LINGUIST List 14.3260

Wed Nov 26 2003

Sum: Loanword Adaptation

Editor for this issue: Steve Moran <>


  1. Jennifer Smith, Loanword Adaptation

Message 1: Loanword Adaptation

Date: Mon, 24 Nov 2003 16:47:33 -0500 (EST)
From: Jennifer Smith <>
Subject: Loanword Adaptation

In (LINGUIST 14.3045), I posted a query about patterns of phonological
loanword adaptation involving syllable-final clusters.

I was looking for examples of: (1) languages that use two different
strategies in adapting CVCC syllables depending on the nature of the
consonants in the cluster, and (2) languages that adapt CVCC by
epenthesizing into the cluster (--> CV.CvC) when both consonants in
the cluster are obstruents.

Many thanks to the following people, who sent information about
languages, references, or both: Ron Artstein, Elena Bashir, Mike
Cahill, Sue Hassel, Shinji Ido, Toby Paff, Marc Picard, Yvan Rose,
Gary Toops, Pete Unseth, Adam Ussishkin, Andrew Wedel, and Moira Yip.
The following is a brief summary of the responses that I received;
please contact me directly if you would like more information or

- -------- 

Type (1): Languages with varying repair strategies

Marc Picard reports that Canadian French borrows fricative-final
clusters unchanged, but resolves ''most other clusters'' with

Andrew Wedel and Shinji Ido sent information on Turkish. Andrew
reports that Turkish tolerates some final clusters both in non-loan
cases and in loanwords, but resolves illicit clusters with epenthesis
into the cluster (CV.CvC). He also mentions some interesting facts
about historical changes in patterns of epenthesis in loanword onset
clusters. Shinji provides an example of an Arabic CVCC word that is
borrowed into Turkish as CV.CvC but is borrowed into Uzbek and Tajik

Mike Cahill describes Konni (Gur; northern Ghana), which generally
only allows nasal codas in the non-loan phonology, and accordingly
borrows CVNC as CVN.Cv but other CVCC as CV.Cv.Cv.

Adam Ussishkin (reporting on research in collaboration with Dafna
Graf) and Ron Artstein sent information about Modern Hebrew. The
treatment of clusters here depends on sonority class: two-obstruent
clusters and sonorant-obstruent clusters are tolerated; two-sonorant
clusters are usually split by epenthesis, although there are
individual lexical items that are exceptions; and, Ron notes,
obstruent-sonorant clusters are also split by epenthesis. Adam points
out that coda clusters (regardless of sonority) are also avoided in
the non-loan phonology of Modern Hebrew, except when they are formed
by the addition of the 2sg feminine past-tense suffix [-t].

- --------

Type (2): Epenthesis into an obstruent-obstruent cluster

Elena Bashir reports that final clusters in Persian loans into Urdu
and Panjabi are systematically adapted as CV.CvC for all consonant
cluster types (including two-obstruent clusters). Also, there is some
variability in the presence of the epenthetic vowel depending on the
speaker's educational background, especially in Urdu.

- --------

Finally, several respondents sent information about other languages
with interesting patterns of loanword-cluster adaptation.

Gary Toops sent an example of cluster avoidance only under
suffixation, in loanwords in Serbian/Croatian/Bosnian: an English loan
cluster is tolerated word-finally in _student_, but split by
epenthesis in gen. pl. _studenata_ ''(both _a_ are long)''.

Moira Yip sent a paper that discusses a number of patterns involved in
loanword adaptation, including a systematic difference in adaptation
strategies (V epenthesis vs. C deletion) in two varieties of Mandarin.
The paper was published in the Journal of the Phonetic Society of
Japan (vol 6, no. 1, 2002).

Toby Paff suggested looking at English loanwords into Spanish, noting
that Spanish essentially avoids coda clusters, and there are a lot of
Spanish-speaking communities in the US that could well be using many
loanwords from English.

Pete Unseth reports that Amharic tolerates certain CC clusters, but
resorts to epenthesis when there would otherwise be a large cluster.

Yvan Rose has very kindly offered to send me detailed information
about French loanwords in Kinyarwanda, which were the subject of his
MA thesis from Université Laval.

Sue Hassel reports that Tswana (Setswana) resolves most loanword
clusters with epenthesis, and that a word-initial CV syllable
resulting from this process is sometimes reinterpreted as a noun-
class prefix.

Thanks again to all who responded.
- Jen

 Jennifer Smith Department of Linguistics 322 Dey Hall, CB #3155 University of North Carolina
 (919) 962-1474 Chapel Hill, NC 27599 USA 
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue