Mandarin native speakers’
intuitive choices when adapting English nonsense words.
5
2. Study 1
Study 1 is a preparatory study which consists in a dictionary-based data analysis.
In order to analyze the distribution of the tones that English loanwords stressed
syllables receive in Mandarin, fifty English loanwords were chosen from Xiàndài
Hànyǔ Cídiǎn (Modern Chinese Dictionary). The expectation was that a large number
of the stressed syllables receive T1, while the other three tones should be evenly
distributed.
The criteria for choosing the target words for this study are the following: 1.
Directly borrowed into Mandarin from English, not through another language or
dialect of Chinese; 2. Adapted into Mandarin after the 1920s when modern Chinese
Mandarin became a steady phonological system and there is seldom phonological
changes since then; 3. Accepted by Xiàndài Hànyǔ Cídiǎn (Modern Chinese
Dictionary), which is the most authoritative dictionary of modern Chinese; 4. Widely
known and used in daily life; 5. Randomly chosen from Xiàndài Hànyǔ Cídiǎn
(Modern Chinese Dictionary). (See Appendix I)
Tones Absolute numbers Proportions
T1 23 46%
T2 12 24%
T3 7 14%
T4 8 16%
Table 2 Distribution of tones that the target words stressed syllables receive
Table 2 shows that T1 has a much larger proportion than any other tones, which
is an answer to the first research question. However, the choice for other three tones
still needs explanation.
Twenty-seven words which do not satisfy the rule were qualitatively analyzed
(See Appendix II) and it showed the following pattern:
6
(1). Eleven of these 27 words (40.7%) whose stressed syllables do not receive T1
in Mandarin is because there is no corresponding T1 characters, and as a result, a
character of another tone has to be chosen for the transcription. That is to say, besides
the phonological factors, in the process of adapting a foreign word into Chinese, there
is a constraint of choosing an appropriate character for transcription because not every
syllable has corresponding characters of all four tones.
(2). The remaining 16 of these 27 words (59.3%) could have chosen a character
of T1 for transcribing a stressed syllable, but still they receive another tone instead.
Therefore, we have to account for these exceptions from the perspective of semantic
constraints:
(2a). The transcribed stressed syllable of eight3 of the 16 words (29.6%) has
only one corresponding character of T1. Seven of these characters (except No.3) are
monosemic, and seven of them (except No.25) are verbs. Thus if we assign one of
them to certain stressed syllable, the meaning of the ideogram itself will be bounded
to the loanword and thus will clash with the real meaning of the word. For instance, if
we adapt marathon (See Appendix II, No.25) as ma.la.song (55.55.55), the only
character we have to transcribe ma(55) is 妈, which has the only meaning “mother”
in Chinese. So it would be very unnatural and difficult for native speakers to relate
this character which means “mother” to the original meaning of the loanword, which
is “a long-distance running race, strictly one of 26 miles 385 yards”. Similarly, the
other 7 words also face this problem, and thus a character of another tone has been
chosen for the transcription.
(2b). Another eight4 of the 16 words (29.6%) have more than one T1 options,
but still they receive another tone instead. Nevertheless, a regular pattern of how to
choose a Chinese character for transcription has been concluded from the data:
(2b-1). Almost all of the chosen characters (except No. 18) are polysemic, which
means that without context, it will be hard to determine its exact meaning. The
advantage of choosing a polysemic character for transcription is that its meaning will
3 Number 3, 25, 26, 28, 29, 30, 31, and 32. See Appendix II. 4 Number 7, 11, 14, 18, 43, 46, 48 and 50. See Appendix II.
7
not be strictly bounded to the loanword and thus usually there will not be a semantic
clash.
(IV) ´chocolate – qiao ke li qiao ke li
敲(T1) 克 力 巧(T3) 克 力
knock 1.skillful
2.by chance
悄(T1) 克 力 3.sweet (words)
silent 4.a family name
For instance, for transcribing the stressed syllable cho of chocolate shown in
(IV), if we assign T1 to it, then it will be very awkward for Chinese native speakers to
relate either of the two monosemic T1 characters to “food”. However, the official
transcription is 巧克力, in which the meaning “sweet” of the polysemic character巧
can be somehow semantically related to chocolate, although originally in Chinese it is
only used for sweet words but not for sweet food.
(2b-2). There is a tendency to choose characters which can be used for proper
nouns in Chinese for transcription. Eighteen of the qualitatively analyzed 27 words
(66.67%) have chosen a character which is also a family name in Chinese (See
Appendix II). Such characters are not only family names, but also usually appear in
other proper names, such as city names. On the other hand, most of the loanwords5
are nouns. Therefore, by using these characters which are frequently used in Mandarin
as nouns (or components of nouns) to transcribe nouns, it will be easier for the
Mandarin native speakers to accept.
(2b-3). There is also a tendency to choose a character whose meaning can be
somehow positively semantically related to the meaning of the loanword.
To illustrate what is “positively semantically related”, we rechecked the 14th
and
the 50th
words (See Appendix I) here. The character 引(yin, T3) for transcription of
5 In this study, 46 of the 50 checked loanwords are nouns. Especially, 8 can be either noun or verb, and 22 can be
either adjective or noun. See Appendix II.
8
en in engine has the meaning of “lead” or “cause”, so it could be naturally related to
“a machine with moving parts that converts power into motion”, i.e. “a machine that
leads a vehicle to move”, or “a machine that causes motion”. In case of yuppies,
which are “well-paid young middle-class professionals who work in a city job and has
a luxurious lifestyle”, the meaning “elegant” or “stylish” of the chosen character 雅
(ya, T3) can be perfectly combined with the original word. Among the eight
loanwords for which there are more than one T1 options but still they receive another
tone, five6 can be positively related to the meaning of the characters they have chosen.
Usually this kind of transcriptions is considered very good since both from the
phonological and semantic point of view, they are very similar to the original word.
Similarly, when the meaning of the character clash with the meaning of the
original English word, it is considered as “negatively semantically related” (See the
mentioned example of chocolate in (IV)). When the meaning of the character is
neither positively nor negatively semantically related to the meaning of the original
English word, then it is considered “neutrally semantically related”.
Based on the results of Study 1, it seems that T1 has priority in the choice of
appropriate characters for transcription. However, the tone adaptation of English
loanwords stressed syllable in Mandarin seems unpredictable since there are a large
amount of exceptions which do not follow the initial high-tone adaptation rule. Still,
the following questions await to be solved: (1) For those stressed syllables which do
not receive T1 in Mandarin, is there any corresponding T1 characters? If there is none,
then naturally they cannot receive T1 in Mandarin. But if there are T1 characters but a
stressed syllable still does not receive T1, then it has to be explained why. (2) For
each transcribed stressed syllable, we have to check the semantic meanings of all the
possible selections of corresponding characters of different tones, so that we can test
whether the semantic constraints do play a role in the choice of ideograms for
transcription. In order to investigate these questions, I carried out a rating experiment
in Study 2.
6 14, 43, 46, 48 and 50. See Appendix II.
9
3. Study 2
3.1 Goals
The results of Study 1 show that about half of the English loanwords stressed
syllables receive T1 in Mandarin, but it still remains unknown that why the rest of the
loanwords do not receive T1 in Mandarin. Therefore, Study 2, which is a
questionnaire-based rating experiment and the central part of this paper, has been
designed to test the native speakers’ intuitive adaptation of English loanwords into
Mandarin by controlling semantic constraints, which means that in each option of an
item, the target characters can be positively, negatively or neutrally semantically
related to the original English word. By this experiment, firstly we intended to find
out whether the adaptation of loanwords stressed syllables as T1 is predictable in
Mandarin, and secondly, if semantic constraints play a role in the process of
transcription which leads to the choice of other tones, then is there a rule that we can
conclude?
3.2 Methods
The current study was based on a rating experiment which was sent to the
participants through email. In every item, there were three options which contained
target characters whose meanings were differently semantically related to one of the
three original English words. The participants were asked to score each option
according to a given grading scale. After finishing rating all items, they were also
asked to write a short comment on how they considered an option as a good one or a
bad one.
10
3.2.1 Participants
Twenty-eight Mandarin native speakers with university background and with a
relatively good level of English participated in the experiment. Participants with
background in linguistics, translation or similar subjects were avoided. The results of
four participants were excluded because two of them had background in translation,
and another two participants didn’t follow the instructions to answer the questions
(See Appendix III).
3.2.2 Materials
Three nonsense English words Kushrey, Barapara and Meerif were used in this
experiment. At the beginning of the questionnaire, the made-up definitions of these
words were given. Two situations of adapting English loanwords stressed syllables
were designed for this experiment. In the first situation there were possible T1
characters to transcribe stressed syllables. The variables were divided into two groups:
T1, and other tones. Since both the selection of T1 and other tones can be neutrally7,
negatively or positively semantically related to the meaning of the loanword, there are
in total 9 combinations, as showing in Table 3:
T1 Other tones Number of items8
Group 1 Neutrally semantically related Neutrally semantically related 1
Neutrally semantically related Positively semantically related 2
Neutrally semantically related Negatively semantically related 3
Group 2 Positively semantically related Neutrally semantically related 4
Positively semantically related Positively semantically related 9
7 In this study, the “neutrally semantically related” characters were chosen by my personal experience as Mandarin
native speaker. Also following the pattern of the qualitative analysis in Study 1, verbs, nouns and monosemic
characters were avoided and polysemic characters with relatively vague and unfixed meanings were preferred. 8 See Appendix III. 9 This combination has been excluded from the experiment because it is difficult to make up a word which satisfy
this criterion.
