LINGUIST List 10.1477

Thu Oct 7 1999

Sum: Linking Elements in Compounds

Editor for this issue: Karen Milligan <>


  1. Andrea Krott, Linking elements in compounds

Message 1: Linking elements in compounds

Date: Thu, 7 Oct 1999 13:38:58 +0200 (MET DST)
From: Andrea Krott <>
Subject: Linking elements in compounds

Dear all,

some time ago I asked for information about languages that use linkers
in compounds. I am very thankful to everybody who responded. Below you
can find a list of all the languages I was pointed to. For each
language I added the name(s) of the person(s) who told me about it, a
short description, and (if possible) references.

People who responded to my query:

Laurie Bauer <>
Kristine Bentzen <>
Antonietta Bisetto <>
Eva Breindl <>
Bozena Cetnarowska <>
Henno Brandsma <>
Andrew Carstairs-McCarthy <>
Doug Cooper <>
Jan Engh <>
Jaco Geldenhuys <>
Jila Ghomeshi <ghomeshicc.UManitoba.CA>
Chris Golston <>
Pius ten Hacken <>
Esther Herrera <>
Lars Johanson <>
Satoshi Stanley Koike <>
Richard Laurent <>
Andrew McIntyre <>
Ingmarie Mellenius <>
Viktor I. Pekar <>
Asya Pereltsvaig <>
Linda Rashidi <>
Norvin W Richards <norvinMIT.EDU>
Jason Roberts <>
Danko Sipka <>
Erica Smale <>
Aaron Smith <>
Joan Smith/Kocamahhul <>
R'emy Viredaz <>
Cecil Ward <>

List of languages:
- ----------------

1.1 Germanic

- Afrikaans

 Laurie Bauer <>
 Andrew Carstairs-McCarthy <>
 Jaco Geldenhuys <>
 Erica Smale <>

 Afrikaans nominal compounds have the linkers -s- and -e-.
 They are very similar to the Dutch linkers.

 R.P. Botha (1968): The Function of the Lexicon in Transformational
 Generative Grammar. The Hague: Mouton.

- West Frisian
 Henno Brandsma <>

 West Frisian has nominal compounds with -s-, -e- and DIM (diminutive
 suffix) and no linker (plus other not productive linkers like -en).
 The -e- is phonological not identical to plural suffix -en as it is
 in Dutch. The choice of the linker is sometimes given by rules (-s-
 after suffix -ert, after nominal infinitives, after suffix -ing
 <+abstract>, and suffix -er <+levend>). Often there are only
 tendencies (see suffixes -heid, -dom etc.). The linker -en- is only
 possible if the plural suffix of the first constituent is -en- or if
 there is an empty plural ending.

 Jarich Hoekstra (1998): Fryske Wurdfoarming. Fryske Akademy, Ljouwert.

- German 
 Eva Breindl <>
 Andrew McIntyre <>

 German has noun-noun compounds with the linking morphemes -s-, -n-,
 -er-, and -e- which are originally genitive singular or plural morphemes.
 They are synchronically not neccessarily interpreted as such. Some
 of them are distributed according to grammatical rules.

 Fleischer, Wolfgang/Barz, Irmhild (1995): Wortbildung, 2.
 Auflage, T�bingen: Niemyer, p. 136-143 with references to special
 literature on the subject (Zepic 1970, Augst 1975, Wurzel 1970,
 Wellmann/Reindl/Fahrmaier 1974, Grube 1976).

 Demske, U., 1995. Word vs. Phrase Structure: The Rise of Genitive
 Compounds in German. FAS Papers in Linguistics vol 3. 1-28. (This is
 worth reading)

- Swedish
 Laurie Bauer <>
 Lars Johanson <>
 Ingmarie Mellenius <> (who kindly sent
 me her thesis about "The Acquisition of Nominal Compounding in Swedish")
 Viktor I. Pekar <>

 Swedish nominal compounds appear with the linking morphemes -o-, -u-, -e-, -s-.
 Sometimes there is variation: 'fotbollklubb' and 'fotbollsklubb'

 Josefsson, Gunl�g 1997. On the principles of word formation in 
 Swedish. Lund: Lund University Press. 

 Teleman, Ulf 1970. Om Svenska Ord. Lund: Gleerups.

 Mellenius, Ingmarie (1997): The Acquisition of Nominal Compounding
 in Swedish. Lund: Lund University Press. 

- Norwegian
 Kristine Bentzen <>
 Jan Engh <>
 Lars Johanson <>

 Norwegian nominal compounds appear with the linkers -s- and -e-.

 Leira, Vigleik: 1992, Ordlaging og ordelement i norsk. Oslo : Samlaget.
 ISBN 82-521-3844-6

 Gundersen, Dag, Jan Engh, Ruth Vatvedt Fjeld: 1995, H�ndbok i norsk :
 skriveregler, grammatikk og spr�klige r�d fra a til �. Oslo :
 Kunnskapsforlaget. ISBN 82-573-0562-6

 Ak�, J�rn-Otto: 1989, Sammensatte ord : bruken av s-fuge i moderne bokm�l.
 Unpublished master's thesis, Institutt for Nordisk spr�k og litteratur,
 Universitetet i Oslo. [The author can be contacted:

- Icelandic
 Laurie Bauer <>

 In Icelandic compounds, 
 a) the stem (or root) may be joined directly to the following
 b) it may be connected by a so-called connective vowel, that
 otherwise does not appear,
 c) or some case forms of the word may be used (never the nominative,
 if it is different from the stem), especially the genitive.

 Stefan Einarsson (1945): Icelandic. Grammar Texts Glossary. John Hopkins
 University Press: Baltimore and London.

- English 
 Pius ten Hacken <>
 Andrew McIntyre <>

 Andrew McIntyre pointed me to the frozen linking morphemes in
 'spokesman', 'sportsman', 'marksman', Pius ten Hacken to the Saxon
 genitive construction.

 ten Hacken, Pius (1994), Defining Morphology: A Principled Approach to
 Determining the Boundaries of Compounding, Derivation, and Inflection,
 Olms, Hildesheim.

 ten Hacken, Pius (1999), 'Motivated Tests for Compounding', Acta
 Linguistica Hafniensia 31:27-58.

1.2 others
- Celtic: Scottish Gaelic
 Cecil Ward <>

 In Celtic languages, there is a linking effect in certain nominal
 compounds where a defining element precedes the head noun, rather
 than following it as is usual (Celtic being head-first). The
 preposed element is typically an adjective, although in general
 adjectives follow the noun, bar a tiny group (compare French).

 1. The following rule seems to be completely regular "lenite the
 noun when an element is preposed" (postposed modifiers being
 the norm).

 2. The meaning seems to be "a specific kind of N" (for example,
 "city") as opposed to "an N that happens to be A" (a town that
 is big).

 3. This seems to be fully productive, although given its semantic
 function, that of defining new "special kinds of x", only a
 limited number are current, although creating neologisms using
 this process is a strategy that is recognisable to hearers. 

 4. The presence of the initial consonant mutation "lenition" is in
 many cases equivalent to the presence of an abstract preceding
 null morpheme. Historically, the initial consonant mutations
 originated as sandhi effects from the presence of a lost ending
 on the preceding word, or that represented the residual effects
 of a word that had been completely elided away.

 For an overview of Celtic initial consonant mutations, see the
 article by King in "The Handbook of Morphology", Spencer and Zwicky,
 ISBN 0631185445; also MacAulay, ed., "The Celtic Languages", CUP.

- ancient Greek and modern Greek
 Antonietta Bisetto <>
 Chris Golston <>
 Richard Laurent <>

 The linker in Greek compounds is usually -o-.

