5 The Morphology of English Loanwords: Mark Irwin

1 Mark Irwin
3
5 The morphology of English loanwords
4
6
1 Introduction
7
The study of Western loanwords in Japanese has a long history, with scholarly articles
8
dating to as far back as Ōtsuki (1884) and loanword dictionaries to Ueda et al. (1915).
9
The earliest research focusing on English loans into Japanese can be traced back to
10
Ichikawa (1929, 1930). Although Ichikawa published in English, for many decades
11
thereafter, scholarly output was predominantly in Japanese and confined to univer-
12
sity or research institute bulletins. International impact was low. It was not until the
13
last quarter of the 20th century that interest took off at a global level, with scholarly
14
articles appearing in English in journals, handbooks and conference proceedings:
15
Quackenbush (1977), Koo and Honma (1989), Hirata (1990), Hirozane (1992), Takagi
16
and Mann (1994), Katayama (1995). It is at this juncture also that theoretical analyses
17
of a morphological and morphophonological nature began to appear: J. Itô (1990),
18
Itô and Mester (1992), Kubozono (1999b, 2001b, 2002).
19
Although this chapter will deal specifically with English loans into Japanese, the
20
morphological processes to be described below are applicable to loanwords (see my
21
definition in Section 1.1) from any language. English borrowings have become so
22
dominant (Section 1.2), however, that there is a tendency to view any loanword as
23
having an English source. This not only means that many Japanese mistakenly think
24
the English for ‘part-time job’ and ‘bread’ are *arbite and *pan,1 but also that the
25
morphological processes described in this chapter can only have an English donor
26
input. This latter assumption, indeed, has given rise to the Japanese term waseieigo
27
和製英語 ‘English made in Japan’ – applicable just as equally to loans borrowed
28
from French, Russian, Korean or whatever – to describe a range of very different
29
phenomena: essentially all the phenomena described in this chapter, as well as
30
mora clipping (Section 3.1), semantic change (Irwin 2011a: 153–155), and any form
31
of irregular phonological adaptation (Section 1.3), especially auditory and spelling
32
loans (Irwin 2011a: 76–80). The term waseieigo goes back to at least Umegaki (1944:
33
304). Although various categorizations – some more thorough than others – have
34
been proposed (Ishiwata 1976; Tanabe 1990; Hirai 2003; Shibasaki, Tamaoka and
35
Takatori 2007; Suzuki 2008), the term has suffered from chronic overuse and a
36
deleterious lack of definition in much academic literature.2 Needless to say, I will
37
forgo it in the discussion below.
38
39
1 Japanese arubaito from German Arbeit ‘work’ (with subsequent semantic narrowing); Japanese paN
40
from Portuguese pão ‘bread’.
41
2 The non-academic literature is even more guilty in this respect. In particular, mass-market publi-
42 cations aimed at the hugely lucrative English education market bandy the term around with total
impunity.
(Unicode 9 16/12/15 17:59) WDG-New (170mm240mm) DGMetaSerifScience (OpenType) 0000 pp. 161–198 1668 Kageyama_05_Irwin (p. 161)
162 Mark Irwin
1 Before considering the morphology of English loanwords in Sections 2 and 3,

2 some preliminaries are necessary. I consider the definition of a loanword in Section
3 1.1, statistical and attestatory detail in 1.2, issues of phonological adaptation and
4 phonemicization in 1.3, and loanword compounding in general in 1.4.
5
7 1.1 Definitions
8
9
Modern Japanese vocabulary is conventionally divided into four major strata: native
10
(wago 和語, etc.), mimetic (giongo 擬音語, gitaigo 擬態語, giseigo 擬声語 ), Sino-
11
Japanese (kango 漢語 , etc.) and loan (gairaigo 外来語, shakuyōgo 借用語, yōgo
12
洋語).3 A number of definitions have been proposed for the loan stratum over the
13
years (Itô and Mester 1999; NKD 2000–2002; Sugimoto 2007: 408, inter alia). How-
14
ever, a robust definition must eschew extraneous factors such as orthography and
15
instead focus on the following three factors:
16
CHRONOLOGY: Sino-Japanese is also, technically, a stratum comprised mostly of
17
borrowings, taken exclusively from Chinese. Crucially, however, it is felt by Japanese
18
speakers to be different from the loan stratum and exhibits very different phono-
19
tactics. Sino-Japanese borrowing was largely complete by the 13th century.4 The
20
Japanese language first came into contact with Western languages in 1543, when
21
Portuguese sailors landed on Tanegashima in the Ōsumi islands (now Kagoshima
22
prefecture).
23
24
PHONOLOGICAL ADAP TATION : A prerequisite for loanword status is that all donor
25
phones be assimilated into Japanese phonology (Section 1.3). As the Japanese pho-
26
nemic inventory is comparatively impoverished and its phonotactics comparatively
27
constrained, phonological adaptation can be severe. Loans whose phones remain
28
unassimilated and pronounced ‘in a foreign manner’ must be considered gaikokugo
29
外国語 ‘foreign words’.5
30 INTELLIGIBILITY: Any loanword whose meaning is unintelligible to the general
31 Japanese speaker community may, arguably, be considered not to have been fully
32 incorporated into the lexicon. The intelligibility factor is, however, highly problematic.
33 Although a soccer enthusiast may be au fait with the technical term ofusaido+toraQ pu
34
35
3 The Japanese terms for the loan stratum have subtle differences in meaning: see Irwin (2011a: 7).
36 Non-stratal models have been proposed by Itô and Mester (1995, 1999) and Fukuzawa and Kitahara
37 (2005).
38 4 For more detail on the history of Sino-Japanese borrowing see inter alia Tōdō (1969); Frellesvig
(2010: 258–294); Yuzawa (2010).
39
5 For a detailed discussion of phonological adaptation see Irwin (2011a: 71–136). In all the examples
40
to follow, borrowed English words (English donor etyma) are given a dotted underline, → indicates
41 ‘adapted to Japanese’, and ← ‘adapted from English’. Thus, ‘screen → sukuriiN ’ means ‘English screen
42 is adapted to Japanese sukuriiN ’. Element boundaries in compounds are indicated with a plus sign.
The morphology of English loanwords 163
1 (← offside trap), what proportion of the speaker community must fail to comprehend
2 a loanword for it to be generally unintelligible? This argument applies not only to
3 technical terms, but also to loanwords used by particular age cohorts, ethnic groups,
4 dialect speakers, or socio-economic groups. It must also be remembered that it is not
5 uncommon for loans to have been intelligible in the past but be unintelligible in the
6 present, while many loans intelligible at the time of writing will fall out of use and
7 be rendered unintelligible to future Japanese speakers.
8 I thus follow Irwin (2011a: 10) and define a Japanese loanword as:
9
10 (1) a foreign word which has undergone adaptation to Japanese phonology,

11 has been borrowed into Japanese after the mid-16th century and whose
12 meaning is, or has been, intelligible to the general speech community.
13
14
15 1.2 Statistics and attestations

16
17
Loanwords make up 2%–12% of the Japanese lexicon by token, and 5%–35% by
18
type, with great variation apparent across media.6 Borrowings from English currently
19
comprise the vast bulk, although this has not always been the case. In the late 19th
20
and very early 20th century English donor words accounted for just over half of all
21
loans, though this proportion had grown to around 75% by the 1950s. For the last
22
half century, since the 1960s, it has hovered virtually unchanged at 80%–85% by
23
type, or 85%–90% by token (Irwin 2011a: 25–27). The most recent survey (Hashimoto
24
2006) cites English donor words at a level of 91% of all loanword tokens. At the
25
same time, the share of other donor languages has plummeted, often drastically
26
(e.g. Dutch and Portuguese to only 1–2%). French donor words currently make up
27
around just 5% of all types (M. Itō 2003: 47; NINJAL 1987: 65) and German a mere
28
4% of all tokens (Hashimoto 2006).
29
The oldest English loanword is probably gere+borotaN ‘Great Britain’ (now
30
gureeto+buriteN ), found in the early 17th century diplomatic chronicle Ikoku Nikki
31
(Chronicle of Foreign Countries), compiled by the Zen monk Ishin Sūden.7 The
32
English had a trading post at Hirado (modern Nagasaki prefecture) between 1613
33
and 1623. English loans did not begin to appear in any number until the mid-19th
34
century, contemporaneous with the appearance of sailors’ pidgins in Nagasaki and
35
Yokohama (Atkinson 1879; Daniels 1948; Inoue 2007) and with the arrival in Uraga
36
37
6 Based on 10 statistical analyses conducted across a range of media since 1980. See Irwin (2011a:
19) for a list (dictionary analyses are ignored).
38
7 By 1604, James I of England and VI of Scotland was styling himself ‘King of Great Brittaine, France
39
and Ireland’. Patterns of phonological adaptation indicate that gere+borotaN cannot be derived from
40 a Portuguese, Spanish, or Dutch donor etymon, the only other European donor language candidates
41 of the time.
42
164 Mark Irwin
1 in 1853 of US Commodore Perry and his Black Ships. An influx of loanwords related
2 to American trends in sport, fashion, and music in the early 20th century saw the
3 donor variety shift from the UK to the US. The post-war US occupation further
4 strengthened this drift, which has since intensified further with globalization and
5 Japanese government education policy. It is no exaggeration to say that any English
6 word or phrase is currently fair game for borrowing, with politicians and bureaucrats
7 frequently criticized for being overly prone to reeling off loanwords for purposes
8 obfuscatory, mystificatory, or abstruse: e.g. iN šideN to ‘incident’, daibaašiti ‘diversity’,
9 suteekuhorudaa ‘stakeholder’.8
10
11
12 1.3 Phonological adaptation and phonemicization

13
14
This huge influx of English loanwords has meant that traditional “conservative”
15
phonemicizations (Yamaguchi 2007: 16; Tsujimura 2007: 22–37) are no longer tenable.
16
I thus employ the “contemporary” phonemicization of Japanese shown in Figure 1.
17
Similar analyses have been posited by Vance (1987; 2008), Irwin (2011a: 72–74) and
18
Tranter and Kizu (2012). Figure 1 contains 21–23 consonant phonemes,9 depending
19
on the inclusion of ŋ, whose phonemic status is open to debate (Vance 2008: 214–
20
222), and v, generally realized [β], which most speakers have great difficulty pro-
21
nouncing and found only in highly innovative pronunciations. Other phonemes
22
requiring comment are:
23
– the voiced velar plosive g is typically [ɡ], although may be [ŋ] or [ɣ] for some
24
speakers, chiefly word-internally. See Kindaichi ([1942] 1967); Hibiya (1995, 1996,
25
1999); Labrune (2012a: 78).
26
– the affricates c z č j are realised [ts dz cɕ ɟʑ].
27
– the fricatives f š are [ɸ ɕ].
28
– h is [ç] before i y, but [h] elsewhere.
29
– the mora obstruent Q is the initial segment of a geminate consonant and has a
30
variety of realizations.
31
– the mora nasal N also exhibits a variety of realizations, including [m n ŋ ɲ N ].
32
See Vance (2008: 96–112).
33
34
35
36
37
8 For greater detail on the history of English donor gairaigo than space here can provide,
see Umegaki (1963: 71–95); Loveday (1996: 59–76); Yamada (2005: 73–196); Irwin (2011a: 53–61).
38
For a summary of attitudes towards (mainly English) gairaigo, see NINJAL (2006); Irwin (2011a:
39
193–206).
40 9 Vowel phonemicization is identical in both conservative and contemporary phonemicizations and
41 not shown in Figure 1. The five vowel phonemes /a e i o u/ are broadly realized [ɑ e i o ɯ].
42
9
Figure 1: The Japanese consonants: A contemporary phonemicization
10
11
Japanese is a mora-timed language, with moras coming in three types: the
12
simple mora (C)(G)V (where G = an approximant glide y or w) and the two special
13
moras Q and N . However, the syllable also has an important role to play (McCawley
14
1968; Kubozono 1999b).10 Syllables come in three types: light, heavy, and super-
15
heavy. Light syllables consist of a single simple mora (C)(G)V. Heavy syllables
16
are bimoraic and consist of a simple mora followed by a vowel or a special mora:
17
thus (C)(G)VV, (C)(G)VQ or (C)(G)VN . Superheavy syllables are trimoraic and highly
18
marked: (C)(G)VVQ , (C)(G)VVN or (C)(G)VNQ . Syllable type has a crucial role to play
19
in compound clipping (Section 3.1).11
20
As noted in Section 1.1, during the borrowing process English donor words
21
undergo phonological adaptation, by which donor phones are adapted to the Japanese
22
phonemic inventory. Most phonological adaptation occurs in the form of phonic sub-
23
stitution, but epenthesis (of a vowel or Q ), as well as deletion, may also take place.
24
Most adaptation pathways have an orthographic source, although auditory sources,
25
while considerably less common, are also found. A detailed description of phonolog-
26
ical adaptation is provided in Irwin (2011a: 76–126).
27
28
29
1.4 Compounding
30
31 Any of the three vocabulary strata outlined in Section 1.1 can combine with one
32 another during compounding to produce hybrid compounds: e.g. the Sino-Japanese +
33 loan hybrid deN ši+buQ ku ‘electronic book’. Such hybrids will not be considered in this
34 chapter.
35
36
37 10 Some (e.g. Vance 2008: 132) have queried the status of the superheavy syllable in standard
(Tokyo) Japanese, while others have refuted the existence of the syllable altogether, most recently
38
Labrune (2012b).
39
11 Itô and Mester (2015) note an extremely rare fourth syllable type, the ‘ultra-superheavy’. These
40 come in one tetramoraic form only, (C)(G)VVNQ , and must contain a morphological juncture imme-
41 diately prior to a Q -initial particle (e.g. Q te) or Q -initial derived suffix (e.g. Q poi). They thus have no
42 role to play in compound clipping.
166 Mark Irwin
1 In the discussion to follow, an important distinction must be drawn between

2 ‘imported’ and ‘assembled’ English loanword compounds (Irwin 2011a: 143). I define
3 imported compounds as those that have been borrowed directly from English: e.g.
4 hunger strike → haN gaa+sutoraiki. Included among imported compounds are a
5 small number whose elements have been switched during the borrowing process:
6
7 (2) a. toaster oven → oobuN +toosutaa ‘toaster oven’

8
b. news flash → furaQ šu+nyuusu ‘news flash’
9
c. cabbage roll → rooru+kyabecu ‘cabbage roll’
10
11
d. off-season → šiizuN +ofu ‘off-season’
12
13
Assembled compounds, meanwhile, are those which have been constructed from
14
two independently borrowed non-compound English loanwords: e.g. resort + lover →
15
rizooto+rabaa ‘holiday romance’. Assembled compounds are frequently labelled in
16
the literature as waseieigo (see Section 1.1). Imported compounds greatly outweigh
17
assembled, although the morphosemantics of the latter are of great interest and
18
explored in Section 2.4. Whether a given compound is imported or assembled is
19
relevant for some truncatory processes (Section 3.2), but irrelevant for others (Sections
20
3.1 and 3.3).
21
The morphology of English loanwords in Japanese may be divided into trunca-
22
tory and non-truncatory phenomena. The latter will be outlined in Section 2. It is the
23
former which are of the greater linguistic interest, however, and analyses of these in
24
Section 3 form the bulk of this chapter.
25
26
27 2 Non-truncatory morphology
28
29 After looking at the core morphology of English loanwords in Section 2.1, discussion
30 will move on to affixes and boundedness in Section 2.2, rendaku in Section 2.3 and
31 morphosemantic compounding in Section 2.4.
32
33
34 2.1 Core morphology

35
36 The overwhelming majority of English loanwords are nouns12 and, like all Japanese
37 nouns, appear unmarked for number or case. Any English loan noun may also
38 appear as a verb by means of the native Japanese verbalizer suru ‘do’. The English
39
40 12 Some 95% of all loanwords are nouns according to NINJAL (1964). There is no reason to suppose
41 English loanwords behave differently, and little likelihood these figures have changed to any signif-
42 icant extent since the survey in question was carried out.
1 etymon may be a noun, verb, gerund, adjective, preposition or even abbreviation or

2 interjection:13
3
4 (3) Noun: harassment → harasumeN to (suru) ‘harassment (to harass)’

5
Verb: get → geQ to (suru) ‘score, win (to win, score, get)’
6
Gerund: heading → heQ diN gu (suru) ‘header (to head (in soccer))’
7
8
Adjective: live → raibu (suru) ‘live (to play live)’
9 Preposition: off → ofu (suru) ‘off (to turn off)’
10 Abbreviation: Q&A → kyuuaN doee (suru) ‘Q&A (to conduct a Q&A)’
11
Interjection: bye-bye → baibai (suru) ‘bye-bye (to say bye-bye)’
12
13
A very few English loanwords appear as either non-suru verbs or as adjectives.
14
These exhibit morphological alternation. As illustrated in (4a) and shown in non-
15
past form, non-suru verbs are formed by suffixing -r to the English loan, unless this
16
ends in -ru, in which case a suffix is unnecessary (4b). These non-suru verbs are
17
conjugated without exception as C-final stems: i.e. wikiQ ta ‘wikied’, never *wikita.14
18
Their preferred stem length is 2–3 moras and longer forms are thus frequently
19
clipped (4c), indicated here and in the rest of this section by a double strikethrough
20
( ). The verbalization process in (4) is highly marked, however. The vast bulk of
21
English loanword verbs appear as in (3) and many of those that do follow the
22
marked pattern in (4) possess doublets that follow the unmarked pattern in (3).
23
English loanwords appearing as adjectives do so as either nominal adjectives
24
(5a) or true adjectives (5c), the former being the unmarked pattern. These adjectives
25
undergo the standard morphological processes found in all Japanese adjectives and
26
may also be converted to adverbs via standard morphological rules (5b, 5d).
27
28
(4) a. diss → disu disu-r-u ‘to diss’
29
b. trouble → toraburu torabu-r-u ‘to have trouble’
30
31 c. taxi → takušii taku-r-u ‘to go by taxi’

32
33 (5) a. wild → wairudo wairudo-na ‘wild (out of control)’

34 b. wild → wairudo wairudo-ni ‘wildly’
35
c. pink → piN ku piN ku-i ‘pink, erotic’
36
d. pink → piN ku piN ku-ku ‘erotically’
37
38
39
40
13 Of course, in many cases the English etymon can belong to more than one part of speech. Here, I
41 have endeavored to use etyma that belong to just one.
42 14 For theoretical discussion of the r-suffix, see Chapter 1 (Kageyama and Saito, this volume).
168 Mark Irwin
1 Typically, any inflectional morphology an English etymon may possess has,

2 until fairly recently, been deleted during the borrowing process. Deletion of noun
3 and verb inflectional morphology is illustrated in (6a–e). Derivational morphology
4 may be deleted also, as exemplified in (6f–g), although such deletion has not been
5 as common. More recent borrowings have tended to preserve both the original in-
6 flectional and derivational morphology, as shown in (6h–k). This tendency appears
7 to be growing stronger due to an increased awareness among Japanese speakers of
8 English morphology. Indeed, a form of hypercorrection may even occur where an
9 English plural morpheme may be accreted during borrowing (6l–m).15
10
11 (6) a. pyjamas → pajama ‘pyjamas’

12
b. women’s lib → uumaN +ribu ‘women’s lib’
13
c. Rubik’s Cube → ruubiQ ku+kyuubu ‘Rubik’s Cube’
14
15
d. corned beef → kooN +biifu ‘corned beef’
16 e. happy ending → haQ pii+eN do ‘happy ending’
17 f. engagement ring → eN geeji+riN gu ‘engagement ring’
18
g. lucky seventh → raQ kii+sebuN ‘lucky seventh (inning)’
19
h. needs → niizu ‘needs, demands’
20
21 i. informed consent → iN foomudo+koN seN to ‘informed consent’

22 j. idling → aidoriN gu ‘(engine) idling’
23
k. empowerment → eN pawaameN to ‘empowerment’
24
l. top → toQ pusu ‘top (garment)’
25
26 m. bun → baN zu ‘(hamburger) bun, bap’

27
28 Although relatively infrequent, reduplication can occur, usually accompanied by

29 truncation. In the native Japanese and Sino-Japanese strata reduplication typically
30 signals plurality or distribution. With loanwords, however, reduplication appears
31 to function in a manner similar to the mimetic stratum, signalling emphasis or
32 intensity:
33
34 (7) a. love → rabu+rabu → raburabu ‘lovey-dovey’

35 b. erotic → eročiQ ku+eročiQ ku → eroero ‘porno’
36
c. grotesque → gurotesuku+gurotesuku → guroguro ‘gross’
37
38
39
40
15 A hypercorrection in the sense that the plural of a countable English noun occurs with vastly
41 greater frequency than the singular. The examples given ignore donor nouns which typically occur
42 in pairs or only rarely in the singular.
1 In their verbalization and adjectivization strategies (3–5), the loanword and

2 Sino-Japanese strata – both borrowed, albeit at greatly different time depths –
3 behave similarly. More divergent strategies are exhibited by the mimetic stratum: its
4 verbalization strategies are the same as both the loanword and Sino-Japanese strata,
5 while its adjectivization strategies are more idiosyncratic. The native stratum is the
6 outlier: its r-suffix, is “semi-productive” (Itô and Mester 2015), while its adjectiviza-
7 tion strategies are, broadly, the opposite of the loanword and Sino-Japanese strata.
8 Verbalization and adjectivization strategies across all four vocabulary strata are
9 summarized in Table 1, along with examples. Markedness is indicated by means of a
10 crude 4–point scale, where 3 = ‘unmarked’, 2 = ‘somewhat marked’, 1 = ‘highly
11 marked’, and 0 = ‘non-existent’.
12
13 Table 1: Verbalization and adjectivization strategies across vocabulary strata

14
LOAN SINO - JAPANESE NATIVE MIMETIC
15
16 suru verbalization 3 3 3 3
17 kokuhaku suru tamuro suru jimejime suru
‘confess’ ‘hang out’ ‘be damp’
18
19 r-suffix 1 1 2 1
koku-r-u futo-r-u paku-r-u
20
‘cough up, fess up’ ‘get fat’ ‘lift, crib’
21
22
nominal adjectives 3 3 1 2
daiji-na čiisa-na gunyagunya-na
23
‘important’ ‘small’ ‘mushy’
24
true adjectives 1 1 3 0
25
šikaku-i čiisa-i
26
‘square’ ‘small’
27
28
29
2.2 Affixes and boundedness
30
31 Table 2 offers some examples of English loanwords, frequently prepositions or

32 affixes, functioning as prefixes or, less commonly, suffixes in Japanese. There are no
33 examples of infixes. In most cases, semantic change has occurred, as is clear from
34 the English gloss. In most cases also, the Japanese affixes are bound. Those few
35 that may also occur unbound frequently show a semantic difference between their
36 bound and unbound form (e.g. bound širubaa- ‘senior citizen, OAP’, but unbound
37 širubaa ‘silver’). A few affixes have not yet reached a sufficient level of produc-
38 tivity for inclusion in Table 2. These include the suffixes šiQ pu ← -ship (e.g. skin →
39 sukiN +šiQ pu ‘close physical contact, esp. between mother and child’) and -fuuru ←
40 -ful (e.g. heart → haato+fuuru ‘heart-warming’).
41
42
170 Mark Irwin
1 Table 2: English loanwords as affixes: A selection

2
AFFIX GLOSS EXAMPLES
3
my → mai ‘personal, mai+kaa ← car ‘private car’
4
one’s own’ mai+hoomu ← home ‘owner-occupied house’
5 mai+buumu ← boom ‘personal fad’
6
no → noo ‘un-, not’ noo+taQ či ← touch ‘steering clear (of a
7 potential problem)’
8 noo+kauN to ← count ‘not counted, not included’
noo+kaQ to ← cut ‘uncut (movie)’
9
PREFIXES
10 the → za ‘the only, the za+terebijoN ← television TV website

best’ za+šiN fonii+hooru ← Osaka concert hall
11
symphony hall
12 za+šinema ← cinema satellite movie channel
13
silver → ‘senior citizen, širubaa+šiito ← seat ‘priority seat’
14 širubaa OAP’ širubaa+saabisu ← service ‘care for the elderly’
15 širubaa+maN šoN ← ‘sheltered housing’
16
mansion
17 over → oobaa ‘surfeit, oobaa+doraQ gu ← drug ‘overdose’

18
complete’ oobaa+dokutaa ← doctor ‘overterm PhD candidate’
oobaa+neQ to ← net ‘reaching over the net
19
(volleyball)’
20
off → ofu ‘limit, price → puraisu+ofu ‘discount’
21
deduction’ book → buQ ku+ofu second-hand book store
22 chain
23 pay → pee+ofu ‘financial cap’
SUFFIXES
24 man → maN ‘person’ gag → gyagu+maN ‘comedian’

25 guard → gaado+maN ‘security guard’
salary → sararii+maN ‘white-collar worker’
26
27 -er → raa ‘obsession’ ketchup → kečaQ pu+raa ‘ketchupaholic’

Uniqlo → yunikuro+raa ‘Uniqlo freak’
28
Wiki → wiki+raa ‘Wikipedia nut’
29
down → dauN ‘reduction’ image → imeeji+dauN ‘damage to reputation’
30
speed → supiido+dauN ‘slowdown’
31 price → puraisu+dauN ‘reduction in price’
32
33
34
The prefix mai- is not restricted to the first person, but can be used with reference
35
to the second or third person also. The suffix -raa is a reanalysis based on the
36
large number of English donor words ending in <-ler, -rer, -lor > → -raa: e.g. Internet
37
Explorer → intaaneQ to+ekusupurooraa, counsellor → kauN seraa, hustler → hasuraa,
38
etc. Truncation is not uncommon with this suffix. A reanalysis similar to -raa can be
39
found with two other suffixes not listed in Table 2: -tariaN ← <(t)arian> and -čiQ ku ←
40
<(t)ic>. These suffixes rarely combine with English loans however, favoring instead
41
non-loan morphemes: e.g. jibe+tariaN ‘young person who squats on the ground
42
with others’ or maN ga+čiQ ku ‘manga-esque’.
1 The loanword stratum is not unique in providing affixes, and their behavior
2 parallels very closely that found in other strata: most are bound and infixes do not
3 occur at all (see Chapter 1 [Kageyama and Saito, this volume] for further detail).
4 Although not occurring as affixes, some English loanwords occur extremely
5 rarely, or never, unbound. Kaa ← car and sukuuru ← school, for example, appear
6 with overwhelming frequency as bound morphemes in loanword compounds: e.g.
7 kaa+diiraa ← car dealer, patorooru+kaa ← patrol car (= ‘police car’), sukuuru+zooN
8 ← school zone, samaa+sukuuru ← summer school. When an unbound morpheme is
9 required, the non-loans kuruma / jidooša ‘car’ and gaQ koo ‘school’ are preferred.
10
11
12 2.3 Rendaku
13
14
Perhaps the mostly widely researched phenomenon observed in Japanese compound-
15
ing is rendaku, or sequential voicing, where the initial voiceless obstruent of a
16
non-initial element may undergo voicing (Vance 2015). However, it has been well
17
established (Nakagawa 1966; Kubozono 1999a; Takayama 2005; Tamaoka et al. 2009;
18
Irwin 2011a: 150–153) that rendaku does not occur to any extent in the loanword
19
stratum. Here, it is restricted to only a very few Portuguese loans borrowed in the
20
17th century and a sole compound containing a truncated English loanword,
21
the now obsolete akageQ to ‘country bumpkin’ (lit. ‘red blanket’), where keQ to is the
22
mora-clipped (see (14) below) output of buraN keQ to ← blanket.
23
24
25
2.4 Morphosemantic compounding
26
Many assembled compounds exhibit severe semantic divergence from the English
27
donor elements from which they are composed. I will term this phenomenon ‘mor-
28
phosemantic compounding’. When encountered by English-speaking, or English-
29
conversant, learners of Japanese, morphosemantic compounds are frequently parsed
30
incorrectly. As Daulton (2008: 19) has noted, this, combined with the fact that
31
Japanese learners of English are frequently inclined to presume morphosemantic
32
compounds are correct English (i.e. imported rather than assembled), leads to diffi-
33
culties with foreign language learning in both directions.
34
I do not include in my definition of morphosemantic compounds those assembled
35
compounds containing an English loanword functioning as an affix (Section 2.2).
36
English loanwords functioning as affixes have attained semantic stability and such
37
stability generally precludes incorrect parsing: an English-speaking, or English-
38
conversant, learner of Japanese has become aware of the meaning of such affixes
39
through having encountered them frequently across a range of compounds.
40
With a few exceptions, morphosemantic compounds are restricted to five broadly
41
definable semantic domains: marketing and media (8), technology (9), automotive
42
172 Mark Irwin
1 (10), sports (11), and sex (12). The motivation behind morphosemantic compounding
2 in these domains is not hard to ascertain. With the first three – media, technology,
3 and automotive – the motive is prestige. English loans are perceived by Japanese
4 speakers as having high status. TV and radio programs, websites, advertisements,
5 high-tech merchandise, and high-end consumer goods, such as automobiles, are
6 heavy users. Need also plays a role: many new technologies and media have no
7 non-loan Japanese equivalents.16 The sheer number of English loans which can be
8 encountered in these three domains (JKS (ed.) 2013) provides ample reproductive
9 fodder for morphosemantic compounding. In the domain of sports, the motive is
10 perhaps best articulated as ‘power’. The prestige accorded the English language
11 and English loans confers a sense of power on morphosemantic coinages, and
12 power is a sine qua non of the sporting world.17 Finally, the prime motivation with
13 the domain of sex (as well as the sex industry and sexual bodily functions) is euphe-
14 mism. A similar trend is found in English, where Latin and French borrowings are
15 especially prevalent. Here, morphosemantic compounds can contain elements of
16 otherwise exceptionally low frequency to further obfuscate meaning and heighten
17 euphemism. The few morphosemantic compounds which fall outside these five
18 domains (8)–(12) also tend towards euphemistic usage and include green + car →
19 guriiN +kaa ‘first-class railway carriage’, ceremony + hall → seremonii+hooru ‘funeral
20 hall’ and hello + work → haroo+waaku ‘government job seekers bureau’.
21
22 (8) time + service → taimu+saabisu ‘special offer, special sale’

23 golden + hour → goorudeN +awaa ‘prime time TV’
24 mic + location → maiku+rokeešoN ‘on-the-spot broadcasting’
25
(9) catch + phone → kyaQ či+hoN ‘call waiting’
26
plus + screwdriver → purasu+sukuryuudoraibaa ‘Phillips screwdriver’
27
ten + key → teN +kii ‘numeric keypad’
28
29 (10) gasoline + stand → gasoriN +sutaN do ‘petrol station, gas station’

30 front + glass → furoN to+garasu ‘windscreen, windshield’
31 door + engine → doa+eN jiN ‘auto open-and-close (taxi)
32 door’
33
34
35 16 Within the automobile domain, another possible motivation may be donor confusion. Many car
36 parts, as well as automobile-related goods and accessories, differ markedly across English varieties:
the gulf between the UK and the US in this respect is huge (Hargraves 2003: 179–187). It can be
37
unclear to Japanese speakers what the ‘correct’ English term actually is and morphosemantic com-
38
pounding can thus be seen as a fudge.
39 17 Note that in another aspect of sport, fair play, English loans are rarely encountered. The Japanese
40 notion of what constitutes gentlemanly conduct or sportsmanship does not require English loan-
41 words for adequate expression.
42
1
(11) out + course → auto+koosu ‘front nine (golf)’
2
toss + batting → tosu+baQ tiN gu ‘pepper drill (baseball)’
3
point + getter → poiN to+geQ taa ‘goal scorer (soccer)’
4
5 (12) love + hotel → rabu+hoteru ‘hotel renting rooms by the hour’

6 delivery + health → deribarii+herusu ‘prostitute call-out service’
7 blue + day → buruu+dee ‘day when one is menstruating’
8
9 Many of the morphosemantic compounds illustrated in (8)–(12) undergo a range of

10 truncatory processes and it is to these that I turn in the following section.
11
12
3 Truncatory morphology
13
14 Three truncatory processes occur with English loanword compounds: compound

15 clipping (Section 3.1), morpho-orthographic truncation (Section 3.2), and ellipsis
16 (Section 3.3). These are illustrated in Table 3 by means of the imported loan
17 compound dejitaru+kamera ← digital camera. The first two of these processes, com-
18 pound clipping and morpho-orthographic truncation, are morphophonological phe-
19 nomena, while the last, ellipsis, is purely morphological.
20
21 Table 3: Truncation types

22
TRUNCATION T YPE INPUT OUTPUT
23
24 compound clipping (Section 3.1) dejitaru+kamera dejikame

25 dejitaru+kamera dejika
26
dejitaru+kamera dekame
dejitaru+kamera dejimera . . . . . . etc.
27
28
morpho-orthographic digital + camera
truncation (Section 3.2) ↓ ↓ DC
29
dejitaru+kamera
30
ellipsis (Section 3.3) dejitaru+kamera dejitaru
31
dejitaru+kamera kamera . . . . . . etc.
32
33
34 If a compound undergoes truncation via one of the three processes illustrated, it

35 typically does so via one and one only. If multiple truncated forms do occur, then
36 usage is skewed towards one of them. The unmarked word for ‘digital camera’ in
37 Japanese is dejikame, via the first of the truncatory options in Table 3. This clipped
38 form is encountered far more frequently than its full form dejitaru+kamera. Indeed,
39 of the three truncation processes to be examined, compound clipping is by far
40 the most frequent. However, other truncations of dejitaru+kamera may be found,
41 typically in youth speech, jargon, slang, and dialect, indicating that truncation pro-
42 cesses possess a strong sociolinguistic component. The other, highly marked, trunca-
174 Mark Irwin
1 tions of dejitaru+kamera shown in Table 3 are only some of the options available:
2 some of those shown may not actually be encountered, while others that may be
3 encountered are not shown. Truncation processes are extremely fluid and the trunca-
4 tions themselves frequently highly transient.
5 All three types of truncation shown may be hybrid, i.e. the compound input
6 consists of a mixture of English and non-English loanwords (e.g. ice + Dutch koffie
7 ‘coffee’ → aisu+koohii → aiko ‘iced coffee’) or a mixture of English loanword and
8 non-loanword (e.g. service + Sino-Japanese zaN gyoo ‘overtime’ → saabisu+zaN gyoo →
9 sabizaN ‘unpaid overtime’). Such hybrids will not be treated here.
10 For truncatory processes to operate implies an awareness on the part of the
11 Japanese native speakers applying the process that the loanword to be truncated is
12 a compound, despite this often being opaque in the case of imported compounds.
13 Such an awareness is the only explanation for the fact that, in compound clipping
14 (Section 3.1), moras are clipped either from the beginning or from the end of individual
15 elements (never from the middle, never at random) and that, in ellipsis (Section 3.3),
16 an entire element is deleted. For morpho-orthographic truncation (Section 3.2) we
17 must presume, in addition, knowledge of the original English spelling.
18
19
20 3.1 Compound clipping

21
22
All examples cited in this section occur in the author’s English Loanword Clipped-
23
Compound Integrated Database (hereafter “El Cid”). All El Cid compounds (n = 432)
24
contain two elements, both of which have been borrowed from English and appear
25
as either an entry (or as a redirection to an entry) in the Wikipedia Japan pages
26
posted since 2009 or in one of the following sources: Umegaki (1956, 1963); Arakawa
27
(ed.) (1977); JKS (ed.) (1977, 2003, 2010); Maruyama et al. (1992); Yonekawa (1997);
28
GJH (eds.) (2001); Watanabe, Skrzypczak and Snowden (2003); Shinmura (ed.) (2008);
29
SSDH (2010).18 As a database, I make no claims for El Cid being 100% comprehensive.
30
That said, I believe it to be an accurate reflection of compound clipping realities with
31
respect to the statistical trends to be discussed below.
32
In previous studies a variety of different names has been employed for what I
33
will term compound clipping. J. Itô (1990: 220) refers to the phenomenon as ‘com-
34
pound abbreviation’, while Labrune (2002: 102–104) calls the output of the process
35
‘compound abbreviated loanwords’. This fluid nomenclature reflects the general
36
lack of agreement on a formal subdivision for the truncatory morphology of loan-
37
words.
38
39 18 Compounds consisting of a truncated reduplicated loanword, as in (7), are not included in El

40 Cid. However, compounds containing elements which may have been borrowed from either English
41 or French are included: e.g. omurecu ‘omelette’, restoraN ‘restaurant’. See Irwin (2011a: 50) for
discussion.
42
1 Restricting discussion in line with El Cid content to compounds containing two

2 elements, I define compound clipping as:
3
4 (13) the reduction of a loanword compound to between 2 and 6 moras (μ), by

5 means of clipping at least one mora from at least one element. Moras are
6 clipped overwhelmingly from the rear of an element and the process applies
7 equally to both imported and assembled compounds.
8
9 There exists a range of possible outputs and these are expressed below as, for example,
10 1μ+1μ. In this case, the output is a 2μ truncation composed of two elements both of
11 which have been clipped back to one mora only. Compound clipping is illustrated in
12 Table 4.
13 Compound clipping should not be confused with mora-clipping. The latter is a
14 purely phonological phenomenon, as illustrated in (14a): see Irwin (2011a: 131–136)
15 or Kubozono (2015) for further examples. In mora-clipping the syllable plays a major
16 role (J. Itô 1990; Itô and Mester 1992; Labrune 2002), whereas in compound clipping,
17 as we shall see below, the syllable plays only a bit part. With mora-clipping, supra-
18 segmental factors (i.e. accent) are critical in determining where a loanword will
19 be clipped (Labrune 2002; Irwin 2011b); with compound clipping, suprasegmental
20 factors appear to be irrelevant. This does not mean, however, that English loanword
21 compounds may not undergo mora-clipping. Although such cases are rare, the
22 examples in (14b) testify to their existence. Such truncation falls outside my defini-
23 tion of compound clipping in (13) and will not be treated further.
24
25 (14) a. chocolate → čokoreeto → čoko

26 television → terebijoN → terebi
27
b. convenience store → koN binieN su+sutoa → koN bini
28
concentric plug → koN seN toriQ ku+puragu → koN seN to
29
30
It is clear from Table 4 that the unmarked clipping process is truncation to a 2μ+2μ
31
type: such types comprise 72% (313/432) of El Cid. The only other type to occur to any
32
significant extent is the 2μ+1μ type, which makes up 16% (67/432) of the database.
33
Other types are extremely rare and occasionally obsolete: e.g. the two 1μ+1μ types
34
shown in Table 4, moga and bea. That compound clippings were heavily skewed
35
towards 2μ+2μ types was pointed out, without the benefit of a database, as long
36
ago as Umegaki (1963: 108–109).
37
In the vast majority of cases, the clipped elements of which clipped compounds
38
are composed do not exist as unbound morphemes. In other words, pi and bairiN are
39
40
41
42
176 Mark Irwin
1 Table 4: Examples of compound clipping

2
T YPE DONOR ENGLISH JAPANESE INPUT TRUNCATED ENGLISH GLOSS EL CID
3
OUTPUT %
4
1μ+1μ modern + girl modaaN + gaaru moga flapper 1.2%
5
base + up beesu+aQ pu bea basic pay increase
6
1μ+2μ light novel raito+noberu ranobe light novel 2.1%
7
mail address meeru+adoresu meado email address
8
1μ+3μ picture + crossword pikučaa+kurosuwaado pikurosu nonogram (logic puzzle) 1.4%
9 Pacific + league pašifiQ ku+riigu pariigu Pacific (baseball) league
10 2μ+1μ Brad Pitt buraQ do+piQ to burapi Brad Pitt 15.5%
11 smart phone sumaato+hoN 19 sumaho smart phone
12 2μ+2μ unique + clothing yuniiku+kuroojiN gu yunikuro Uniqlo 72.5%

costume + play kosučuumu+puree kosupure cosplay
13
2μ+3μ cutlet + sandwich kacurecu+saN doiQ či kacusaN do wiener schnitzel sandwich 4.2%
14
husband + hunt hazubaN do+haN to hazuhaN to husband hunting
15
2μ+4μ self + motor serufu+mootaa serumootaa self-starting motor 0.7%
16
3μ+1μ black + lagoon buraQ ku+raguuN burakura Black Lagoon (computer game) 0.5%
17
3μ+2μ cool + business kuuru+bizunesu kuurubizu dressing down for the heat 1.2%
18 auto + bicycle ooto+baišikuru ootobai motorbike
19 3μ+3μ plastic tile purasučiQ ku+tairu purasutairu plastic tile 0.5%
20 4μ+2μ bilingual + girl bairiN garu+gyaru bairiN gyaru young female returnee 0.5%
21
22
23
found only bound in compounds, never as unbound mora-clipped forms (14) of the
24
loans pikučaa ‘picture’ and bairiN garu ‘bilingual’.20
25
The most favored total length for clipped output is 4μ (74%, 321/432), although
26
3μ output is not uncommon (18%, 76/432). Both initial (401/432, 93%) and final ele-
27
ments (329/432, 76%) are clipped most often to 2μ, this proclivity being far stronger
28
among initial elements. In terms of element balance, nearly three-quarters of El Cid
29
compound clippings have both elements clipped to an identical length. As far as the
30
remaining quarter is concerned, a longer first element (76/432, 18%) is favored over a
31
longer final element (36/432, 8%). All these trends are summarized in Figure 2.
32
The dominant 2μ+2μ output witnessed for compound clipping, as well as the
33
fact that both initial and final elements are most commonly clipped to 2μ, are
34
all yet further examples of the importance of the foot (=2μ) in Japanese phonology,
35
morphology, and morphophonology. Not only is the foot the most frequently found
36
length for native Japanese nouns (see Itô and Mester 2015), it is also the preferred
37
length for loanword back-clippings (Irwin 2011b; see (14a) for an example). It also
38
plays an important role in non-loan compound clipping (Shibatani 1990: 254–256),
39
40
41 19 Also, sumaato+foN .
42 20 The small number of exceptions include kacu ‘cutlet’ and saN do ‘sandwich’.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38 Figure 2: Analysis of compound clipping output

39
40
41
42
178 Mark Irwin
1 in the accentuation patterns found in compounds (Tsujimura and Davis 1987), in

2 rendaku immunity (Rosen 2003; Irwin 2009a), in mimetic reduplication (Poser
3 1990: 94–95; Hamano 1998: 25–38), in the sociolinguistic phenomenon of jazz argot
4 (Tateishi 1989; Itô, Kitagawa and Mester 1996), and in hypocoristic formation (Mester
5 1990; Poser 1990: 81–93).
6 The question remains, however, as to why approximately one-sixth of all clipped
7 compounds are of the 2μ+1μ type. Clearly, the importance of the foot means a similar
8 proportion of 3μ+1μ or 4μ+3μ types would be unexpected. But why 2μ+1μ in partic-
9 ular? Why not 1μ+2μ or 2μ+3μ?
10 One possible motivation (Kubozono 2002: 94) is that, in Japanese in general,
11 long vowels (VaVa) tend to be shortened in word-final position. Thus, in 2μ+1μ type
12 clipped compounds, the final element is clipped to 1μ in order to avoid a 2μ CVaVa
13 heavy syllable. Had they been clipped to the dominant 2μ+2μ output, 79% (53/67)
14 of 2μ+1μ types in El Cid would have had a final element consisting of a single heavy
15 syllable (Section 1.3), and 60% (40/67) a final element consisting of a CVaVa heavy
16 syllable (see Figure 3). On the other hand, a similar analysis of the actual 2μ+2μ out-
17 put found in El Cid raises difficulties in maintaining the argument that word-final
18 long vowels are disfavored. Here, the proportions were smaller: 40% (124/313) had
19 a final element consisting of a single heavy syllable and 8% (26/313) a final element
20 consisting of a CVaVa heavy syllable. Further, as shown in (15), there exist final ele-
21 ments in the Japanese input which may be clipped to either 1μ or 2μ, the 2μ variant
22 containing a final CVaVa heavy syllable. Here, neither the syllable structure, phono-
23 tactics, nor suprasegmentals of the initial element appear to play any role in the
24 choice of variant.
25
26 (15) a. geemu ← game

27 2μ+1μ sabaibaru+geemu → sabage ‘survival game’
28 2μ+1μ neQ to+geemu21 → netoge ‘internet game’
29 2μ+2μ retoro+geemu → retogee ‘retro computer game’
30 2μ+2μ mobairu+geemu → mobagee ‘mobile phone game’
31
b. boodo ← board
32
2μ+1μ pokeQ to+boodo → pokebo ‘add-on keyboard for mobile’
33
2μ+1μ sunoo+boodo → sunobo ‘snowboard’
34
2μ+2μ sunoo+boodo → sunoboo ‘snowboard’
35
2μ+2μ sukeeto+boodo → sukeboo ‘skateboard’
36
37
Another possible motivation, avoidance of a word-final mora obstruent, generally
38
illicit in Japanese, must be discounted. As well as accounting for only a small number
39
40
41 21 See DELETION OF THE MORA OBSTRUENT Q below for an explanation of the Q -deletion found in
42 this clipped compound.
1 (16%, 11/67) of 2μ+1μ types, there exist alternative strategies for word-final mora
2 obstruent avoidance which preserve an unmarked 2μ+2μ output (see DELETION OF
3 THE MORA OBSTRUENT Q below).
4 It would thus appear there are no transparent constraints or rules that can
5 account for a dominant 2μ+2μ type clipped compound further truncating its second
6 element to yield a 2μ+1μ type. There is no obvious explanation as to why 2μ+1μ
7 types should be the second most favored compound clipping type. It is likely we
8 are witnessing sociolinguistic variation based on some as yet unknown variable or
9 variables. Here, and with other truncatory morphology to be discussed below, it is
10 hoped future sociolinguistic research can shed greater light on a confused situation.
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25 Figure 3: Final element syllable types (hypothetical and actual)

26
27 Several further issues related to compound clipping require further discussion:

28
29
HALF- CLIPPINGS : Truncations where only one element is clipped – recall my defini-
30
tion of compound clipping in (13) – are termed ‘half-clippings’ (Irwin 2011a: 144).
31
These include pariigu, hazuhaN to, serumootaa, kuurubizu, ootobai, purasutairu and
32
bairiN gyaru (the unclipped element is underlined) in Table 4. Half-clippings are not
33
uncommon and comprise some 18% (77/432) of El Cid.
34
The motivation behind half-clipping, in many cases, appears to be preservation
35
of the foot. In fully three-quarters of all half-clippings (75%, 58/77) the unclipped
36
element is 2μ, this despite the fact that 1μ-elements are perfectly licit (1μ-elements
37
make up 11% (94/864) of all elements in El Cid). The strong trend towards preserving
38
the foot is even clearer when viewed from another angle: of all the 2μ-elements
39
appearing in the El Cid Japanese loan input, only 6% (4/65) were clipped to 1μ.
40
In other cases, the motivation behind half-clipping is obscure. In the remaining
41
one quarter of half-clippings, the unclipped element is typically 3μ (23%, 18/77) or,
42
180 Mark Irwin
1 rarely, 4μ (1%, 1/77) in length.22 Although in the majority of these cases (68%, 13/19)
2 clipping to 2μ would yield a heavy syllable, unlike mora-clipping (14a) there is no
3 constraint against clipped elements consisting of a single heavy syllable in com-
4 pound clipping. Thus, any argument that half-clipping finds its motivation in the
5 avoidance of a single heavy syllable must be rejected: there is no reason why we
6 should not find *hazuhaN , *serumoo, *kuubizu or *oobai for hazuhaN to, serumootaa,
7 kuurubizu, and ootobai. Indeed, being unmarked 2μ+2μ truncations, we should in
8 fact expect *hazuhaN , *serumoo, *kuubizu and *oobai as the optimal output. That
9 this does not happen is curious.
10 N ON - FINAL VOWEL DELETION IN NON - LIGHT SYLLABLES : Some 4% (19/432) of
11 El Cid clipped compounds undergo some form of non-final vowel deletion in addi-
12 tion to truncation. Such deletion occurs only within heavy or super-heavy syllables
13 (i.e. non-light syllables) and comes in two types.
14 The more common of the two (79%, 15/19) is deletion of the second vowel in an
15 element-initial heavy syllable, typically (13/15) in an initial element: e.g. birth control →
16 baasu+koN torooru → basukoN ; personal computer → paasonaru+koN pyuutaa →
17 pasokoN ; sailor pants → seeraa+paN cu → serapaN , etc. In all cases, this deleted
18 second vowel is identical to that preceding it, i.e. it cannot be claimed that the
19 heavy syllable contains a diphthong. This type of deletion may thus be termed
20 ‘vowel shortening’. Interestingly, initial-element vowel shortening occurs solely
21 with imported compounds, while the only two examples of final-element vowel
22 shortening occur with assembled compounds: hello + morning → haroo+mooniN gu →
23 haromoni (name of TV program); and bloomers + sailor (suit school uniforms) →
24 buruumaa+seeraa → burusera (shop selling school girl’s uniforms and underwear
25 complete with photo ID of wearer). There are no examples of vowel shortening occur-
26 ring in both elements of a clipping.
27 The less common type of non-final vowel deletion (21%, 4/19) occurs only in
28 super-heavy syllables ending in -auN . Here, rather than the syllable-final N , it is the
29 preceding u which is deleted (Kuwamoto 1998a; Kubozono 2001a). There are only
30 four examples in El Cid, including soundtrack → sauN do+toraQ ku → saN tora and
31 Blue Mountain (coffee brand) → buruumauN teN → burumaN . This type of deletion
32 occurs with equal frequency in both initial and final elements and across both
33 imported and assembled compounds.
34 The motivation behind the former type of non-final vowel deletion, vowel shorten-
35 ing, is unclear. On the one hand, all the elements in which vowel deletion operates
36 are reduced to 2μ, thus achieving the optimal element length of one foot. On
37 the other hand, however, there are examples of the second vowel in a non-final
38 heavy syllable not undergoing deletion, in spite of the fact that such deletion would
39
40 22 There are no unclipped elements (i.e. no El Cid Japanese loan input) 1μ in length. This is due to
41 the phonological adaptation pathways which apply during the borrowing process (Section 1.3) and
42 which mitigate in general against the creation of 1μ English loans. There exists only one, za ← the,
which I treat as an affix (Section 2.2).
1 yield an unmarked 2μ+2μ output: e.g. verification card → berifikeešoN +kaado →

2 berikaado, as well as kuurubizu and ootobai in Table 3.
3 With the latter type of non-final vowel deletion, that of u in super-heavy syllables
4 ending in -auN , the motivation is clearly one found across the Japanese language in
5 general, that of avoiding super-heavy syllables. Of course, such a syllable could just
6 as easily be avoided by deleting the syllable-final mora nasal N . Note, however,
7 an identical type of non-final vowel deletion can occur during the phonological
8 adaptation of English loanwords (Ichikawa 1929, 1930; Lovins 1973: 81–82; Kubozono
9 1995, 1999b; Irwin 2011a: 100–101): e.g. (sports) ground → guraN do; (cosmetic) foun-
10 dation → faN deešoN .
11 D ELETION OF THE MORA OBSTRUENT Q : A very small number (1%, 5/432) of
12 clipped compounds evince Q -deletion. Since, in Japanese in general, word-final Q is
13 licit only in the mimetic stratum and in emphatic forms, it is unsurprising that the
14 mora obstruent Q is deleted in cases where it would become a word-final mora. The
15 two cases where this occurs (16a) yield 2μ+2μ type truncations. However, deletion is
16 not the only strategy found for avoiding illicit word-final Q . Clipping the final ele-
17 ment to yield a 2μ+1μ output is also possible: see the examples in (16b). Why some
18 Japanese input follows the truncation pattern in (16a) while other input follows the
19 pattern in (16b) is unclear.23 Once again, it is possible we are dealing with an
20 unknown sociolinguistic variable.
21
22 (16) a. 2μ+2μ
23 American football → amerikaN +fuQ tobooru → amefuto (*amefuQ )
24 Harry Potter → harii+poQ taa → haripota (*haripoQ )
25
b. 2μ+1μ
26
Photoshop → foto+šoQ pu → fotošo (*fotošoQ, *fotošopu)
27
potato chips → poteto+čiQ pusu → poteči (*potečiQ, *potečipu)
28
29
The other three cases of Q -deletion in El Cid occur word-internally, despite the fact
30
that word-internal Q is perfectly licit in Japanese in general. Nevertheless, the moti-
31
vation behind its deletion is transparent. Were they to be clipped to the dominant
32
2μ+2μ type output, the examples in (17a) show that retention of initial-element-final
33
Q would result in either an illicit Q m, a quasi-illicit Q r, or a marked Q g gemination.24
34
35
36 23 A further potential strategy is Q → cu (Kuwamoto 1998b) under the influence of kana orthography,
37 where Q っ is written as miniscule cu つ. Although there are no examples in El Cid, an example of a
38 clipped hybrid compound where this strategy is apparent is: Sino-Japanese daN zeN ‘absolute’ + top
→ daN zeN +toQ pu (だんぜん + トップ) → daN tocu (だんトツ) ‘decisive, runaway’.
39
24 Geminate nasals are phonemically N m and N n. The existence of a geminate liquid Q r is moot
40
(Irwin 2011a: 73; Labrune 2012a: 136; Tranter and Kizu 2012: 270), hence my use of ‘quasi-illicit’.
41 Geminate voiced obstruents (Q b, Q d, Q g) are restricted to the loanword stratum and thus marked in
42 standard Japanese. Even here, they may be subject to devoicing (Arisaka 1940: 94; Quackenbush
1989; Vance 2008: 108–110; Kawahara 2011a, 2011b, 2012).
182 Mark Irwin
1 Evidence for this motivation is further strengthened by the examples in (17b), where
2 word-internal Q is not deleted and the resulting clipped compounds contain licit,
3 unmarked geminations. Disfavored geminations such as those in (17a) may, never-
4 theless, be dealt with elsewhere not by deletion, but by further clipping the initial
5 element to yield a 1μ+2μ output, as seen in (17c). This is despite the fact that, in
6 some cases, an unmarked 2μ+2μ output would contain an unmarked gemination,
7 such as Q s.
8
9 (17) a. (inter)net radio → neQ to+rajio → netoraji (*neQ raji)

10 (browser) bookmark → buQ ku+maaku → bukuma (*buQ ma)
11 (inter)net game → neQ to+geemu → netoge (*neQ ge)
12
b. hit parade → hiQ to+pareedo → hiQ pare
13
shocking pink → šoQ kiN gu+piN ku → šoQ piN
14
mid-century → miQ do+seN šarii → miQ seN ‘designer furniture’
15
16 c. Netscape → neQ to+sukeepu → nesuke (*neQ suke)

17 sex + friend → seQ kusu+fureN do → sefure (*seQ fure)
18
19 BLENDS : Here, the front rather than the rear – recall the definition in (13) – of
20 one element is clipped by at least one mora. Blends are infrequent, occurring in 6%
21 (24/432) of clipped compounds in El Cid. In all these cases bar one, it is the front
22 of the final element which is clipped.25 Further, in the vast majority of cases (88%,
23 21/24) the Japanese input is an assembled compound. No particular output is
24 dominant. The initial element is clipped to 2μ in only 71% (17/24) of blends and the
25 final element to 2μ in only 54% (13/24) of cases: compare the data for El Cid as a
26 whole in Figure 2. Overall, only 42% (10/24) of blends are of the 2μ+2μ type.
27 Some examples of blends are shown in (18). Here, the assembled compound
28 (18a) means ‘an apartment with a garage’, (18c) ‘a type of marathon run while carry-
29 ing food’ and (18d) ‘a Japanese working in Asia’.
30
31 (18) a. 1μ+3μ
32 motor + apartment → mootaa+apaato → mopaato
33
b. 2μ+2μ
34
child idol → čairudo+aidoru → čaidoru
35
36 c. 2μ+3μ
37 marathon + picnic → marasoN +piQ kuniQ ku → maraniQ ku
38
39
d. 3μ+2μ
40
Japanese + Asian → japaniizu+ajiaN → japaniaN
41
42 25 The sole exception is model + casual → moderu+kajuaru→derukaji ‘casualwear as worn by

models’. This blend also undergoes MORA SPLITTING .
1 MORA SPLITTING : Mora-splitting may occur only where the Japanese input con-
2 tains a kana digraph, i.e. a single mora written using two kana (see Section 3.2), the
3 second of which is written in miniscule: ジャ <ja>, ティ <ti>, フェ <fe>, etc. The mora
4 in question is split by clipping the second miniscule kana, resulting in a vowel
5 change: ja > ji, ti > te, fe > fu, etc. In the examples shown below, the kana digraph
6 is boxed:
7
8 (19) memory stick → memorii+sutiQ ku → memosute ‘USB stick,

9 flash memory’
10 メモリー + スティックメモステ
11
mass communication → masu+komyunikeešoN → masukomi ‘mass media’
12
マス + コミュニケーションマスコミ
13
14 differential gear → defarenšaru+gia → defugia ‘differential gear’

15 デファレンシャル + ギアデフギア
16
17 Mora-splitting is infrequent, occurring only 13 times in El Cid and accounting for

18 3% (13/432) of the database. Of these, 10 are 2μ+2μ types where all syllables are light.
19 In a nonce compound survey, Irwin (2009b) concluded that mora-splitting was
20 highly marked. The El Cid data, however, is inconclusive. If one confines output to
21 2μ+2μ types only, then there are 17 cases where the Japanese input contains a mora
22 digraph as the second mora in either the first or second element, i.e. where the
23 Japanese input is of the form λλλ. . .+λλ. . . or λλ. . .+λλλ. . . ,26 λ indicating a reduced
24 kana graph and λλ a kana digraph. Of these, 10 (59%) exhibit mora-splitting and
25 7 (41%) do not.
26 P ERNICIOUS HOMOPHONY: Compared to donor English, Japanese is phonemically
27 impoverished.27 Consequently, during phonological adaptation (Section 1.3), several
28 English donor phones may converge on a single Japanese phoneme: e.g. Eng. r and l
29 both → Jp. r; Eng. ʤ, ð (before certain vowels) and z (before certain vowels) all → Jp.
30 j; a range of English vowels → Jp. a (for greater detail see Irwin 2011a: 95–96). Homo-
31 phonous English loanword elements exist prior to compound clipping – the reduc-
32 tion occasioned by the clipping process itself only exacerbates homophony to the
33 extent that it may become pernicious. The case of the clipped element koN is
34 particularly well-known, deriving as it does from at least 11 different unclipped
35 English elements (Irwin 2011a: 145). Other examples include iN and paN :
36
37
38
26 There is no input of the form λλλ. . .+λλλ. . . , i.e. where a mora digraph appears as the second
39 mora in both the first and second element.
40 27 Japanese has 26–28 phonemes (see Figure 1), as against 40–45 for English depending on variety
41 and analysis.
42
184 Mark Irwin
1 (20) inter-college → iN taa+kareQ ji → iN kare

2 inferiority complex → iN ferioritii+koN pureqkusu → iN koN
3 instrument panel → iN sutorumeN to+paneru → iN pane
4 natural + punch → načuaru+paN či → načupaN (band)
5 running pants → raN niN gu+paN cu → raN paN
6 Smashing Pumpkins → sumaQ šiN gu+paN pukiN su → sumapaN (band)
7
8 Such pernicious homophony is a source of confusion not only for foreign learners of
9 Japanese, but also for Japanese native speakers. It does not appear to be the case
10 that the avoidance of pernicious homophony plays a role in deviations from the
11 unmarked 2μ+2μ output found in compound clippings.
12 Compound clipping is not confined to the loanword stratum, but occurs freely
13 in the Sino-Japanese stratum also: e.g. tookyoo+daigaku → toodai ‘Tokyo Univer-
14 sity’. Although it may be assumed the process was active – or at the very least
15 activatable – from the mid-19th century when English loans began to be borrowed
16 (Section 1.2), in fact the earliest attestations for any of the El Cid compound clippings
17 date as far back as only the late 1920s or early 1930s: e.g. zenesuto ‘general strike’,
18 omuraisu ‘rice-stuffed omelette’, moga ‘flapper’, haN suto ‘hunger strike’ (Arakawa
19 (ed.) 1977; NKD 2000–2002).
20 Finally, compound clippings have even been borrowed back into English:
21 ‘cosplay’ ← kosupure in Table 4 being one well-known example.
22
23
24 3.2 Morpho-orthographic truncation

25
26
Japanese possesses three ‘native’ scripts. Kanji – Chinese characters borrowed from
27
around the 5th century – and two moraic scripts that evolved from these characters
28
in the 9th century, hiragana and katakana.28 The conventional domains of use of
29
these scripts are highly complex: see Irwin (2011a: 166–168) for a summary. In brief,
30
though, one of the major domains of the katakana script is to write loanwords, and
31
loanwords are written predominantly in katakana script.
32
Recently, the Roman alphabet has been making inroads as a loanword script,
33
although not, as one might expect, as an alternative orthography for a full, or even
34
a truncated, loan. Thus, <needs> for (6h) or <idling> for (6j) are not making any
35
inroads on the conventional katakana orthographies of <ニーズ> and <アイドリング >.
36
Instead, a compound containing two or more English loan elements may be abbre-
37
viated through having its orthography transposed to the original Roman script, and
38
all but the initial letters of each element deleted. To my knowledge, no detailed
39
description or thoroughgoing analysis of this phenomenon has been published to
40
date. What follows is a first attempt at a synthesis.
41
42 28 Thus, none of these three scripts is ultimately ‘native’. See Seeley (2000) for a history of Japanese
writing systems.
1 This process – which is not obligatory – is illustrated in Figure 4, where the two
2 English loan elements forming the input are ofisu ‘office’ and redii ‘lady’ (STAGE ➊ ).
3 These are assembled (see Section 1.4) into ofisu+redii (STAGE ➋ ), the script trans-
4 posed to the original English (STAGE ➌), the non-initial letters of each element
5 deleted and the initials retained (STAGE ➍). When these initials are converted to
6 majuscule the process yields the morpho-orthographic truncation OL ‘female office
7 worker’ (STAGE ➎ ). In this example, the result is pronounced oo+eru (← O+L).
8
10
11
12
13
14
Figure 4: The morpho-orthographic truncation process
15
16
Two important caveats must be applied to the morpho-orthographic truncation
17
process illustrated in Figure 4. The first is that the STAGE ➎ output does not exist in
18
English, at least not as an acronym derived from an identical input.29 Where it does
19
exist and the input is identical, I assume it is a direct borrowing and the process in
20
Figure 4 has not taken place. This is the case with, for example, enu+jii+oo ‘NGO’,
21
pii+dii+efu ‘PDF’, ee+ii+dii ‘AED = automated external defibrillator’, pii+tii+ee
22
‘PTA’, jii+dii+pii ‘GDP’, ii+tii+šii ‘ETC = electronic toll collection’ or oo+bii ‘OB’,
23
all of which exist in English and must be borrowings. Thus, although on the surface
24
OB and OL appear orthographically similar, their underlying morphology is different.
25
Furthermore, when the corpus of Roman alphabet acronyms found in Japanese
26
is viewed as a whole (Irwin 2011a: 188–191), the former OB-type direct borrowings
27
significantly outweigh the latter OL-type morpho-orthographic truncations.
28
The second caveat is that the process in question is not one of translation. For
29
example, the names of political parties in Japan all possess Roman letter acronyms:
30
LDP, DPJ, JCP, etc. It is presumed these were created in Japan and are direct trans-
31
lations of the their Japanese originals ( jimiN too 自民党 ‘Liberal Democratic Party’,
32
miN šutoo 民主党 ‘Democratic Party of Japan’, nihoN kyoosaN too 日本共産党
33
‘Japanese Communist Party’, etc.). These acronyms are not, however, used in
34
Japanese and cases such as these will not be considered as instances of morpho-
35
orthographic truncation. Grey areas nevertheless exist, notably with the names of
36
corporations. For example, JNR (← <japan.national.railways>), the precursor of JR
37
(← <japan.railways>), dates back to 1948, but is likely – though I have no proof –
38
the American GHQ translation of the Japanese name niQ poN kokuyuu tecudoo 日本
39
40
41 29 i.e. although OL does indeed exist in English, it does so as an acronym for ‘offensive line’ in
42 American football, not as an acronym for ‘office lady’.
186 Mark Irwin
1 国有鉄道. The multinational NEC Corp. has a long history dating back to 1899, but
2 whether NEC is a translation of the Japanese corporate name niQ poN deN ki 日本電気
3 or an original morpho-orthographic truncation (← <nippon.electric.corporation>) is
4 unclear. Such problematic acronyms will, faute de mieux, be considered as morpho-
5 orthographic compounds in the discussion to follow.
6 Several other issues concerning the process schematized in Figure 4 require
7 further elucidation. These are formalized in Table 5, along with examples:
8
9 OMISSION OF STAGE ➊ Morpho-orthographic truncation may also be applied to