11
Positively semantically related Negatively semantically related 5
Group 3 Negatively semantically related Neutrally semantically related 6
Negatively semantically related Positively semantically related 7
Negatively semantically related Negatively semantically related 8
Table 3 Combinations of different semantic conditions when T1 was available
In the second situation there was no T1 character and thus the participants had to
choose from other tones. Three combinations were designed according to the different
degree of semantic relation, shown in Table 4:
T2, T3 or T4 Number of items
Group 4 All neutrally semantically related 9
At least one positively semantically related 10
At least one negatively semantically related 11
Table 4 Combinations of different semantic conditions when T1 was not available
Therefore, eleven items corresponding to these combinations were designed. In
each question, one of the three nonsense made-up words and its pinyin transcription
was given, and the participants were asked to score every option. For non-stressed
syllables, the same characters were used for transcription. The only difference
between two options was that for each option, a different character was used for
transcribing the stressed syllable. Before answering the questionnaire, the participants
were told they didn’t have to consider whether the pronunciation of the transcription
was similar to the original word or not since the aim of the questionnaire was to test
the semantic constraints. An example of an item of the questionnaire is shown in (V)
(See Appendix III for the full questionnaire with gloss in English):
(V) Meerif – Mi.le.fu
A. 米乐芙 B. 眯乐芙 C. 蜜乐芙
12
3.2.3 Procedure
At the beginning of the questionnaire, the made-up definitions of three nonsense
English words were given, shown in (VI), so that when the participants were to
choose a character for transcription, they had to take both the meaning of the
character and the meaning of the original word into consideration.
(VI) Made-up definitions of the nonsense words used in the experiment
Kushrey ['kʌʃri] is a seafood dish of rice and clam cooked together, topped with
pasta and garlic juice.
Barapara [Barə'parə] is a synchronized group dance, which consists of mostly
arm movements, and very little lower body movement is involved.
Meerif ['mi:rif] is a dessert formed into a loaf shape, and usually made of
mixture of cooked rice, honey and sugar.
In each item, there appeared one of the three words that have been transcribed
into pinyin already, and the participants were asked to use numbers to score the
options which correspond to such transcription. The grading scale is as following:
Number 1 2 3 4 5
Meaning Unacceptable Bad Normal Good Perfect
Table 5 Grading scale for the participants
When the participants were to score the options, they were told not to consider
whether the pronunciation of the adapted form was good or not, and they only need to
consider whether the choice of characters were appropriate or not. Some options
reappeared in different items, and the participants were told to make sure that the
scores they gave to the same option were the same.
13
After finishing the questionnaire, the participants were asked to write a short
comment on how they consider an item as good or bad with 20-100 words.
3.3 Results
3.3.1. Results of Group 1, Group 2 and Group 3
(1). Effect of preference for T1 when no semantic constraint appeared
Figure 1. Average scores of the four tones when no semantic constraint appeared
First, we compared the average scores of the situation in which all four tones
were available and no semantic constraint appeared, i.e. the meanings of all the
options were neutrally semantically related to the meaning of the loanword. As shown
in Figure 1, the scores that the participants gave to the four tones were 2.94, 2.67, 2.13,
and 2.26 respectively. Remind that according to the grading scale, in the vertical axis
“2” means “bad” and “3” means “normal”. Thus we can consider the average score of
T1 as very close to “normal”, and it is clear that the preference scale of the four tones
from high to low is T1, T2, T4, and T3.
(2). Effect of semantic constraints in the adaptation
2
2.2
2.4
2.6
2.8
3
T1 T2 T3 T4
14
Figure 2. Average scores of positively semantically related options
Figure 3. Average scores of negatively semantically related options
Second, we compared the average scores of the four tones in different semantic
conditions. As shown in Figure 2 and Figure 3, when the target characters were
positively semantically related to the meaning of the loanwords, the average scores
were much higher (M=3.88), and when they were negatively semantically related, the
results were the other way around (M=1.35).
The preference scale from high to low changed in different semantic conditions:
When the targets characters were positively semantically related, it was T3, T4, T1
and T2; When they were negatively semantically related, it was T1, T3, T4 and T2.
3
3.2
3.4
3.6
3.8
4
4.2
4.4
4.6
4.8
5
T1 T2 T3 T4
1
1.2
1.4
1.6
1.8
2
T1 T2 T3 T4
15
However, we cannot simply use this conclusion to make prediction, because how a
participant judged an option as positively or negatively semantically related to the
original word can be very subjective. We can imagine that once the target character is
semantically – no matter positively or negatively - related to the loanword, for every
participant, he/she has an own criteria to score and thus it could be very subjective.
Therefore, the final average score may have been affected by it. That is to say, the
deviation of the scores of both positively and negatively semantically related
characters is much larger than that of the neutrally semantically related characters, and
thus we claimed that the former ones are not reliable to make prediction.
As an illustration, in Item 10 (See Appendix III), the nonsense English word
Meerif was given with a made-up definition of “a dessert formed into a loaf shape,
and usually made of mixture of cooked rice, honey and sugar”, and the participants
were asked to score for every option. The target character in one option was 蜜(mi T4,
‘honey’), which indicates one ingredient of this dish, and thus was expected it to be
considered as positively semantically related to the original word. Nevertheless, there
were still two participants who only gave it 1 point although the average score of this
option is 4.08. On the other hand, another option 米(mi T3, ‘rice’) which also
indicates one ingredient of this dessert, was expected to be considered as positively
semantically related too, but there was one participant who gave it only 1 point
although the average score is 4.5.
We concluded that when the targets characters are positively semantically related
to the loanwords, they got significantly higher scores. When they are negatively
semantically related, the results were the other way around. However, for the
mentioned reasons, we cannot get a reliable preference scale for further prediction just
based on the results.
3.3.2. Results of Group 4
(1). Effect of preferences for the other three tones when no semantic constraint
appeared
16
Figure 4. Average scores of the other three tones when no semantic constraint appeared
Once there was no T1 option, the participants had to choose from other three
tones. When no semantic interference appeared, as shown in Figure 3, the scores of
the three tones were 2.71, 2.13, and 2.35 respectively, which was very similar to the
results in Figure 1. The preference scale from high to low is still T2, T4, and T3.
(2) Effect of semantic constraints in the adaptation
Figure 5. Average scores of the other three tones when positively semantically related
2
2.2
2.4
2.6
2.8
3
T2 T3 T4
3
3.2
3.4
3.6
3.8
4
4.2
4.4
4.6
4.8
5
T2 T3 T4
17
Figure 6. Average scores of the other three tones when negatively semantically related
As shown in Figure 5 and Figure 6, once the semantic interference appeared, the
results show a similar pattern to that of Section 3.3.1. The preference scale for
positively semantically related target characters was T3, T4, and T2, and for
negatively semantically related target characters was T2, T3, and T4. However, for the
same reason as mentioned before, we cannot use these preference scales to make
correct prediction since the participants may have scored them by subjective criteria.
4. Conclusion
In the current paper we dealt with the tone adaptation of English loanwords
stressed syllables in Mandarin with a small-scale qualitative analysis and a rating
experiment. The results from Study 1 and Study 2 showed a different pattern than tone
adaptation in Cantonese (Silverman 1992, Lai et al. 2011, Davis et al. 2012). Study 1
showed that about half of the stressed syllables receive T1 in Mandarin. Furthermore,
Study 2 showed that when adapting English loanwords stressed syllables into
Mandarin, the preferred choice is T1 if no semantic constraint appears, and the
preference scale from high to low is T1, T2, T4 and T3 (See Figure 1). When T1 is not
1
1.2
1.4
1.6
1.8
2
T2 T3 T4
18
available and another tone has to be chosen, the preference scale remains as T2, T4
and T3 (See Figure 4). However, we claim that the preference scale in conditions
where the target characters are positively or negatively semantically related to the
original words is not reliable to make prediction because how the native speakers
judge a transcription to be good or not depends on their subjective criteria, and the
degree of relation between a target character and the meaning of the original word is
also different.
According to the comments of the participants at the end of the questionnaire, 8
of the 24 participants mentioned that they made the choices solely by their intuition,
and 17 of them mentioned semantic constraints (e.g. whether the meaning of the
target character and the meaning of the loanword corresponds, or whether the target
character seems attractive if it appears in the name of a dish, etc). Therefore, when the
participants had to take the semantic factors into consideration, the adaptation stopped
being a purely phonological process. On the contrary, when all targets characters are
neutrally semantically related to the original words, no semantic constraints appear
and the participants’ choices are considered as pure phonological transcription process.
Thus the preference scales that we concluded from Figure 1 and Figure 4 are reliable.
In order to understand the difference between Cantonese and Mandarin in tone
adaptation of English loanwords, we may have to get to know that Cantonese and
English have been in close contact for over two centuries since Hong Kong became
British colony and that Cantonese is primarily a spoken language and hence
loanwords are normally transmitted orally rather than graphically (Kenstowicz 2012),
which means that in the adaptation of English loanwords in Cantonese, the semantic
constraints are avoided because instead of choosing an appropriate character for
transcription, the people only transcribed them orally. Therefore, in most cases the
preferred choice is T1 in Cantonese.
In summary, the main findings of the current study is that when adapting English
loanwords stressed syllables into Mandarin, it is predictable that T1 is preferred when
no semantic constraint appears. However, once the semantic constraints appear, the
degree of acceptance depends on the native speakers’ subjective criteria, and thus in
19
many cases the tone adaptation of English loanwords stressed syllables remains
unpredictable in Mandarin.
References
Chao, Yuen Ren. 1968. A Grammar of Spoken Chinese. University of Carlifonia
Press.
Davis, Stuart; Natsuko; Tu, Jung-yueh. Towards a Taxonomy of Loanwords
Prosody. Catalan Journal of Linguistics, 2012, vol. 11, 13-39.
Duanmu, San. 1990. A Formal Study of Syllable, Tone, Stress and Domain in
Chinese Languages. Massachusetts Institute of Technology.
Guo, Hugo Ling-yu. 1999. Mandarin Loanword Phonology and Optimality Theory:
Evidence from Transliterated American State Names and Typhoon Names. National
Chengchi College, Taiwan.