 Rivista di Linguistica, volume 4, number 1, 1992 (edited by
 Sergio Scalise and published by Rosenberg and Sellier in Italy).
 There, you can find an article by Angela Ralli on Modern Greek

 Smith's Greek Grammar (1920: �870ff.)

- Finno-Ugric: Finnish 
 Laurie Bauer <>

 Finnish allows case forms of the modifying noun.

- Altaic: Turkish
 Joan Smith/Kocamahhul <>
 Pius ten Hacken <> 

 Turkish has a compound marker with the same shape as the possessive
 marker for the third person singular, an attached '-i'. 
 example: 'okul kitab-i' (textbook)
 Some of these compounds are frozen and have become a single word: 
 'ayak' (foot) + 'kap' (container) > 'ayakkabi' (shoe)

 Kornfilt, Jacklin (1997): Turkish. London et al.: Routledge.

- North Caucasian: Kabardian
 R'emy Viredaz <>

 Kabardian has the connectives -ah-, -m-, -n-, and -r- which appear
 between two segments. They are not stressed and they are always in
 non-syllabic juncture with a following segment. Their use is
 sometimes facultative and varies dialectally.

 Kuipers, Aert H. (1960): Phoneme and Morpheme in Kabardian, The Hague, 78-80.

 Rieks Smeets (1984): Studies in West Circassian Phonology and Morphology, Leiden.

- Indo-Iranian:
 a) Sanskrit
 Laurie Bauer <>

 Sanskrit allows case forms of the modifying noun.

 b) Persian 
 Jila Ghomeshi <ghomeshicc.UManitoba.CA> 
 Linda Rashidi <>
 Norvin W Richards <norvinMIT.EDU>

 These people informed me about the Persian Ezafe construction:
 The esafe is a single vowel which links nouns to their modifiers
 (which can be adjectives or other nouns, but not phrasal) and to
 possessors. It is phonologically attached to the head or the
 preceding element but is semantically part of the post-modifier.
 It does not appear in compounds and it is not a morpheme, but it
 is a linker.

 ten Hacken, Pius (1994), Defining Morphology: A Principled Approach to
 Determining the Boundaries of Compounding, Derivation, and Inflection,
 Olms, Hildesheim.

 ten Hacken, Pius (1999), 'Motivated Tests for Compounding', Acta
 Linguistica Hafniensia 31:27-58.

 Ghomeshi, Jila (1997): Non-projecting nouns and the Ezafe
 Construction in Persian. NLLT (Natural Language & Linguistic
 Theory), Vol. 15, pp. 729-788.

2. Slavic:

- Russian
 Richard Laurent <>
 Viktor I. Pekar <>
 Asya Pereltsvaig <>
 Russian compounds have the linkers -o- and -e-. They are determined
 on the basis of the preceding consonant (-e- after a 'soft'
 consonant). They also appear in adjectival compounds and they are
 productively used. They only and always appear between roots.
 Compounds consisting of full words do not have linking vowels.
 Neither -o- nor -e- has any meaning.

 short grammar of Russian by Unbegaun. (As the author never 
 bothers to transliterate, a reader would do well to know Cyrillic.)

- Polish
 Bozena Cetnarowska <>
 Viktor I. Pekar <>

 First, the most common linker in Polish is -o-. It links noun-noun
 compounds and adjective-adjective compounds. It is homophone to a
 neuter marker, but it appears after nouns which do not have
 an -o- ending. Second, there are also the linkers -i- or -y-. They
 appear in verb-noun compounds and are usually (not always) the
 thematic vowel of the verb in question.

- Serbo-Croatian
 Danko Sipka <>

 Serbo-Croation compound parts are connected by an -o-.

3. Austronesian
- Yapese
 Jason Roberts <>
 Some of the productive pattern of compounding in Yapese are:
 - intransitive verb + long /e:/ or /e":/ + noun
 - transitive verb + long /o:/ + noun
 - noun with possessive suffix -n 'his, its' + long /e:/, /e":/, /i:/
 or /a:/ + noun
 All three types have an alternative variant without linking element.