10 imported compounds. In such cases, STAGE ➊ is omitted: e.g. <home.page> → HP
11 or <commercial.message> → CM.
12
HYBRID INPUT STAGE ➊ may contain numerals: e.g. <basement.2> (see Table 5),
13
<3.livingroom.diningroom.kitchen> → 3LDK ‘(apartment with) three bedrooms, a liv-
14
ing room and a dining room/kitchen’. STAGE ➊ may also contain a non-loan ele-
15
ment: e.g. <nippon.telegraph.telephone> → NTT, <all.nippon.airways> → ANA. The
16
element <nippon> ‘Japan’ is commonly found in company names.
17
18 INCOMPLETE PROCESSING Here, one of the elements in the STAGE ➊ input fails
19 to proceed to STAGE ➋ and remains written, unabbreviated, in native script: e.g.
20 jee+araato ← Japan+alert’ (written < Jアラート>) ‘National Disaster Warning
21 System’.30 Incomplete processing may be combined with HYBRID INPUT (see above):
22 e.g. daburyuu+hai (written <W杯 >) ‘(soccer) World Cup’, where the English loan
23 input is the initial element waarudo (← world) and the incompletely processed input
24 the final element Sino-Japanese hai ‘cup’ written in kanji.
25
PRONUNCIATION ➎ output may be pronounced in one of two ways: (i) as an
STAGE
26
alphabet string, where the letters of the Roman alphabet are read according to their
27
phonologically adapted English names (e.g. ee-bii-šii-dii-ii for A-B-C-D-E, etc.);31 or
28
(ii) as if they were an English word. Option (i) is the unmarked pronunciation: e.g.
29
oo+eru for OL or tii+bii+esu for TBS.32 Examples of option (ii) include mekusuto
30
for MEXT (see Table 5) and mosu for MOS(Burger) (← <mountain.ocean.sun>). With
31
compounds containing numerals, the numeral may be read as either the English
32
loan (waN +eru+dii+kee for 1LDK ← <1.livingroom.diner.kitchen>, ‘(apartment with)
33
one bedroom, a living room and a diner-kitchen’) or as Japanese (bii+iči for B1 ←
34
<basement.1>, ‘lower ground floor’).
35
36
37
30 One could also argue here that japaN (← Japan) is employed so frequently it has become regular-
38
ized as the quasi-prefix J, read jee, meaning ‘Japan(ese), national, nationwide’.
39 31 Some speakers use Roman letter names adapted from other donor languages, most especially
40 German: e.g. dee for D.
41 32 The phonologically adapted English values of Roman letters are overwhelmingly bimoraic, or
42
one foot. See the discussion in Section 3.1.
1 SPELLING At STAGE ➌, the script is always transposed to the original English spell-
2 ing, never one of the standard romanization systems employed for writing Japanese
3 (e.g. Hepburn or kunrei). Thus, in Figure 4, the second element レディー is not trans-
4 posed to redii (or the like),33 leading to a STAGE ➍ <o+r>, and thus yielding the
5 incorrect STAGE ➎ output *OR, to be pronounced *oo+aaru (← O+R).34 Knowledge
6 of the original English spelling is therefore a prerequisite for morpho-orthographic
7 truncation. While incorrect STAGE ➎ output produced by incomplete or flawed
8 knowledge of original English spelling cannot be ruled out, I am unaware of any ex-
9 amples. Although found only infrequently, some morpho-orthographic truncations
10 may also be written in native katakana script: e.g. スイカ for SUICA (see Table 5).
11 Here, the operation of a further putative STAGE ➏ must be presumed, where the pro-
12 nunciation of the STAGE ➎ alphabet string is transposed back into native script.
13
14
NON - INITIALS AND REBUSES At STAGE ➍ , non-initial letters of an element may be
15
retained or an initial letter deleted. Such compounds are relatively uncommon: e.g.
16
<super.urban.intelligent.card> → SUICA (see Table 5).35 Further, two initials may be
17
combined rebus-fashion. Once again, such compounds are rare: e.g. MEXT, where
18
<C.S.S> → <X> (see Table 5).
19 There exist compounds whose derivation is opaque. For example, I have neither
20 heard nor read LCC or PK in English: they do not exist in my variety. Yet they do (or
21 did) exist in certain other varieties, or in technical jargon. It is unclear whether the
22 Japanese CG, LCC, and PK are direct borrowings from CG, LCC, and PK in certain
23 varieties and jargons and thus have not undergone morpho-orthographic truncation,
24 or whether <computer.graphics>, <low.cost.carrier> and <penalty.kick> have independ-
25 ently undergone the process beginning at STAGE ➋ (see OMISSION OF STAGE ➊
26 above).
27 Morpho-orthographic truncation is a productive process and any English loan
28 may serve as STAGE ➊ or STAGE ➋ input. Most truncations currently in use are of
29 recent coinage, with few created before the 1970s. Ascertaining their first appearance
30 is fraught with philological difficulties, however. Most are listed only in specialist
31 dictionaries, such as those used for the El Cid database (Section 3.1), and few appear
32 in corpora. The most comprehensive Japanese dictionary, NKD (2000–02), lists only
33 a handful, and offers attestation dates for even fewer: the first written attestation for
34 BG is 1960, for CM 1963 and for OL 1974. Older, now generally obsolete attestations do
35 exist however. Umegaki (1944: 321), for example, cites CM (← <communist.manifesto>),
36 os (← <old.style>) and bc (← <birth.control>), suggesting morpho-orthographic truncation
37
38
33 Japanese has no lateral phonemes and all donor laterals are adapted to Japanese /r/ (Irwin 2011a:
39
91). This /r/ is transcribed as <r> in all romanization systems.
40 34 In the majority of cases, however, transposition to the original English spelling and transposition
41 to a romanization system will in fact yield the same STAGE ➎ output.
42 35 In the case of SUICA, and indeed some other railway smart cards, the output is also an inten-
tional pun: see Hibiya (2012) for details.
11
21
12
22
17
15
31
13
1
14
27
25
32
23
19
16
2
41
18
37
35
33
29
26
7
5
42
24
3
10
28
39
36
38
34
20
9
6
8
4
30
40
Table 5: Examples of morpho-orthographic truncation
188
pronuncia-
pronunciation:
tion: (1) as (2) as
alphabet English
STAGE ➊ STAGE ➋ STAGE ➌ STAGE ➍ STAGE ➎ string word English gloss
Mark Irwin
ビジネスガールビジネス + ガール business.girl b.g BG bii+jii businesswoman

jii+ cluster of national
ゴルデンウイークゴルデン + ウイーク golden.week g.w GW
daburyuu holidays in early May
セックスフレンドセックス + フレンド sex.friend s.f SF esu+efu casual sex partner
コミュニケーショ communication. satellite broadcasting
c.s CS šii+esu
ン + サテライト satellite system
インターチェンジ inter.change i.c. IC ai+šii expressway junction
expressway parking
パーキング + エリア parking.area p.a. PA pii+ee
omission of stage ➊ standard process

area
とうきょう + ブロード tokyo.
東京ブロードキャス Tokyo Broadcasting
(Unicode 9 16/12/15 17:59) WDG-New (170mm240mm) DGMetaSerifScience (OpenType) 0000

キャスティング + シズ broadcasting. t.b.s TBS tii+bii+esu
ティングシズテム System (TV channel)
テム system
オール日本エアウェオール + にっぽ all.nippon.
a.n.a ANA ee+enu+ee ana ANA Corp.
hybrid input
ーズん + エアウェーズ airways
bii+ni
ベースメント2 ベースメント + 2 basement.2 b.2 B2 2nd lower ground floor
bii+cuu
Japan National Soccer
ジャパンリーグジャパン + リーグ japan.リーグ j . リーグ Jリーグ jee+riigu
League
ビクトリーゴールビクトリー + ゴール victory.ゴール v.ゴール Vゴール bui+gooru golden goal (in soccer)
processing
incomplete
ユーズド + カー used.カー u .カー Uカー yuu+kaa second-hand car
pp. 161–198 1668 Kageyama_05_Irwin (p. 188)

11
21
12
22
17
15
31
13
1
14
27
25
32
23
19
16
2
41
18
37
35
33
29
26
7
5
42
24
3
10
28
39
36
38
34
20
9
6
8
4
30
40
Table 5: (continued)
pronuncia-
pronunciation:
tion: (1) as (2) as
alphabet English
STAGE ➊ STAGE ➋ STAGE ➌ STAGE ➍ STAGE ➎ string word English gloss
スーパーアーバンスーパー + アーバ
super.urban. rechargeable railway
インテリジェントカン + インテリジェン s.u.i.ca SUICA suika
intelligent.card smart card
ードト + カード
マルティーアミニティマルティー + アミニ multi.amenity. double-decker bullet
m.a.x MAX maQ kusu
ーエクスプレスティー + エクスプレス express train carriage
ministry.
ミニステリエヂュケーミニステリ + エジュケ education. Japanese Ministry of
ションカルチャースーション + カルチャ culture. MECSST > Education, Culture,
noninitials and rebuses

m.e.c.s.s.t mekusuto
ポーツサイエンスー + スポーツ + サイ sports. MEXT Sports, Science &
テクノロジーエンス + テクノロジー science. Technology
technology
NPO enu+pii+oo non-profit organization
acquired immuno-
AIDS eezu
deficiency syndrome
(Unicode 9 16/12/15 17:59) WDG-New (170mm240mm) DGMetaSerifScience (OpenType) 0000

United Nations
UNESCO yunesuko Educational, Scientific
& Cultural Organization
direct borrowing ðprocess

does not applyÞ
The morphology of English loanwords
189
pp. 161–198 1668 Kageyama_05_Irwin (p. 189)

190 Mark Irwin
1 may be traced to as far back as before World War II. Note that os and bc are cited in
2 miniscule, rather than majuscule. One may thus speculate that in early cases of mor-
3 pho-orthographic truncation STAGE ➎ did not apply.
4 Frequency data is also problematic. Nakayama, Kiryū and Yamaguchi (2007: 377),
5 an analysis of all Tokyo editions of the Mainichi Shinbun published 1994–2003, list
6 only 68 Roman acronyms, the vast majority of which are direct borrowings and thus
7 irrelevant. The most frequent morpho-orthographic truncations are, in order, JR, NTT,
8 TBS, NEC, CM, JT, BS, CS, JCO, IC, JA, OL, JAS, CG, JTB, JAL, and JICA.36 Of these, all
9 but four (CM, IC, OL, CG) are corporations, organizations, or institutions, of which
10 more than half contain an initial J for ‘Japan’.
11 Of interest sociolinguistically is that taboo constraints do not apply (Allan and
12 Burridge 2006): in any society where English is the official or majority language,
13 taboo and quasi-taboo acronyms are scrupulously avoided.37 This is not the case
14 with morpho-orthographic truncation, as the examples in (21) attest. English native-
15 speakers involved in or dealing with these entities may feel extremely uncomfortable
16 or, when it comes to February 14th, enjoy the irony.
17
18 (21) COC Centre of Community (Japanese Ministry of Education project)

19 PIS Public Information Style (Corporation)
20 TOSS Teacher’s Organization of Skill Sharing
21 NIG National Institute of Genetics
22 VD Valentine’s Day
23
24 Processes broadly similar to morpho-orthographic truncation occur in other

25 Japanese vocabulary strata. Here, the process is identical to that outlined in Figure
26 4, with the caveat that the STAGE ➊ input is non-loan. Examples include 女子 + 高生
27 → joši.koosei → j.k → JK ‘highschool girl’ and 空気 + 読めない → kuuki.yomenai →
28 k.y → KY ‘out of touch with reality’ (lit. ‘can’t read the air’). Indeed, this non-loan
29 morpho-orthographic truncation is colloquially known as keewaišiki nihoN go ‘KY-
30 style Japanese’, after the second example cited. KY-style Japanese also employs
31 rebuses, such as majide.moo.muri → m.m.m → MMM > 3M ‘totally out of the ques-
32 tion’, and dates back to at least the 1960s. See Kitahara (ed.) (2008) or Blockbuster
33 and GRK (eds.) (2008) for further examples.
34 Areally, morpho-orthographic truncation is not limited to Japanese: elsewhere in
35 East Asia, the phenomenon occurs in Chinese (Liu 2001, 2002) and Korean (Lee and
36
37
36 No definitions are given by Nakayama et al. While it is safe to presume that, for example, JTB
38
refers only to JTB Corp. (← <japan.travel.bureau>), this is not necessarily the case with others,
39 such as IC (see Table 5), which may also refer to direct borrowings with an identical structure (← IC
40 ‘integrated circuit’). Their frequency ranking should thus be treated with caution.
41 37 Unless as a deliberate marketing ploy: e.g. the quasi-taboo branding FCUK for ‘French Connec-
42
tion UK’.
1 Ramsey 2000: 3; Kim-Renaud 2012: 164–165) and quite probably, though as yet undo-
2 cumented, in other languages employing non-Roman script.
3
5 3.3 Ellipsis
6
7
The third and final type of truncation found in English loanword compounds is
8
ellipsis. Here, an entire element, either the final (22a) or the initial (22b), is deleted.
9
The former process is considerably more common than the latter. There are also a
10
few rare cases where the middle element in a three-element English loanword is de-
11
leted: this is illustrated in (22c). The loan input may be either imported or assembled.
12
13
(22) a. superimpose → suupaa+iN poozu → suupaa ‘subtitles’
14
magic + pen → majiQ ku+peN → majiQ ku ‘marker’
15
half-caste → haafu+kaasuto → haafu ‘person of mixed race’
16 b. screwdriver → sukuryuu+doraibaa → doraibaa ‘screwdriver (tool)’
17 game + cube → geemu+kyuubu → kyuubu ‘(Nintendo) GameCube’
18 sewing machine → sooingu+mišiN → mišiN ‘sewing machine’
19
c. must-have + item → masuto+habu+aitemu → masutoaitemu ‘must-have’
20
soft + ice cream → sofuto+aisu+kuriimu → sofutokuriimu ‘ice cream in a cone’
21
ballpoint pen → booru+poiN to+peN → boorupeN ‘pen’
22
23
Differentiating ellipsis from mora-clipping (14) may be problematic in cases
24
where an imported English loan compound is opaque; i.e. where its status as a
25
compound is unclear to a Japanese speaker. This is often the case where the loan
26
in question is actually a prefix+base (rather than a compound) and the prefix is
27
infrequent:
28
29
(23) software → sofuto(+)wea → sofuto
30
infrastructure → iN fura(+)sutorakučaa → iN fura
31
forehand → foa(+)haN do → foa
32
33
Finally, as with compound clipping (20), ellipsis can give rise to pernicious homo-
34
phony:
35
36
(24) superimpose → suupaa+iN poozu → suupaa ‘subtitles’
37
supermarket → suupaa+maakeQ to → suupaa ‘supermarket’
38
superheterodyne → suupaa+heterodaiN → suupaa ‘superheterodyne receiver’
39
software → sofuto(+)wea → sofuto ‘software’
40
soft + ice cream → sofuto+aisu+kuriimu → sofuto ‘ice cream in a cone’
41
softball → sofuto+booru → sofuto ‘softball’
42
192 Mark Irwin
1
4 Conclusions and future research perspectives
2
3
This chapter has presented a comprehensive survey of the morphophonological and
4
semantic properties of English loanwords in terms of the word formation processes
5
in which they participate. It is hoped that future work on the morphology of English
6
loanwords will focus on the various truncation processes described in Section 3.
7
More specifically:
8
– The high fluidity witnessed in the truncation patterns illustrated in Table 3
9
requires thorough diachronic and synchronic analysis. What motivates the
10
skewing towards a particular truncation type? Are the factors involved socio-
11
linguistic, phonological, morphophonological, suprasegmental, a combination
12
of several of these, or something else entirely?
13
– The operation of these same processes appears to require an awareness by the
14
Japanese speaker that the object of truncation is a compound, despite in many
15
cases the lack of any linguistic signal to this effect. The operation of morpho-
16
orthographic truncation implies a further awareness of English spelling. These
17
observations throw up a number of interesting issues in need of further research.
18
– Possible motivations proposed in Section 3.1 for the prevalence of 2μ+1μ type
19
clipped compounds (Table 4, ex. (15)) are inadequate. The same applies to those
20
offered for NON - FINAL VOWEL DELETION IN NON - LIGHT SYLLABLES and for
21
DELETION OF THE MORA OBSTRUENT Q . Further research is required to bring to
22
light these unknown, probably sociolinguistic, variables.
23
– Morpho-orthographic truncation is a new, under-researched, highly fluid phe-
24
nomenon. The description offered in Section 3.2 is tentative and painted with
25
a broad brush. More data collection is required and any future corpus-based
26
analyses are certain to throw up theoretically important hypotheses and results.
27
28
29
30 Acknowledgments
31
32 The author wishes to express his thanks to Matthew Zisk for his helpful comments
33 on an earlier draft of this chapter.
34
35
36
References
37
38 Allan, Keith and Kate Burridge. 2006. Forbidden words: Taboo and the censoring of language.
39
Cambridge: Cambridge University Press.
Arakawa, Sōbei (ed.). 1977. Gairaigo jiten [Loanword dictionary]. Tokyo: Kadokawa Shoten.
40
Arisaka, Hideo. 1940. On’inron [Phonology]. Tokyo: Sanseidō.
41
42
1 Atkinson, Hoffman. 1879. Revised and enlarged edition of exercises in the Yokohama dialect. Yoko-
2 hama: Japan Gazette Office.
Blockbuster and GRK = Gendai Ryakugo Kenkyūkai [Research Group on Modern Abbreviations] (eds.).
3
2008. KYgo jiten [Dictionary of KY words]. Tokyo: Byakuya Shobō.
4
Daniels, F. J. 1948. The vocabulary of the Japanese ports lingo. Bulletin of the School of Oriental and
5 African Studies 12. 805–823.
6 Daulton, Frank. 2008. Japan’s built-in lexicon of English-based loanwords. Clevedon: Multilingual
7 Matters.
8
Frellesvig, Bjarke. 2010. A history of the Japanese language. Cambridge: Cambridge University Press.
Fukuzawa, Haruka and Mafuyu Kitahara. 2005. Ranking paradoxes in consonant voicing in Japanese.
9
In Jeroen van de Weijer, Kensuke Nanjō and Tetsuo Nishihara (eds.), Voicing in Japanese, 105–
10
121. Berlin: Mouton de Gruyter.
11 GJH = Gakken Jiten Henshūbu [Gakken Dictionary Editorial Board] (eds.). 2001. Pāsonaru eiwa
12 katakanago jiten [Personal English-Japanese and katakana word dictionary]. Tokyo: Gakushū-
13 kenkyūsha.
Hamano, Shoko. 1998. The sound-symbolic system of Japanese. Stanford: CSLI Publications and
14
Tokyo: Kurosio Publishers.
15
Hargraves, Orin. 2003. Mighty fine words and smashing expressions. Oxford: Oxford University Press.
16 Hashimoto, Waka. 2006. Asahi Shinbun shasetsu no gairaigo [Loanwords in Asahi Shinbun editorials].
17 Dōshisha Kokubungaku [Doshisha University Journal of Japanese Literatures] 64. 178–186.
18 Dōshisha University.
19
Hibiya, Junko. 1995. The velar nasal in Tokyo Japanese: A case of diffusion from above. Language
Variation and Change 7. 139–152.
20
Hibiya, Junko. 1996. Denasalization of the velar nasal in Tokyo Japanese: Observations in real time.
21
In Gregory Guy, Crawford Feagin, Deborah Schiffrin and John Baugh (eds.), Towards a social
22 science of language. Papers in honor of William Lavov, volume I: Variation and change in
23 language and society, 161–171. Amsterdam: John Benjamins.
24 Hibiya, Junko. 1999. Variationist sociolinguistics. In Natsuko Tsujimura (ed.), The Handbook of
Japanese linguistics, 101–120. Oxford: Blackwell.
25
Hibiya, Junko. 2012. Kotoba no tayōsei ni me o hiraku [Looking at the diversity of words]. In Junko
26
Hibiya (ed.), Hajimete manabu shakaigengogaku [Introduction to sociolinguistics], 1–13. Kyoto:
27 Mineruva Shobō.
28 Hirai, Mitsuko. 2003. Waseieigo no keitairon [The morphology of made-in-Japan English]. Journal of
29 Rehabilitation and Health Sciences 1. 54–59.
30
Hirata, Yukari. 1990. Perception of geminated stops in Japanese word and sentence levels. Bulletin
of the Phonetic Society of Japan 194. 23–28.
31
Hirozane, Yoshito. 1992. Perception by Japanese speakers of some English sounds as the Japanese
32
choked sound /Q /. Bulletin of the Phonetic Society of Japan 201. 15–19.
33 Ichikawa, Sanki. 1929. Foreign influences in the Japanese language. Tokyo: Japanese Council,
34 Institute of Pacific Relations.
35 Ichikawa, Sanki. 1930. The pronunciation of English loan-words in Japanese. In N. Bøgholm, Aage
Brusendorff and C.A. Bodelsen (eds.), A grammatical miscellany offered to Otto Jespersen on
36
his seventieth birthday, 179–190. Copenhagen: Levin and Munksgaard.
37
Inoue, Aya. 2007. Grammatical features of Yokohama pidgin Japanese: Common characteristics of
38 restricted pidgins. In Naomi Hanaoka McGloin and Junko Mori (eds.), Japanese/Korean Linguistics
39 15. 55–66. Stanford: CSLI Publications.
40 Irwin, Mark. 2009a. Prosodic size and rendaku immunity. Journal of East Asian Linguistics 18. 179–
41
196.
42
194 Mark Irwin
1 Irwin, Mark. 2009b. Mora splitting in loanword compounds. Yamagata Daigaku Jinbungakubu
2 Kenkyū Nenpō [Yamagata University Faculty of Literature and Social Sciences Annual Research
Report] 6. 71–85. Yamagata University.
3
Irwin, Mark. 2011a. Loanwords in Japanese. Amsterdam: John Benjamins.
4
Irwin, Mark. 2011b. Mora clipping of loanwords in Japanese. Journal of Japanese Linguistics 27. 71–81.
5 Ishiwata, Toshio. 1976. Waseieigo to kokusai tsūyōgo [Made-in-Japan English and current interna-
6 tional language]. In Bunkachō (ed.), Gairaigo [Loanwords], 65–73. Tokyo: Bunkachō Bunkabu
7 Kokugoka.
8
Itô, Junko. 1990. Prosodic minimality in Japanese. In Michael Ziolkowski, Manuela Noske and Karen
Deaton (eds.), Papers from the 26th Regional Meeting of the Chicago Linguistic Society Volume
9
2: The Parasession on the Syllable in Phonetics and Phonology, 213–239. Chicago: Chicago
10
Linguistic Society.
11 Itô, Junko and Armin Mester. 1992. Weak layering and word binarity. Report LRC-92-09, University of
12 California, Santa Cruz. Republished (2003) in Takeru Honma, Masao Okazaki, Toshiyuki Tabata
13 and Shin-ichi Tanaka (eds.), A new century of phonology and phonological theory: A Festschrift
for Professor Shosuke Haraguchi on the occasion of his sixtieth birthday, 26–65. Tokyo:
14
Kaitakusha.
15
Itô, Junko and Armin Mester. 1995. Japanese phonology. In John A. Goldsmith (ed.), The handbook of
16 phonological theory, 817–838. Cambridge, MA: Blackwell.
17 Itô, Junko and Armin Mester. 1999. The phonological lexicon. In Natsuko Tsujimura (ed.), The hand-
18 book of Japanese linguistics, 62–100. Oxford: Blackwell.
19
Itô, Junko and Armin Mester. 2015. Word formation and phonological processes. In Haruo Kubozono
(ed.), Handbook of Japanese phonetics and phonology, 363–395. Berlin: De Gruyter Mouton.
20
Itô, Junko, Yoshihisa Kitagawa and Armin Mester. 1996. Prosodic faithfulness and correspondence:
21
Evidence from a Japanese argot. Journal of East Asian Linguistics 5. 217–294.
22 Itō, Masamitsu. 2003. Kayōkyoku no naka no gairaigo gaikokugo [Loanwords and foreign words in
23 pop songs]. Nihongogaku 22(7). 40–48.
24 JKS = Jiyūkokuminsha [Free Citizens Company] (ed.). 1977. Gendai yōgo no kiso chishiki 1977 [Essen-
tial modern vocabulary 1997]. Tokyo: Jiyūkokuminsha.
25
JKS = Jiyūkokuminsha [Free Citizens Company] (ed.). 2003. Gendai yōgo no kiso chishiki 2003
26
[Essential modern vocabulary 2003]. Tokyo: Jiyūkokuminsha.
27 JKS = Jiyūkokuminsha [Free Citizens Company] (ed.). 2010. Gendai yōgo no kiso chishiki 2010
28 [Essential modern vocabulary 2010]. Tokyo: Jiyūkokuminsha.
29 JKS = Jiyūkokuminsha [Free Citizens Company] (ed.). 2013. Gendai yōgo no kiso chishiki 2013 [Essen-
30
tial modern vocabulary 2013]. Tokyo: Jiyūkokuminsha.
Kageyama, Taro and Michiaki Saito. this volume. Chapter 1: Vocabulary strata and word formation
31
processes. In Taro Kageyama and Hideki Kishimoto (eds.), Handbook of Japanese lexicon and
32
word formation. Berlin: De Gruyter Mouton.
33 Katayama, Motoko. 1995. Loanword accent and minimal reranking in Japanese. Phonology at Santa
34 Cruz 4. 12–28. University of California, Santa Cruz.
35 Kawahara, Shigeto. 2011a. Aspects of Japanese loanword devoicing. Journal of East Asian Linguistics
20. 169–194.
36
Kawahara, Shigeto. 2011b. Japanese loanword devoicing revisited: A rating study. Natural Language
37
& Linguistic Theory 29. 705–723.
38 Kawahara, Shigeto. 2012. Lyman’s Law is active in loanwords and nonce words: Evidence from
39 naturalness judgment studies. Lingua 122. 1193–1206.
40 Kim-Renaud, Young-Key. 1997. The Korean alphabet. Honolulu: University of Hawai‘i Press.
41
Kindaichi, Haruhiko. [1942] 1967. Gagyō bionron [On the velar nasal]. In Kindaichi Haruhiko (ed.),
Nihongo on’in no kenkyū [Studies in Japanese phonology], 168–197. Tokyo: Tōkyōdō.
42
1 Kitahara, Yasuo (ed.). 2008. KY-shiki nihongo [KY-style Japanese]. Tokyo: Taishukan Shoten.
2 Koo, John and Yayoi Homma. 1989. Consonant gemination in English loanwords of Japanese. Journal
of the Association of Teachers of Japanese 23. 125–134.
3
Kubozono, Haruo. 1995. Gokeisei to on’in kōzō [Word formation and phonological structure]. Tokyo:
4
Kurosio Publishers.
5 Kubozono, Haruo. 1999a. Nihongo no onsei [The sounds of Japanese]. Tokyo: Iwanami Shoten.
6 Kubozono, Haruo. 1999b. Mora and syllable. In Natsuko Tsujimura (ed.), The Handbook of Japanese
7 linguistics, 31–61. Oxford: Blackwell.
8
Kubozono, Haruo. 2001a. On the markedness of diphthongs. Kobe Papers in Linguistics 3. 60–73.
Kobe University.
9
Kubozono, Haruo. 2001b. Epenthetic vowels and accent in Japanese: Facts and paradoxes. In Jeroen
10
van de Weijer and Tetsuo Nishihara (eds.), Issues in Japanese phonology and morphology, 113–
11 142. Berlin: Mouton de Gruyter.
12 Kubozono, Haruo. 2002. Prosodic structure of loanwords in Japanese: Syllable, structure, accent and
13 morphology. Journal of the Phonetic Society of Japan 6(1). 79–97.
Kubozono, Haruo. 2015. Loanword phonology. In Haruo Kubozono (ed.), Handbook of Japanese
14
phonetics and phonology, 313–361. Berlin: De Gruyter Mouton.
15
Kuwamoto, Yūji. 1998a. Ryakugokeisei kara mita nihongo no chōjūonsetsu no kōzō ni tsuite [Trun-
16 cation and the structure of trimoraic syllables in Japanese]. Gengo Kagaku Ronshū [Papers in
17 Linguistic Science] 2. 25–35. Tohoku University.
18 Kuwamoto, Yūji. 1998b. Nihongo ni okeru fukugōgoryakugo no on’inkōzō [The phonological struc-
19
ture of Japanese clipped compounds). On’in Kenkyū 1. 161–168.
Labrune, Laurence. 2002. The prosodic structure of simple abbreviated loanwords in Japanese: A
20
constraint-based account. Journal of the Phonetic Society of Japan 6(1). 98–120.
21
Labrune, Laurence. 2012a. The phonology of Japanese. Oxford: Oxford University Press.
22 Labrune, Laurence. 2012b. Questioning the universality of the syllable: Evidence from Japanese.
23 Phonology 29. 113–152.
24 Lee, Iksop and S. Robert Ramsey. 2000. The Korean language. Albany: State University of New York
Press.
25
Liu, Yongquan. 2001. New development of Chinese lexicon. Journal of Asian Pacific Communication
26
11(2). 255–262.
27 Liu, Yongquan. 2002. On the problems of Chinese lettered words. Sino-Platonic Papers 116. 1–13.
28 Loveday, Leo. 1996. Language contact in Japan: A sociolinguistic history. Oxford: Oxford University
29 Press.
30
Lovins, Julie Beth. 1973. Loanwords and the phonological structure of Japanese. Chicago, IL: Univer-
sity of Chicago dissertation.
31
Maruyama, Takao, Tadao Kobayashi, Chiaki Yamazaki and Masanori Terauchi. 1992. Katakanago o
32
eigo ni suru jiten [Dictionary of katakana words and their English translations]. Tokyo: Taishukan
33 Shoten.
34 McCawley, James D. 1968. The phonological component of a grammar of Japanese. The Hague:
35 Mouton.
Mester, Armin. 1990. Patterns of truncation. Linguistic Inquiry 21. 478–485.
36
Nakagawa, Yoshio. 1966. Rendaku, rensei (kashō) no keifu [A genealogy of sequential voicing and
37
sequential non-voicing (working label)]. Kokugokokubun 35. 302–314. Kyoto University.
38 Nakayama, Eriko, Rika Kiryū and Masaya Yamaguchi. 2007. Shinbun ni miru kikangairaigo [Core
39 loanwords found in newspapers]. In Kokuritsu Kokugo Kenkyūjo [National Institute for the
40 Japanese Language] (ed.), Hōkoku 126: Kōkyōbaitai no gairaigo – gairaigo iikae teian o sasaeru
41
chōsa kenkyū [Report #126: Loanwords in the public media – basic research on ‘paraphrasing
loanwords’], 343–378. Tokyo: Kokuritsu Kokugo Kenkyūjo.
42
196 Mark Irwin
1 NINJAL = Kokuritsu Kokugo Kenkyūjo [National Language Research Institute] (ed.). 1964. Hōkoku 25:
2 Gendai zasshi 90–shu no yōgoyōji daisanbunsatsu III: bunseki [Report #25: Vocabulary and
Chinese characters in 90 magazines of today III: Analysis of the results]. Tokyo: Shūei Shuppan.
3
NINJAL = Kokuritsu Kokugo Kenkyūjo [National Institute for the Japanese Language] (ed.). 1987.
4
Hōkoku 89: Zasshi yōgo no hensen [Report #89: Vocabulary trends in magazines]. Tokyo: Shūei
5 Shuppan.
6 NINJAL = Kokuritsu Kokugo Kenkyūjo [National Institute for the Japanese Language] (ed.). 2006.
7 Gairaigo iikae tebiki [A guide to paraphrasing loanwords]. Tokyo: Gyōsei.
8
NKD = Shōgakukan (ed.). 2000–2002. Nihon kokugo daijiten [The great dictionary of the Japanese
language]. Tokyo: Shōgakukan.
9
Ōtsuki, Fumihiko. 1884. Gairaigogen kō [Notes on loanword etymologies]. Gakugei Shirin 14. 122–
10
139, 370–384, 572–590. Tokyo University.
11 Poser, William. 1990. Evidence for foot structure in Japanese. Language 66. 78–105.
12 Quackenbush, Hiroko. 1977. English loanwords in Japanese: Why are they difficult for English-
13 speaking students? The Journal of the Association of Teachers of Japanese 12. 149–173.
Quackenbush, Hiroko. 1989. Gairaigo no sokuonka ni tsuite [Gemination in Japanese loanwords]. In
14
Hiroshima Daigaku Kyōikugakubu [Hiroshima University Faculty of Education] (eds.), Ryūgakusei
15
nihongokyōiku ni kansuru rironteki jisenteki kenkyū [Theoretical and practical research on
16 Japanese language education for foreign students], 1–7. Hiroshima: Hiroshima Daigaku Kyōikuga-
17 kubu Ryūgakuseinihongokyōiku / Nihongokyōiku Gakka. Hiroshima University.
18 Rosen, Eric Robert. 2003. Systematic irregularity in Japanese rendaku: How the grammar mediates
19
patterned lexical exceptions. Canadian Journal of Linguistics / Revue Canadienne de Linguistique
48. 1–37.
20
Seeley, Christopher. 2000. A history of writing in Japan. Honolulu: University of Hawai‘i Press.
21
Shibasaki, Hideko, Katsuo Tamaoka and Yūki Takatori. 2007. Amerikajin wa waseieigo o donogurai
22 rikai dekiru ka [How much can Americans understand made-in-Japan English?]. Nihongo
23 Kagaku 21. 89–110.
24 Shibatani, Masayoshi. 1990. The languages of Japan. Cambridge: Cambridge University Press.
Shinmura, Izuru (ed.). 2008. Kōjien dai-6 han. Tokyo: Iwanami Shoten.
25
SSDH = Sanseidō Henshūjo [Sanseidō Editorial Institute] (ed.). 2010. Konsaisu katakanago jiten dai
26
4 ban [Concise katakana word dictionary: 4th edition]. Tokyo: Sanseidō.
27 Sugimoto, Tsutomu. 2007. Gairaigo [Loanwords]. In Yoshifumi Hida, Yoshihide Endō, Masanobu Katō,
28 Takeyoshi Satō, Kiyoto Hachiya and Tomiyoshi Maeda (eds.), Nihongogaku kenkyū jiten [The
29 research encyclopaedia of Japanese linguistics], 408–411. Tokyo: Meiji Shoin.
30
Suzuki, Shunji. 2008. Waseieigo no kenkyū: sono kōzō to shisō [A study of Japanized English: Its
structure and thought]. Kokusai Tankidaigaku Kiyō [Bulletin of Kokusai Junior College] 23. 1–
31
47. Kokusai Junior College.
32
Takagi, Naoyuki and Virginia Mann. 1994. A perceptual basis for the systematic phonological corre-
33 spondences between Japanese load (sic.) words and their English source words. Journal of
34 Phonetics 22. 343–356.
35 Takayama, Tomoaki. 2005. A survey of rendaku in loanwords. In Jeroen van de Weijer, Kensuke
Nanjo and Tetsuo Nishihara (eds.), Voicing in Japanese, 177–190. Berlin: Mouton de Gruyter.
36
Tamaoka, Katsuo, Mutsuko Ihara, Tadao Murata and Hyunjung Lim. 2009. Effects of first-element
37
phonological length and etymological-type features on sequential voicing (rendaku) of second
38 elements. Journal of Japanese Linguistics 25. 17–38.
39 Tanabe, Yōji. 1990. Waseieigo no keitaibunrui [A morphological categorization of made-in-Japan
40 English]. Waseda Daigaku Nihongokenkyūkyōiku Sentā Kiyō [Bulletin of Waseda University
41
Center for Japanese Language] 2. 1–26. Waseda University.
42
1 Tateishi, Koichi. 1989. Theoretical implications of Japanese musicians’ language. In Jane Fee and
2 Kathryn Hunt (eds.), Proceedings of the eighth West Coast Conference on Formal Linguistics,
384–398. Stanford: CSLI Publications.
3
Tōdō, Akiyasu. 1969. Kango to nihongo [Sino-Japanese and Japanese]. Tokyo: Shūei Shuppan.
4
Tranter, Nicolas and Mika, Kizu. 2012. Modern Japanese. In Nicolas Tranter (ed.), The languages of
5 Japan and Korea, 268–312. Abingdon: Routledge.
6 Tsujimura, Natsuko. 2007. An introduction to Japanese linguistics. Oxford: Blackwell.
7 Tsujimura, Natsuko and Stuart Davis. 1987. The accent of long nominal compounding in Tokyo
8
Japanese. Studies in Language 11(1). 199–206.
Ueda, Kazutoshi, Junjirō Takakusu, Kurakichi Shiratori, Naojirō Murakami and Shōzaburō Kanazawa.
9
1915. Nihon gairaigo jiten [Dictionary of Japanese loanwords]. Tokyo: Sanseidō Shoten.
10
Umegaki, Minoru. 1944. Nihongairaigo no kenkyū zōhoban [A study of Japanese loanwords: Revised
11 and enlarged edition]. Osaka: Seinentsūshinsha.
12 Umegaki, Minoru. 1956. Ingo jiten [Dictionary of Cant, Argot, and Secret Language]. Tokyo: Tōkyōdō.
13 Umegaki, Minoru. 1963. Nihongairaigo no kenkyū [A study of Japanese loanwords]. Tokyo: Kenkyusha.
Vance, Timothy. 1987. An introduction to Japanese phonology. Albany: State University of New York
14
Press.
15
Vance, Timothy. 2008. The sounds of Japanese. Cambridge: Cambridge University Press.
16 Vance, Timothy. 2015. Rendaku [Sequential voicing]. In Haruo Kubozono (ed.), Handbook of Japanese
17 phonetics and phonology, 397–441. Berlin: Mouton de Gruyter.
18 Watanabe, Toshirō, Edmund Skrzypczak and Paul Snowden (eds.). 2003. Kenkyusha’s new Japanese-
19
English dictionary. 5th edition. Tokyo: Kenkyusha.
Yamada, Yūichirō. 2005. Gairaigo no shakaigaku [The sociology of Japanese loanwords]. Yokohama:
20
Shunpūsha.
21
Yamaguchi, Toshiko. 2007. Japanese linguistics: An introduction. London: Continuum.
22 Yonekawa, Akihiko. 1997. Wakamono kotoba jiten [Dictionary of youth speech]. Tokyo: Tōkyōdō.
23 Yuzawa, Tadayuki. 2010. Kodai nihonjin to gaikokugo [The ancient Japanese and foreign languages].
24 Tokyo: Bensei Shuppan.
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42

5 The Morphology of English Loanwords: Mark Irwin

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5 The Morphology of English Loanwords: Mark Irwin

Uploaded by

Copyright:

Available Formats

1 Mark Irwin

1 Before considering the morphology of English loanwords in Sections 2 and 3,

10 (1) a foreign word which has undergone adaptation to Japanese phonology,

15 1.2 Statistics and attestations

12 1.3 Phonological adaptation and phonemicization

1 In the discussion to follow, an important distinction must be drawn between

7 (2) a. toaster oven → oobuN +toosutaa ‘toaster oven’

34 2.1 Core morphology

1 etymon may be a noun, verb, gerund, adjective, preposition or even abbreviation or

4 (3) Noun: harassment → harasumeN to (suru) ‘harassment (to harass)’

31 c. taxi → takušii taku-r-u ‘to go by taxi’

33 (5) a. wild → wairudo wairudo-na ‘wild (out of control)’

1 Typically, any inﬂectional morphology an English etymon may possess has,

11 (6) a. pyjamas → pajama ‘pyjamas’

21 i. informed consent → iN foomudo+koN seN to ‘informed consent’

26 m. bun → baN zu ‘(hamburger) bun, bap’

28 Although relatively infrequent, reduplication can occur, usually accompanied by

34 (7) a. love → rabu+rabu → raburabu ‘lovey-dovey’

1 In their verbalization and adjectivization strategies (3–5), the loanword and

13 Table 1: Verbalization and adjectivization strategies across vocabulary strata

31 Table 2 oﬀers some examples of English loanwords, frequently prepositions or

1 Table 2: English loanwords as aﬃxes: A selection

10 the → za ‘the only, the za+terebijoN ← television TV website

17 over → oobaa ‘surfeit, oobaa+doraQ gu ← drug ‘overdose’

24 man → maN ‘person’ gag → gyagu+maN ‘comedian’

27 -er → raa ‘obsession’ ketchup → kečaQ pu+raa ‘ketchupaholic’

22 (8) time + service → taimu+saabisu ‘special oﬀer, special sale’

29 (10) gasoline + stand → gasoriN +sutaN do ‘petrol station, gas station’

5 (12) love + hotel → rabu+hoteru ‘hotel renting rooms by the hour’

9 Many of the morphosemantic compounds illustrated in (8)–(12) undergo a range of

14 Three truncatory processes occur with English loanword compounds: compound

21 Table 3: Truncation types

24 compound clipping (Section 3.1) dejitaru+kamera dejikame

34 If a compound undergoes truncation via one of the three processes illustrated, it

20 3.1 Compound clipping

39 18 Compounds consisting of a truncated reduplicated loanword, as in (7), are not included in El

1 Restricting discussion in line with El Cid content to compounds containing two

4 (13) the reduction of a loanword compound to between 2 and 6 moras (μ), by

25 (14) a. chocolate → čokoreeto → čoko

1 Table 4: Examples of compound clipping

12 2μ+2μ unique + clothing yuniiku+kuroojiN gu yunikuro Uniqlo 72.5%

38 Figure 2: Analysis of compound clipping output

1 in the accentuation patterns found in compounds (Tsujimura and Davis 1987), in

26 (15) a. geemu ← game

25 Figure 3: Final element syllable types (hypothetical and actual)

27 Several further issues related to compound clipping require further discussion:

1 yield an unmarked 2μ+2μ output: e.g. veriﬁcation card → beriﬁkeešoN +kaado →

9 (17) a. (inter)net radio → neQ to+rajio → netoraji (*neQ raji)

16 c. Netscape → neQ to+sukeepu → nesuke (*neQ suke)

42 25 The sole exception is model + casual → moderu+kajuaru→derukaji ‘casualwear as worn by

8 (19) memory stick → memorii+sutiQ ku → memosute ‘USB stick,

14 diﬀerential gear → defarenšaru+gia → defugia ‘diﬀerential gear’

17 Mora-splitting is infrequent, occurring only 13 times in El Cid and accounting for

1 (20) inter-college → iN taa+kareQ ji → iN kare

24 3.2 Morpho-orthographic truncation

9 OMISSION OF STAGE ➊ Morpho-orthographic truncation may also be applied to

ビジネスガ ール ビジネス + ガール business.girl b.g BG bii+jii businesswoman

omission of stage ➊ standard process

(Unicode 9 16/12/15 17:59) WDG-New (170mm240mm) DGMetaSerifScience (OpenType) 0000

pp. 161–198 1668 Kageyama_05_Irwin (p. 188)

noninitials and rebuses

NPO enu+pii+oo non-proﬁt organization

(Unicode 9 16/12/15 17:59) WDG-New (170mm240mm) DGMetaSerifScience (OpenType) 0000

ビジネスガールビジネス + ガール business.girl b.g BG bii+jii businesswoman