Kenstowicz, Michael J. 2012. Cantonese Faithfulness in VC Rime Constraints.
Catalan Journal of Linguistics, 2012, vol. 11, 65-96.
Lai, Wiene Wing-sze; Wang, Dongning; Yan, Nan; Chan, Victor; Zhang, Lan.
2011. Influence of English Donor Word Stress on Tonal Assignment in Cantonese
Loanwords – An Acoustic Account. University of Hong Kong.
Li, Chris Wen-chao. 2007. Foreign names into native language: How to Transfer
Sound Between Languages – Transliteration, Phonological Translation, Nativization,
and Implications for Translation Theory. San Francisco State University.
Miao, Ruiqin. 2005. Loanwords Adaptation in Mandarin Chinese: Perceptual,
Phonological and Sociolinguistic Factors. Stony Brook University.
Silverman, Daniel. 1992. Multiple Scansion in Loanword Phonology: Evidence from
Cantonese. Phonology 9, 289-328.
Wan, Hong. 2007. The Third Wave of Loanwords in Standard Chinese. Nankai
University Press. Tianjin, China.
20
Yang, Pengxi. 2007. Hanyu Wailaici Yanjiu (汉语外来词研究 , “Research of
Loanwords in Chinese”). Shanghai Renmin Publishing House. Shanghai, China.
Appendix I
50 examples of tone adaptation of English loanwords in Mandarin
Number Original word Pinyin
transcription
Corresponding
characters
Corresponding
tones
Tone of the
stressed
syllable
1 ´bar ba 吧 55 T1
2 bi´kini bi.ji.ni 比基尼 214.55.35 T1
3 ´bit bi.te 比特 214.51 T3
4 ´brandy bai.lan.di 白兰地 35.35.51 T2
5 ´bungee beng.ji 蹦极 51.35 T4
6 car´toon ka.tong 卡通 214.55 T1
7 ´chocolate qiao.ke.li 巧克力 214.51.51 T3
8 ´clone ke.long 克隆 51.35 T2
9 ´coffee ka.fei 咖啡 55.55 T1
10 ´cookie qu.qi 曲奇 55.35 T1
11 ´cool ku 酷 51 T4
12 ´DINK ding.ke 丁克 55.51 T1
13 ´disco di.si.ke 迪斯科 35.55.55 T2
14 ´engine yin.qing 引擎 214.35 T3
15 ´gene ji.yin 基因 55.55 T1
16 ´golf gao.er.fu 高尔夫 55.214.55 T1
17 gui´tar ji.ta 吉他 35.55 T1
18 ´Gypsy Ji.bu.sai 吉卜赛 35.214.51 T2
21
19 ´hacker hei.ke 黑客 55.51 T1
20 ´hippies xi.pi.shi 嬉皮士 55.35.51 T1
21 ´humor you.mo 幽默 55.51 T1
22 hys´teria xie.si.di.li 歇斯底里 55.55.214.214 T3
23 ´jazz jue.shi 爵士 35.51 T2
24 ´jeep ji.pu 吉普 35.214 T2
25 marathon ma.la.song 马拉松 214.55.55 T3
26 mo´saic ma.sai.ke 马赛克 214.51.51 T4
27 ´microphone mai.ke.feng 麦克风 51.51.55 T4
28 ´mini mi.ni 迷你 35.214 T2
29 ´modern mo.deng 摩登 35.55 T2
30 ´mousse mo.si 摩斯 35.55 T2
31 ´motor mo.tuo 摩托 35.55 T2
32 ´pie pai 派 51 T4
33 ´poker pu.ke 扑克 55.51 T1
34 ´pudding bu.ding 布丁 51.55 T4
35 ´quark kua.ke 夸克 55.51 T1
36 ´radar lei.da 雷达 35.35 T2
37 ´salad se.la 色拉 51.55 T4
38 sar´dine sha.ding 沙丁 55.55 T1
39 ´sandwich san.ming.zhi 三明治 55.35.51 T1
40 ´sauna sang.na 桑拿 55.35 T1
41 ´shampoo xiang.bo 香波 55.55 T1
42 ´shock xiu.ke 休克 55.51 T1
43 ´show xiu 秀 51 T4
44 ´sofa sha.fa 沙发 55.55 T1
45 ´sonar sheng.na 声纳 55.51 T1
46 ´tank tan.ke 坦克 214.51 T3
47 ´TOEFL tuo.fu 托福 55.35 T1
22
48 ´vitamin wei.ta.ming 维他命 35.55.51 T2
49 ´whisky wei.shi.ji 威士忌 55.51.51 T1
50 ´yuppies ya.pi.shi 雅皮士 214.35.51 T3
Appendix II
Possible characters for transcription of the words in Study 1
Number Stressed
syllable
Possible selection
of characters10
Meanings
3 bi T1 逼 1. v. force 2. v. threat 3. get close to
T2 鼻 n. nose
T3 比 1. v. compare 2. v. gesture 3. v. emitate 4. v. compete
5. prep. than 6. a family name
彼 1. pron. that 2. pron. he
笔 n. pen
鄙 1. adj. vulgar 2. adj. humble
T4 币 n. currency
必 1. adv. certainly 2. v. must 3. a family name
毕 1. v. finish 2. a family name
闭 1. v. close 2. v. obstruct 3. v. cease 4. a family name
毙 v. die
碧 1. green jade 2. adj. green
蔽 v. cover
弊 1. v. cheat 2. n. disadvantage
壁 n. wall
避 v. avoid
10 All the characters in the chart are chosen according to the List of the Most Commonly Used 2500 Chinese
Characters, published by the Ministry of Education of P.R.China in 1988. These 2500 characters cover 99.48% of
daily readings necessities.
23
臂 n. arm
4 lan T1 None
T2 兰 1. n. orchid 2. a family name
拦 v. block
栏 1. n. fence 2. n. column
蓝 1. adj. blue 2. a family name
篮 1. n. basket 2. a family name
T3 览 v. view
懒 adj. lazy
T4 烂 adj. rotten
滥 1. n. overflow 2. adj. excessive
5 beng T1 None
T2 None
T3 None
T4 蹦 v. jump
7 qiao T1 悄 adj. silent
敲 v. knock
T2 乔 1. adj. tall 2. a family name 3. v. disguise
侨 1. reside abroad 2. a family name
桥 1. n. bridge 2. a family name
瞧 look at
T3 巧 1. adj. skillful 2. by chance 3. adj. sweet (words) 4. a
family name
T4 None
8 long T1 None
T2 龙 n. dragon
隆 1. adj. prosperous 2. adj. grand 3. a family name
聋 adj. deaf
笼 n. cage
24
T3 拢 1. v. close 2. v. arrive
垄 1. ridge in a field 2. a family name
T4 None
11 ku T1 枯 adj. withered
哭 v. cry
T2 None
T3 苦 adj. bitter
T4 库 n. storehouse
裤 n. pants
酷 1. adv. extremely 2. adj. oppresive
13, 22 di T1 低 adj. low
滴 v. drop
T2 迪 a family name
的 adj. certain
敌 n. enemy
笛 n. flute
T3 抵 1. v. resist 2. v./n. mortage 3. v. equal 4. v. reach
底 1. n. bottom 2. n. end (of year or month)
T4 地 1. n. earth 2. n. land 3. n. floor 4. n. location 5. n.
region
弟 younger brother
帝 n. emperor
递 v. pass
第 1. prefix used before a cardinal to form a ordinal 2. a family
name
14 yin T1 因 1. n. reason 2. conj. because 3. a family name
阴 1. adj. lunar 2. adj. negative 3. adj. hidden 4. a family
name
音 1. n. sound 2. n. news 3. a family name
25
姻 n. marriage
T2 银 n. silver
T3 引 1. v. lead 2. v. leave 3. v. quote 4. v. cause
饮 v. drink
隐 v. hide
T4 印 1. n. mark 2. v. print 3. a family name
18 ji T1 击 v. attack
饥 n. hunger
机 n. machine
肌 n. muscle
鸡 n. chicken
积 v. accumulate
基 n. base
激 1. v. stimulate 2. v. arouse 3. a family name
T2 及 come up
吉 1. adj. lucky 2. a family name
级 n. level
极 1. n. pole 2. n. utmost 3. adv. mighty
即 adv. scilicet
急 v. worry
疾 n. disease
集 1. v. gather 2. n. market 3. n. episode 4. a family name
T3 几 adj. several
挤 v. squeeze
脊 1. n. ridge 2. n. spine
T4 计 1. v. calculate 2. v./n. plan 3. a family name
记 v. note
纪 1. n. discipline 2. n. era
技 1. n. skill 2. a family name
26
忌 1. v. envy 2. v. avoid
际 1. n. edge 2. prep. among
季 n. season
剂 1. v. adjust 2. n. remedy
迹 n. track
既 adv. already
继 1. v. continue 2. a family name
寄 v. send
绩 n. merit
23, 24 jue T1 None
T2 决 v. decide
觉 v. feel
绝 v. extinguish
掘 v. dig
爵 1. n. nobility 2. a family name
T3 None
T4 None
25 ma T1 妈 n. mother
T2 麻 1. n. hemp 2. adj. numb
T3 马 1. n. horse 2. adj. large 3. a family name
码 n. size
蚂 n. ant
T4 骂 v. curse
26 sai T1 塞 fill in
T2 None
T3 None
T4 赛 1. v. compete 2. v. win 3. a family name
27 mai T1 None
T2 埋 v. bury
27
T3 买 1. v. buy 2. a family name
T4 迈 1. v. stride 2. adj. old
麦 1. v. wheat 2. a family name
卖 1. v. sell 2. v. a family name
脉 n. pulse
28 mi T1 眯 v. squint
T2 迷 1. adj. confused 2. n. enthusiast 3. be crazy about
something
谜 n. puzzle
T3 米 n. rice
T4 秘 n. secret
密 1. adj. close 2. adj. dense 3. a family name
蜜 n. honey
29, 30,
31
mo T1 摸 v. touch
T2 模 1. n. norm 2. n. mold 3. a family name
摩 1. v. touch 2. v. rub 3. a family name
磨 v. grind
魔 n. devil
T3 抹 v. mop
T4 末 n. end
莫 do not (do something)
漠 1. n. desert 2. adj. indifferent
墨 1. n. ink 2. adj. black
默 1. adj. silent 2. a family name
32 pai T1 拍 v. pat
T2 排 1. v. arrange 2. n. row
牌 1. n. cards 2. n. brand 3. a family name
T3 None
T4 派 1. n. clique 2. n. style 3. n. group 4. v. assign 5.
28
measure word (for landscape, sound, language, etc.)
34 bu T1 None
T2 None
T3 卜 1. v. divine 2. a family name
补 1. v. repair 2. a family name
捕 1. v catch 2. a family name
T4 不 negative prefix
布 1. n. cloth 2. family name 3. v. announce 4. v.
arrange
步 1. n. footstep 2. n. phase 3. a family name
部 1. n. section 2. n. unit 3. a family name
36 lei T1 None
T2 雷 1. n. thunder 2. explosive weapon 3. a family name
T3 垒 n. citadel
T4 泪 n. tear
类 1. n. category 2. a family name
累 adj. tired
37 se T1 None
T2 None
T3 None
T4 色 1. n. color 2. n. kind 3. n. scene 4. a family name
43 xiu T1 休 1. v. cease 2. v. rest 3. a family name
修 1. v. repair 2. v. decorate 3. a family name
羞 adj. shy
T2 None
T3 朽 adj. rotten
T4 秀 1. adj. elegant 2. adj. smart 3. adj. outstanding 4. n.
elite 5. a family name
袖 n. sleeve
29
绣 v. embroider
锈 n. rust
46 tan T1 贪 v. covet
摊 spread out
滩 n. beach
T2 坛 n. jug
谈 1. v. talk 2. a family name
弹 v. spring
T3 坦 1. adj. flat 2. adj. frank 3. adj. calm
T4 叹 v. sigh
探 1. v. explore 2. v. visit 3. n. spy
48 wei T1 危 1. adj. dangerous 2. v. harm 3. a family name
威 1. n. power 2. a family name
微 adj. tiny
T2 违 v. disobey
围 1. v. encircle 2. a family name
唯 adv. only
维 1. v. connect 2. v. maintain 3. n. dimension
T3 伟 1. adj. great 2. a family name
伪 adj. fake
尾 n. tail
委 1. v. assign 2. n. committee 3. a family name
T4 卫 v. guard
为 prep. for
未 not yet
味 n. taste
位 1. n. position 2. a family name
畏 v. fear
胃 n. stomach
30
喂 v. feed
50 ya T1 压 v. press
押 v. pledge
鸦 n. crow
鸭 n. duck
T2 牙 n. tooth
芽 n. shoot (of plant)
崖 n. cliff
T3 哑 adj. mute
雅 1. adj. elegant 2. adj. stylish 3. a family name
T4 亚 adj. inferior
Appendix III
The full questionnaire with gloss in English
Kushrey ['kʌʃri] is a seafood dish of rice and clam cooked together, topped with pasta and
garlic juice.
Barapara [Barə'parə] is a synchronized group dance, which consists of mostly arm
movements, and very little lower body movement is involved.
Meerif ['mi:rif] is a dessert formed into a loaf shape, and usually made of mixture of
cooked rice, honey and sugar.
Here are three English words which have not been adapted by Mandarin. In each question,
one of them will appear with three homophonic transcriptions in Chinese characters. Please read
carefully and score the options with number 1 to 5. The grading scale is shown as following:
Number 1 2 3 4 5
Meaning Unacceptable Bad Normal Good Perfect
31
When you score the options, you do not have to consider whether the pronunciation of the
adapted form is good or not, and you only need to consider whether the choice of characters are
appropriate or not. Some options may reappear, please make sure the scores you give to the same
item in different questions are the same.
After finishing the questionnaire, please write a short comment on how you consider an item
as good or bad with 20-100 words.
1. Kushrey – Ke.sha.li
A. 科(T1)沙丽 (1. Branch or division of studies 2. Short for ancient Chinese imperial exams 3.
Family (Biology) 4. A family name)
B. 可(T3)沙丽 (1. v. agree 2. auxiliary verb with the meaning of “possibility” 3. prefix, used with
a verb to form an adjective 4. adv. but)
C. 克(T4)沙丽 (1. v. can 2. overcome 3. a family name)
2. Kushrey – Ge.sha.li
A. 戈(T1)沙丽 (1. dagger axe 2. a family name)
B. 蛤(T2)沙丽 (clam)
C. 格(T2)沙丽 (1. n. grid 2. n. size 3. n. case (linguistics) 4. a family name)
3. Kushrey – Ke.sha.li
A. 咳(T2)沙丽 (v. cough)
B. 科(T1)沙丽 (1. Branch or division of studies 2. Short for ancient Chinese imperial exams 3.
Family (Biology) 4. A family name)
C. 渴(T3)沙丽 (adj. thirsty)
4. Kushrey – Ge.sha.li
A. 割(T1)沙丽 (cut off)
B. 格(T2)沙丽 (1. n. grid 2. n. size 3. n. case (linguistics) 4. a family name)
C. 各(T4)沙丽 (1. pron. each 2. a family name)
5. Kushrey – Ku.sha.li
A. 哭(T1)沙丽 (v. cry)
B. 枯(T1)沙丽 (adj. withered)
C. 苦(T3)沙丽 (adj. bitter)
6. Barapara – Pa.la.pa.la
32
A. 啪啦啪(T1)啦 (onomatopoeia for applause or clap)
B. 怕啦怕(T4)啦 (v. fear)
C. 趴啦趴(T1)啦 (lean over)
7. Barapara – Pa.la.pa.la
A. 爬啦爬(T2)啦 (v. crawl)
B. 怕啦怕(T4)啦 (v. fear)
C. 啪啦啪(T1)啦 (onomatopoeia for applause or clap)
8. Meerif – Mi.le.fu
A. 米(T3)乐芙 (1. n. rice 2. a family name)
B. 眯(T1)乐芙 (v. squint)
C. 蜜(T4)乐芙 (1. n. honey 2. adj. sweet)
9. Meerif – Mi.le.fu
A. 弥(T2)乐芙 (1. adj. full 2. a family name)
B. 密(T4)乐芙 (1. adj. close 2. adj. dense 3. a family name)
C. 秘(T4)乐芙 (1. n. secret 2. adj. rare)
10. Meerif – Mi.le.fu
A. 米(T3)乐芙 (1. n. rice 2. a family name)
B. 蜜(T4)乐芙 (1. n. honey 2. adj. sweet)
C. 密(T4)乐芙 (1. adj. close 2. adj. dense 3. a family name)
10. Meerif – Mi.le.fu
A. 糜(T2)乐芙 (adj. rotten)
B. 秘(T4)乐芙 (1. n. secret 2. adj. rare)
C. 弥(T2)乐芙 (1. adj. full 2. a family name)
Appendix IV
Results of the questionnaire
1. Kushrey – Keshali
1 2 3 4 5 Average Total
科沙丽 3 8 6 5 2 2.79 67
可沙丽 5 2 6 10 1 3 72
33
克沙丽 3 5 10 4 2 2.88 69
2. Kushrey – Geshali
1 2 3 4 5 Average Total
戈沙丽 1 2 13 6 2 3.25 75
蛤沙丽 4 2 5 9 4 3.29 79
格沙丽 4 8 7 2 3 2.67 64
3. Kushrey - Keshali
1 2 3 4 5 Average Total
咳沙丽 23 1 0 0 0 1.04 25
科沙丽 3 8 6 5 2 2.79 67
渴沙丽 19 4 1 0 0 1.25 30
4. Kushrey – Geshali
1 2 3 4 5 Average Total
割沙丽 22 2 0 0 0 1.08 26
格沙丽 4 8 7 2 3 2.67 64
各沙丽 13 8 2 1 0 1.63 39
5. Kushrey – Kushali
1 2 3 4 5 Average Total
哭沙丽 21 0 1 1 1 1.38 33
枯沙利 19 2 0 2 1 1.5 36
苦沙丽 18 4 1 0 1 1.42 34
6. Barapara – Palapala
1 2 3 4 5 Average Total
啪啦啪啦 1 2 7 9 5 3.63 87
怕啦怕啦 18 5 1 0 0 1.29 31
趴啦趴啦 11 6 3 4 0 2 48
7. Barapara – Palapala
1 2 3 4 5 Average Total
爬啦爬啦 16 7 0 0 1 1.46 35
怕啦怕啦 18 5 1 0 0 1.29 31
啪啦啪啦 1 2 7 9 5 3.63 87
8. Meerif – Milefu
1 2 3 4 5 Average Total
米乐芙 1 0 0 8 15 4.5 108
眯乐芙 14 7 3 1 0 1.71 47
秘乐芙 6 8 8 1 1 2.29 55
34
9. Meerif – Milefu
1 2 3 4 5 Average Total
弥乐芙 5 6 5 6 2 2.75 66
密乐芙 6 5 9 3 1 2.5 60
秘乐芙 6 8 8 1 1 2.29 55
10. Meerif – Milefu
1 2 3 4 5 Average Total
米乐芙 1 0 0 8 15 4.5 108
蜜乐芙 2 1 4 3 14 4.08 98
密乐芙 6 5 9 3 1 2.5 60
11. Meerif – Milefu
1 2 3 4 5 Average Total
眯乐芙 14 7 3 1 0 1.71 47
秘乐芙 6 8 8 1 1 2.29 55
弥乐芙 5 6 5 6 2 2.75 66