English loanwords in Japanese.

10 

Loading....

Loading....

Loading....

Loading....

Loading....

Full text

(1)

Bobby Ruijgrok Student no.: s9303944

Course: Klank & Klankstructuur (BA) Teacher: Marc van Oostendorp Date: may/june-2009

En

g

lish lo

a

nwords in J

a

p

a

nese.

1 Introduction. 2

2 Adapting English words in Japanese. 2

3 Stress assignment. 5

4 Conclusion. 9

(2)

1# # Introduction

During the seminar Microvariation we $eshed out language variation that can be observed within one particular language. Language variation can have distinct sources. For instance, language can vary between speakers because they have contrasting social backgrounds or because they live in dif-ferent parts of a country. Even, a language may vary within one individual. All variation we came across we tried to underpin in terms of Optimality Theory. This term paper is a report of a little study in loanword phonology. In some cases such phonology can be predictable. Sounds that are not available in the adapting language, for example, are replaced by the ‘nearest’ sound at hand. The Italian [spagɛti] sounds in Dutch as [spaχeti] since /g/ does not belong to its sound inventory. Fur-ther, loanwords are subject to phonological rules, that is, to the constraint setting, of the adapting language. The English credit card pronounced as [krɛdətkɑrd] sounds in Dutch as [krɛdətkɑrt] since it does not allow voiced consonants in coda positions. In Tokyo Japanese (TJ), the language I will look at in this paper, word accent is realized in a completely di)erent way from English. That is, stress is placed on syllables where we would not expect it. It will appear that in both languages stress is assigned in quite the same way. In the next section I mention the processes that are involved in adapting English loanwords in Japanese, namely, sound change, vowel insertion and stress as-signment. Section 3 is dedicated to the last process which I will speak of in more detail. I conclude my *ndings in section 4.

2# # Adapting English words in Japanese.

Although the English were in Japan in the 17th century no English words were adapted at that time. The Japanese elite started using English words in the late eighteen hundreds as part of an attempt to import Western lifestyle. From then on the amount of English words used grew and spread gradually with a stop in the 1930’s when the government put a ban on foreign culture. After World War Two the usage of loanwords in Japanese exploded. Most of the approximate 25.000 loanwords in Japa-nese language are of English origin (Kay, 1995). Shinohara (2000) shows that when French loan-words are adapted in Japanese the Japanese phonology is at work at three levels. At the segmental level French segments are substituted by Japanese segments. Since Japanese favors CV syllable se-quences we can see at the syllabic level the occurrence of vowel epenthesis either to avoid codas or consonant clusters. Further, pre-*nal syllables are lengthened. At the accentual level pitch is as-signed. Apart from the lengthening of pre-*nal syllables all other processes are recognizable in the adaptation of English words. Markedness constraints that are assumed to bring about vowel epen-thesis are typically NOCODA and *COMPLEX (Kubuzono, 2001). (1), in which epenthetic vowels are between <>, explicates these constraints. Dots mark syllable boundaries. Further, some sound changes can be observed.

(1)#

Constraint Formula English Japanese

NOCODA CVC → CV.C<V> ticket tʃi.ke.t<o>

*COMPLEX CC → C<V>.C(V) three s<u>.ri:

(3)

More examples of sound changes and vowel epenthesis, due to one ore both constraints, are given in (2).

(2)#

In (1-2) we see that /u, o, i/ are in use as epenthetic vowels. Typically /o/ and /i/ are found in dis-tinct phonological contexts. The dental stops [t,d] trigger /o/ to be inserted, whereas the palato-alveolar a)ricates [tʃ, dʒ1] and sometimes [k] trigger /i/ (Kubuzono, 2001). I think that we could

consider /u/ as the default epenthetic vowel. Some authors con*rm this, however, the matter ap-pears to be more complex (Kenstowicz & U)mann, 2006). If we assume that in addition to the above mentioned markedness constraints two other (faithfulness) constraints are at work, we get a con-straint ranking as in (3) to obtain CV constellations. MAX(IMALITY) involves ‘no deletion’ as every segment of S1 should have a correspondent in S2 and DEP(ENDENCY) involves‘no epenthesis’ as every

segment of S2 should have a correspondent in S1 (McCarthy & Prince, 1995).

(3)# NOCODA ⟩⟩*COMPLEX ⟩⟩MAX ⟩⟩DEP

In the tableau in (4) we see the emergence of the optimal output of the English ‘stress’. Note that one or more constraints deciding which vowel is inserted are not shown here.

(4)

1Neither Kay (1995) norKubuzono (2001) verify this. However, my informant (Nakamura Mari, from Fukuoka)

produced [judʒ<i>] ‘judge’.

/stress/ NOCODA *COMPLEX MAX DEP

stress *! *

stressu *! *

☞ sutoresu **

sessu *! *

se *!*

English Japanese Constraints

beech bii.tʃ<i> NOCODA

taxi ta.ku.ʃi *COMPLEX

Christmas k<u>.ri.s<u>.ma.s<u> NOCODA/*COMPLEX

ink in.k<u> NOCODA

stress s<u>.t<o>.re.s<u> NOCODA/*COMPLEX

cake ke:.k<i> NOCODA

guide gai.d<o> NOCODA

(4)

While the candidate stress obeys MAX and DEP it violates the higher ranked markedness constraints since it has a coda and a complex onset. Stressu violates DEP as it displays an epenthetic vowel, how-ever, it is violating *COMPLEX *rst, hence the fatal violation mark. As the violations (two times DEP) of sutoresu are less violent than the *rst two and last two candidates it is the optimal form. In both sessu and se items are deleted to conform to NOCODA and *COMPLEX, at the same time, this is banned by MAX.

Together with segment substitutions and the emergence of epenthetic vowels stress assignment plays a signi*cant role in English loanwords. On the one hand it seems that these words are biased to an accented pattern contrasting native words, that are generally unaccented. On the other hand stress is realized in a completely di)erent way. Kubozono (2006) proposes that phonetic as well as phonol-ogical factors are relevant to account for this distinction. He assumes that “English pronunciation serves as the input to loanword accentuation’” and that this is “preserved as faithfully as possible.” The phonological factor is assumed to be responsible for the di)erent locus of the accentuation as we will see in (6). In the table in (5) we see a comparison of di)erent word types in TJ. SJ stands for Sino-Japanese words that are words that were adapted from Chinese centuries ago.

(5)#

# Word type and accent pattern in trimoraic nouns (N=7937) # #

From (5) we could also conclude that the older the word in TJ the less chance that it has accent. It might be that words loosen accentuation gradually in the course of time. Stress seems to shift as we see in (6) (exept for pakku). Diacritics represent accent.

(6)#

Word type Accent pattern

Accented (%) Unaccented (%)

Native (2220) 29 71

SJ (4939) 49 51

Loan (778) 93 7

English Japanese English Japanese

gríll g<ú>.ri.r<u> vítamin bi.tá.min

páck pák.k<u> Chrístmas k<u>.ri.s<ú>.ma.s<u>

reláy rí.re sándwich san.d<o>.ít.t<i>

gláss g<ú>.ra.s<u> píneapple pai.náp.pu.r<u> stréss s<u>.t<ó>.re.s<u> Eúrope yoo.róp.p<a> Líverpool r<i>.b<a>.púu.r<u> bádminton ba.d<o>.mín.ton

data from Kubozono (2006) data from Kubozono (2006)

(5)

Note that although Japanese prefer CV sequences, some consonants, being part of a geminate, are allowed in coda positions. Nasals may be in coda position anyhow, even at the end of words. Fur-ther, accent is sometimes realized on syllables containing epenthetic vowels, that is, on syllables that are not present in the original English words. In the next section we will see how epenthetic vowels and hence syllable structure interacts with stress assignment in English loanwords in TJ.

3# # Stress assignment.

As we have seen stress in English loanwords in TJ can have di)erent loci. Haraguchi (1991, p.24) distinguishes two types of loanwords: those that respect the original accent and those that conform to the Japanese accent rules. We will see, however, that the *rst type constitute a relatively small group. Any rule or constraint that is invoked to account for stress assignment in Japanese should take the mora as a descriptive unit, since Japanese is a mora language rather than a syllable lan-guage (Kubozono, 2001). McCawley (1968, p.134) proposes the ‘antepenultimate rule’ while he con-siders Japanese a counting syllable language” like Classical Latin, as opposed to “mora-counting mora languages” like Classical Greek. Kubozono (2006) con*rms that this rule is applied to words of all lexical strata in Japanese2 if accented.

(7)# Put an accent on the syllable containing the antepenultimate mora, or the third mora from the # end of the word.

Although Japanese should be considered as a pitch-accent language (accented words have an abrupt pitch fall), rules that can account for the shape of pitch are very similar to the accent rules of stress-accent languages such as English (McCawley, 1968, p.137). Still, I did not found out how native speakers of Japanese perceive accent in English words or whether they are totally oblivious to stress-accent. This is important, I think, as we would like to know what the exact input is as underlying form. Given the data in (5) we have reason to believe that there might be prosodic cue. If we take a look again at the words in (6) we see that (7) is applied correctly. However, there are some loan-words that show a di)erent pattern as we inspect (8).

(8)#

2 Traditionally, four strataare recognized: Yamato (native morphemes), Sino-Japanese (centuries-old loans

from Chinese, foreign (mostly from English) and Ideophonic (extensive system of sound-symbolic expressions) (Ito & Mester 2003, p.38).

English Japanese English Japanese

play p<u>.rée gray g<u>.rée

blue b<u>.rúu dry d<o>.rái

three s<u>.ríi twin t<u>.ín

(6)

The words in (8) share some features that may explain why they are exceptions. They all consist of two syllables: the *rst is a light syllable (L, one mora) the second is a heavy syllable (H, two moras). Further, we observe an epenthetic vowel in the *rst syllable. Statistical analysis con*rms the impor-tance of epenthetic vowels in the penultimate syllable (indicated by lH) in trimoraic loanwords as we take a look at (9).

(9)#

# Percentage of the accent pattern in (8) shown by each syllable structure

The table in (9) reveals that of 424 trimoraic loanwords only 8% show an accented pattern as ob-served in (8). Of the LH-group the words with an epenthetic vowel in the initial syllable favor this accent pattern almost categorically. If the *rst syllable of a word of the LH-group contains an under-lying vowel these syllables generally take accent. A few examples are given in (10).

(10)

Although Kubozono (2001) did not include data of HL syllables in the table in (9), in another study he presents statistical data that show that those words never have *nal accent (as in 9). Since HL syllables are included here this explains the di)ering percentages of the total of words bearing *nal accent (2% as opposed to 8% above).

(11)#

# Syllable structure and accent pattern in accented three-mora loanwords (N=722 words)

LH and lH LH lH LLL Total

23 of 146 4 of 124 19 of 22 12 of 278 35 of 424

(16%) (3%) (86%) (4%) (8%)

data from Kubozono (2001)

English Japanese English Japanese

Hawáii há.wai Irán í.ran

sedán sé.dan reláy rí.ree

púdding pú.rin

data from Kubozono (2006)

Syllable structure Accent pattern

Initial accent (%) Medial accent (%) Final accent (%)

LLL (275) 93 6 1

HL (325) 100 - 0

LH (122) 89 - 11

Total (722) 96 2 2

(7)

In sum, the antepenultimate rule seems to work most of the time and, following Kubozono (2006), it appears to be “fundamentally similar, if not entirely identical” to the compound accent rule (CA) that is assumed to be at work in four-mora (or more) native words. Such words are morphologically complex. The CA rule is originally from McCawley (1968, p.180) and is restated by Kubozono (2006) as in (12). N1 represents the left half of a compound.

(12)# a.# If N2 is either monomoraic or bimoraic, a CA falls on the *nal syllable of N1.

examples: # ti.no.mi + ko # # → # ti.no.mí-go# # ‘milk drinking, baby; baby at the breast’

# # ku.wa.ga.ta+mu.si# →# ku.wa.ga.tá-mu.si#‘stag beetle’

# b.# If N2 is three moras long or longer, a CA falls on the initial syllable of N2.

examples:# nó+ne.zu.mi# # →# no-né.zu.mi## ‘milk drinking, baby; baby at the breast’

# # sín+yo.ko.ha.ma# # →# sin-yó.ko.ha.ma# ‘milk drinking, baby; baby at the breast’

In terms of constraints Kubozono (2006) generalizes (12) as in (13a-b) and adds (13c) to include compound nouns of which the last syllable of N2 bears accent originally.

(13)# a.# Accent the rightmost non-*nal foot (Non*nality-foot, Edgemostness).

# b.# Within the rightmost, non-*nal foot, accent the syllable that is closer to the word-## # # internal morpheme boundary (Align-CA).

# c.# Keep the accent of the *nal member unless it is on the very *nal syllable (Max-accent, # # # Non*nality-syllable).

examples3:# á.ka+[kái]## →# [a.ká]-[gai] ## # ‘red, shell*sh; arch shell’

# # ten.zyoo+[ka.wá]#→# ten.[zyóo]-[ga.wa]# ‘ceiling, river; raised bed river’

(13a) is for example seen in ‘stress’ which becomes a four mora word in Japanese: su.[tó.re].su.4 The

only loanwords that disobey (13a) are typically words ending in LLH# and HLH# that get accent patterns as LĹH# (see ‘vitamin’ in (6)) and HĹH# as in /kan.gá.roo/ ‘kangaroo’. Kubozono assumes that (13a) could only work here if we posit a degenerate foot. What is striking, is that LLH# and HLH# words are subject to age related variation in TJ and that it seems that TJ accent is in fact moving away from (7) (and hence (13a)): TJ youth rather applies the accent rule for Latin and Eng-lish as stated in (14).

(14)# Place an accent on the penultimate syllable if this syllable is heavy: if it is light, place an ac-# # cent on the antepenultimate syllable.

Whereas elderly people pronounce ‘Amazon’ as /a.má.zon/ (LĹH#) and ‘cardigan’ as /kaa.dí.gan/ (HĹH#), younger people tend to pronounce these words as /á.ma.zon/ (ĹLH#) and /káa.di.gan/ (H́LH#). The situation isn’t as dramatic as it seems. Assuming, as we do here, that syllables can be

3 [ ] represent foot structure.

4 Shinohara (2000) de*nes the default accentuation as the head of a non-*nal bimoraic trochaic foot which

clari*es the accent on -to- rather than on -re-. Kubozono (2006) doesn’t include this information in (14a) or we should assume that -to- is “closer to the word-internal morpheme boundary”.

(8)

either L or H there are 23 possibilities for the last three syllables in a word. They are summed up in

(15).

(15)# a. HHH# b. HHL# c. LHH# d. LHL# e. HLH# f. HLL# g. LLH# h. LLL#

Applying (7) or (14) yields only a di)erent pattern in (16e) and (16g). In other words, the TJ accent rule is, following Kubozono (2006), “fundamentally similar to the accent rule of Latin and English.” In the table in (16) we see the results of the eight groups after these rule have applied.

(16)#

This change in accent rule application has consequences for word types as (16e) and (16g) only. Just as we have seen in (9) and (11) there seems to be something special with loanwords ending in LH. Elderly people act in accordance with the pattern of (7), as opposed to the younger people that fol-low the pattern of (14). As a consequence, we could predict that /bi.tá.min/ and /kan.gá.roo/ will be pronounced as /bÍ.ta.min/ and /kán.ga.roo/ respectively in the near future. Without scrutinizing the issue, an explanation could be that TJ youth are more familiar with English because of education and maybe globalization. If so, we could expect that they will be more faithful to the English words. It might be that at some point in time TJ speakers will pronounce a word such as ‘stress’ as

/su.to.ré.su/. The penultimate syllable should be lengthened then, just as in the adapting of French words: ‘stress’ is changed from LLL# to LHL# underlyingly. Yet, being only faithful to the phonology of English, that is, obeying rules such as (14), we would predict a pronunciation similar to what we hear today. This would support the dominance of (14). Another possibility is that /sutoresu/ might loose accent altogether. We have to be patient to *nd out what the future is of ‘stress’ in TJ.

In terms of OT (14) can be covered by the following constraints (Shinohara, 2000). (17)# Head-Left (HEAD-L): Trochaic feet.

# Align (F, R, PrWd, R) (ALIGN-R): Align the right edge of every foot with the right edge of a # # prosodic word (PrWd).

# Non-Finality (NONFIN): No prosodic head (accented foot (F) or accented syllable (S)) of PrWd # # is *nal in PrWd.

# Foot-Binarity (MAX/MINBIN): Feet are binary at some level of analysis. # Parse-Syllable (PARSE-S): Parse every syllable into a foot.

Shinohara (2000) demonstrates that “the basic ranking to generate antepenultimate accent” is as in (18).

(18)# Foot-Binarity, Head-Left, Non-Finality ⟩⟩ Align-Right ⟩⟩ Parse-Syllable.

Rule a. b. c. d. e. f. g. h.

(7) HH́́H# HH́L# LHH#́ LH́L# HĹH# H́LL# LĹH# ĹLL#

(9)

Further, we have seen that the constraint NOCODA,*COMPLEX andDEP are important in relation to the epenthetic vowels. Since it is much beyond the scope of this paper I su0ce to say that there should be one speci*c ranking that can account for the upper pattern in (16). Maybe NONFIN could be ranked lower as it prohibits *nal prosodic heads, or maybe a constraint concerning lengthening is ranked higher.The variation could then be explained in terms of slight changes of constraint rank-ings. As said before, TJ is a pitch-accent language (McCawley, 1968, p.137), however, McCarthy (2008) argues (for other reasons) that such languages should be treated as tone languages, in which tone is assumed to have autosegmental properties. He assumes that tone is a phonological object rather than a phonological relation like stress. Consequently, tone should be associated with seg-ments independently of any metrical structure, although McCarthy (2008) admits that there is evi-dence for Japanese foot structure: “The absence of stress cannot entail the absence of feet.” Accent rules we have seen so far involve conditions on metrical structure. Let us assume that TJ accent is in fact a matter of tone association contrary to to the analysis above. In adapting English words stress could be perceived as a cue for lexical tone, which is then associated along the rules of tone associa-tion that we assume as well for the moment. Haraguchi (1977) is an example of an attempt to ac-count for an autosegmental representation of tone in Japanese. I will not explore any autosegmental theory here, however, my point is that the actual change in (16) might be due to the fact that elderly TJ people only apply tone association rules, whereas TJ youth is loosing representation of tone be-cause of English in$uences. In other words, (7) might be a metrical description of a tone associating rule, while (14) is not. TJ could then be considered to be deviating further from other Japanese dia-lects such as Tsuruoka Japanese which are assumed to have tonal systems (Haraguchi, 2001). Nev-ertheless, in this section it has become clear that TJ accent can be formalized properly with con-straints that impose on metrical structure.

4# # Conclusion.

In TJ, the language we looked at in this paper, stress is placed on syllables of English loanwords where we would not expect it. After I highlighted the processes that are involved in adapting English loanwords in Japanese, namely, sound change, vowel insertion and stress assignment, I delved fur-ther into the last process. Remarkable is that English loanwords are generally more accented than words from other Japanese lexical strata. Perhaps the English input is adapted as faithfully as possi-ble. Still, there might be a chance that English loanwords get less accented in the decades to come. By looking at the rules that are assumed to be involved in TJ stress assignment we found an explana-tion for the di)erent loci of stress: it is predictable to a great extend. Only tirmoraic words ending in LH syllables behave inconsistently, though they are relatively small in amount. Further, four mora words of this kind are subject to change. On the whole, TJ word accent seems to be realized in a completely di)erent way from English, the assignment of stress by younger speakers appears to be based on the same rule that is assumed to be in use in English. The abundance of epenthetic vowels could be hold responsible for the fact that metrical structures change in loanwords and hence cause stress shift. Further studies could be done on the perception of stress. Do Japanese perceive English accent di)erently than Japanese accent? Such research might shed more light on the question to what extend Japanese stress should be motivated by an autosegmental theory.

(10)

5# # References.#

Haraguchi, S., 1977. The tone pattern of Japanese: an autosegmental theory of tonology. Kaitakusha, #

# Tokyo.

Haraguchi, S., 1991. A theory of stress and accent. (Studies in Generative Grammar 37). Foris, # Dordrecht.

Haraguchi, S., 2001. “The accent of Tsuruoka Japanese reconsidered.” In: Van de Weijer, J., # Nishihara, T. (Eds.), Issues in Japanese Phonology and Morphology. Mouton de Gruyter, # Berlin, New York, pp. 47–65.

Ito, J. & Mester, A., 2003. Japanese Morphophonemics: markedness and word structure. The MIT Press, # # Cambridge.

Kay, G., 1995. “English loanwords in Japanese.” World Englishes, 14:67-76.

Kenstowicz, M. & U)mann, C., 2006. “Epenthetic vowel quality in loanwords: empirical and formal # # issues.” Lingua, 116:1079-1111

Kubozono, H., 2001. “Epenthetic vowels and accent in Japanese: facts and paradoxes.” In: Van de # # Weijer, J., Nishihara, T. (Eds.), Issues in Japanese Phonology and Morphology. Mouton de # # Gruyter, Berlin, New York, pp. 113–142.

Kubozono, H., 2006. “Where does loanword prosody come from? A case study of Japanese loanword # # accent.” Lingua, 116; 1140-1170.

McCarthy, J.J., 2008. “The serial interaction of stress and syncope.” Natural Language & Linguistic # # Theory, 26:499-546.

McCarthy, J.J. & Prince, A., 1995. “Faithfulness and reduplicative identity.” In: Beckman, J., Walsh # # Dickey, L., Urbanczyk (Eds.), University of Massachusetts Occasional Papers, 18:249-384.

# #

McCawley, J.D., 1968. The phonological component of agrammar of Japanese. Mouton, Den Haag. Shinohara, S., 2000. “Default accentuation and foot structure in Japanese: evidence from Japanese # # adaptations of French words.” Journal of East Asian Linguistics, 9:55-96.

Figure

Updating...

References

Updating...

Related subjects :