 Jensen (1977) Yapese Reference Grammar, p102.

- Kosraean/Kusaien 
 Jason Roberts <>

 In Kusaien there is a linker -in- in noun-noun compounds and in
 noun-intransitive verb compounds.
 Lee (1975) Kusaiean Reference Grammar, p.213-214.

- Tagalog
 Norvin W Richards <norvinMIT.EDU>

 In Tagalog there are two main linkers: the linker 'na'/-ng which
 appears between modifiers (adjectives and relative clauses) and the
 nouns they modify, among other places, and the linker -ng which
 appears in compounds. Both are phonologically based: -ng (both
 types) appears after a first constituent ending in /h/, /'/, or /n/,
 'na' or the zero form of the compound linker appears after first
 constituents ending in other consonants.

 Schachter and Otanes' _Tagalog Reference Grammar_

 Edward Rubin (1994): Modification: a syntactic analysis and its consequences.

 Norvin Richards (1999): Complementizer cliticization in Tagalog and

4. Asia
- Daic: Thai
 Doug Cooper <>

 Thai has linking morphemes that pop up in the middle of many
 compounds of Pali/Sanskrit origin. This occurs because Thai tends to
 discourage certain kinds of finals, eg. short vowels.
 As a standalone word, the loan may have its final vowel (or
 consonant + vowel) either removed from the orthography, or
 'silenced' with a special character, or simply ignored. Then, when
 the word appears in a compound, the final is either added,
 'unsilenced,' or read.

 There are some cases in which the final vowel/consonant+vowel is
 regularly suppressd even in compounds, e.g., when the supressed
 final vowel leads into a leading vowel.

 Richard Noss _Thai Reference Grammar_

 Gedney _Indic Loanwords in Spoken Thai_ 

 a whole lot of Thai references at

- Austro-Asiatic: Cambodian
 Laurie Bauer <>

 Cambodian (Khmer) appear to have linking elements very much like the
 Germanic ones.

- Japanese
 Satoshi Stanley Koike <>

 In Japanese there are three linking phenomena in compounds: 1)
 sequential voicing of the initial consonant of the second
 constituent, 2) an epenthetic consonant between the two
 constituents, and 3) a vowel change of the final consonant of the
 first constituent. Koike's explanation for these phenomena is a lost
 genitive postposition _no_ which occured between the two constituents.

 Koike, Satoshi Stanley (1996): Sequential voicing in Japanese and
 adjacency. Proceedings of ConSOLE IV, 143--50.

 Koike, Satoshi Stanley (199?): A monosemy approach to the Japanese
 particle _no_: functional categories as linkers and antisymmetry
 in natural language. PhD thesis.
5. Southamerica
- Mixe-Zoque: Zoque
 Esther Herrera <>

 In Zoque there is a nominal compound formation where a vowel appears
 between the two nouns. The vowel is a result of a vowel spreading,
 which means that the vowel of the left constituent is repeated. 

 Herrera, Z. Esther (1995), "Palabras Estratos y
 Representaciones: Temas de Fonologia Lexica en Zoque, El Colegio de

6. Niger-Kongo (Kwa): 
- Yoruba
 Laurie Bauer <>

 Yoruba appears to have linking elements very much like the Germanic ones.


See also:
Bauer, Laurie 1978 (?) On teaching compound nouns. Moderna Spr�k

Dressler, W & Barbaresi, L. 1986: How to fix the interfixes. In: Acta
Linguistica Academiae Scientiarum Hungaricae 36: 53-67.

Rivista di Linguistica, volume 4, number 1, 1992 (edited by Sergio
Scalise and published by Rosenberg and Sellier in Italy)


Andrea Krott M.A.

Interfaculty Research Unit for Language and Speech &
Max Planck Institute for Psycholinguistics
Wundtlaan 1
PB 310, 6500 AH Nijmegen
The Netherlands

Tel: +31 - (0)24 - 3612160
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue