Bound morphology in English (and Beyond)

24
CHAPTER FIVE BOUND MORPHOLOGY IN ENGLISH (AND BEYOND): COPY OR COGNATE? Anthony Grant This paper discusses the sources of bound inflectional (and also some bound derivational) morphology in Present-Day Standard English and thirteen other languages. Evidence suggests that the correlation between borrowing basic vocabulary and the incorporation of items of borrowed morphology into the borrowing language’s structure (where it can be used with inherited as well as with other stems) is weak, with some lan- guages which have borrowed a lot of basic vocabulary acquiring little if any inflectional morphology from their model languages, although deri- vational morphology is more likely to be transferred than inflectional morphology. Transfer of patterns from one language to another using morphemes which are already part of the basic language’s inventory is a fairly common practice. 1 Introduction This paper discusses the sources of bound inflectional (and also some bound derivational) morphology in Present-Day Standard English and a number of other languages.1 I am attempting to test the extent to which inflectional and productive derivational morphological strata in a lan- guage are more immune to borrowing than the lexicon is. To this end I compare patterns and levels of copying of bound morphological items in English with those patterns of morphemic borrowing found in a sample of 13 other ‘heavy copying’ languages from around the world. These lan- guages were selected because they have each replaced a large proportion of their pre-existing basic vocabulary with copies from other languages. The quest is to see whether heavy lexical copying of this kind correlates 1 Thanks to Peter Bakker, Éva Á. Csató, Lars Johanson, Babs Richardson, Martine Rob- beets and Nikki van de Pol for comments. 99-122_JOHANSON-ROBBEETS_F6.indd 99 99-122_JOHANSON-ROBBEETS_F6.indd 99 3/30/2012 9:21:31 AM 3/30/2012 9:21:31 AM

Transcript of Bound morphology in English (and Beyond)

CHAPTER FIVE

BOUND MORPHOLOGY IN ENGLISH (AND BEYOND): COPY OR COGNATE?

Anthony Grant

This paper discusses the sources of bound inflectional (and also some bound derivational) morphology in Present-Day Standard English and thirteen other languages. Evidence suggests that the correlation between borrowing basic vocabulary and the incorporation of items of borrowed morphology into the borrowing language’s structure (where it can be used with inherited as well as with other stems) is weak, with some lan-guages which have borrowed a lot of basic vocabulary acquiring little if any inflectional morphology from their model languages, although deri-vational morphology is more likely to be transferred than inflectional morphology. Transfer of patterns from one language to another using morphemes which are already part of the basic language’s inventory is a fairly common practice.

1 Introduction

This paper discusses the sources of bound inflectional (and also some bound derivational) morphology in Present-Day Standard English and a number of other languages.1 I am attempting to test the extent to which inflectional and productive derivational morphological strata in a lan-guage are more immune to borrowing than the lexicon is. To this end I compare patterns and levels of copying of bound morphological items in English with those patterns of morphemic borrowing found in a sample of 13 other ‘heavy copying’ languages from around the world. These lan-guages were selected because they have each replaced a large proportion of their pre-existing basic vocabulary with copies from other languages. The quest is to see whether heavy lexical copying of this kind correlates

1 Thanks to Peter Bakker, Éva Á. Csató, Lars Johanson, Babs Richardson, Martine Rob-beets and Nikki van de Pol for comments.

99-122_JOHANSON-ROBBEETS_F6.indd 9999-122_JOHANSON-ROBBEETS_F6.indd 99 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

100 anthony grant

with heavy morphological copying, and if so, then to see to what extent these coexist. Features from one speech tradition which fijind their way into another can be classifijied as being transfers of fabric or transfers of pattern (Grant 1999, 2002). The two categories are not mutually exclusive; a feature may belong to both.

These languages all fall to a greater or lesser degree into the category of “fusion languages” proposed in Weinreich (1973, 1: 32–38), in which elements of various origins can combine freely with one another despite their multifarious origins. This is a scalar rather than an absolute category since all languages are more or less fusional, as all those for which we have sufffijicient data to ascertain this have at least absorbed some loanwords. I explore here the consequences of relative degrees of what we may call the gradience of fusion. At word level, some languages blend elements of diffferent origins into a single word more fully than others.

Before I explore this matter further, a caveat is in order. Not all lan-guages which show a high degree of borrowing of basic lexicon are ‘fusion languages’2 in the same way. There are stable mixed languages such as Michif which have absorbed complete sets of morphological forms for use with items from a particular form-class from one of their component languages (Bakker and Mous 1994). There is also the typologically unique case of Berbice Dutch which has taken all its bound morphology (and a high proportion of its basic and less basic vocabulary) from Eastern Ijo while also using many free morphemes from Dutch, Lokono/Arawak and latterly Guyanese Creole English (Kouwenberg 1994). Other lan-guages have absorbed many exogenous patterns rather than exogenous morphemes. The Oceanic language Takia (Ross 1996, 2009) remodelled much of its morphosyntactic structure on nearby Papuan languages such as Waskia but borrowed very little lexically and nothing morphologically from them.3

There are also creoles such as Mauritian Creole French which have developed their own sets of morphological forms through the subsequent grammaticalisation of items which were previously only lexical forms (Grant and Baker 2007). But the fourteen fusion languages which are sur-

2 Fusion languages are not be confused with fusional morphology, in which several structural features can be combined into a single unanlysable morph, as happens in Latin conjugations and inflections.

3 Lexical borrowing seems to have prevailed in the opposite direction, from Oceanic languages to Waskia, but Takia has borrowed lexicon from other languages such as Bar-gam, Gedaged, German and especially Tok Pisin.

99-122_JOHANSON-ROBBEETS_F6.indd 10099-122_JOHANSON-ROBBEETS_F6.indd 100 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

bound morphology in english (and beyond) 101

veyed here are a diffferent case, and two issues are important here. Firstly they have accreted their borrowed bound morphemes onto the compo-nents of a pre-existing (and continuously maintained) morphological system. Secondly these borrowed morphemes are used with lexical items of all origins, not merely those which share an origin with the borrowed morpheme.

2 The Languages in the Sample

All the languages in the sample take at least 10% of the items on their 207-item Swadesh lists of purportedly hard to borrow vocabulary (Swadesh 1955, Comrie and Smith 1977, 72–77) from other languages; they are basic languages because they have received transferred material. Major model languages are italicised and major data sources identifijied; the total list of sources used for each language is much greater.

Table 1: The language sample

1. (British) English (Indo-European: Germanic; worldwide); French, Latin, Norse, Low German, Greek (OED online; West 1948; abbreviation: En).

2. Kalderash Romani (Indo-European: Indic; E Europe and beyond); Iranian, Armenian, Greek, South Slavic, Hungarian, Romanian, German (Gjerdman and Ljungberg 1963, Boretzky 1994; Kal).

3. Hindi-Urdu (Indo-European: Indic; India and Pakistan); Arabic, Persian, Sanskrit (ILCAA 1968, McGregor 1993, Saxena 2010; Hi).

4. Standard Swahili, based on KiUnguja (Bantu; Tanzania and Kenya); Northern Swahili; Arabic (Perrott revised Russell 2002; Sw).

5. Acehnese (Austronesian: Malayo-Chamic; N. Sumatra); Malay, no longer extant Mon-Khmer languages (Daud and Durie 1998; Ac).

6. Tagalog (Austronesian; Philippines); Kapampangan, Malay, Spanish, English, (Rubino 1998, McFarland 1989 for the 2000-word lexical sample; Tg).

7. Chamorro (Austronesian; Guam and other Mariana Islands, Micronesia); Spanish and unidentifijied Austronesian substrate (Topping, Ogo and Dungca 1975; Blust 2000; Ch).

8. Yapese (Austronesian; Yap, W. Micronesia; Palauan and pre-Palauan, Chuukic (Jensen, Pugram and Defeg 1977, Ross 1996; Yp).

9. Ifijira-Mele (Austronesian: Oceanic; Vanuatu); South Efate (Clark 1998; IM).10. Ngandi (Ngandi-Nunggubuyu; Arnhem Land, Australia); Ritharngu,

Warndarrang (Heath 1978a; Ng).11. Pipil/Nawat (Uto-Aztecan; El Salvador, moribund); Spanish (Campbell 1985;

Pi).12. Garifuna/Island Carib/Black Carib (Arawak; Belize and Honduras); Kari’na;

Antillean Creole French, Spanish, English (Cayetano 1993; Ga).

99-122_JOHANSON-ROBBEETS_F6.indd 10199-122_JOHANSON-ROBBEETS_F6.indd 101 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

102 anthony grant

I have also used some parallel comparative lexical data for certain lan-guages from the Highland Chamic language Jarai (paralleling the data from Acehnese), the Romani variety Sinto Piemontese (paralleling Kal-derash data) and the Aztecan language Mexicanero Nahuatl of Durango (paralleling Pipil data). The sources which I have used for Jarai are Lafont (1968), Headley (1965), Siu (2009), and Thurgood (1999), while for Sinto Piemontese I have used Formoso and Calvet (1987), and Fochi and Soravia (1995), and my data for Mexicanero come from Canger (2001) and Castro Medina (2000). My structural and textual data for Sinto Piemontese, Jarai and Mexicanero are considerably sparser than those which I have for the other languages in the sample, and they are therefore included here to demonstrate the high degree of borrowability within their various clades (Romani, Aceh-Chamic and Nahua respectively)4

Overall the proportion of Swadesh list items which have been bor-rowed into at least one of these languages as the usual means of express-ing the concept in question is extremely high. Indeed, the only items on the cumulative Swadesh list for which forms not borrowed into any of the above languages are the ones with the glosses ‘to eat, eye, here, louse, right-side, there, where, who, to wipe’, and labels for all these are been borrowed into some other language outside this sample.

Apart from Mexicanero, which as a geographically isolated Nahua lan-guage exhibits a much greater degree of borrowing from Spanish than Pipil does, the other languages added to the sample for their scores of bor-rowed lexicon (Sinto Piemontese and Jarai) show similar degrees of bor-rowing to those of Kalderash and Acehnese respectively. Note that much of Jarai’s lexicon is unsourced while the main loan source for Acehnese is Malay, which has not afffected Jarai.

4 Although Siwi and Taqbaylit/Kabyle Berber are the varieties purportedly most influ-enced by Arabic, I have also examined and counted data on the Ayt Seghrouchen variety of Tamazight.

Table 1 (cont.)

13. Cuzco Quechua (Quechuan; Peru); Jaqi, Spanish (Cusihuamán 1976; CQ).14. Berber languages of Egypt, Algeria and Morocco4 (Afroasiatic); Arabic,

Latin, Punic. (Laoust 1931 for Siwi, abbreviation: Si, Dallet 1985 for Taqbaylit/Kabyle, Penchoen 1973 for Ayt Ndhir Tamazight, Abdel-Massih 1971 for Ayt Seghrouchen Zenatiya).

99-122_JOHANSON-ROBBEETS_F6.indd 10299-122_JOHANSON-ROBBEETS_F6.indd 102 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

bound morphology in english (and beyond) 103

Table 2: Loan percentages in the lexical samples of languages discussed in Grant (2006), with data from some added languages (these are prefijixed by *)

100-item list loan percentage

215-item list loan percentage

Loan % in a larger etymo-logised lexical

sample, n is the language sample size

English 11 16 72+; n = 80,000; 52% loans on a sample of 3984 words

Kalderash Romani 14 23 52+; n = c. 3500*Piedmontese  Sinto (Romani)

16 20.5 50+; n = c. 1600

Hindi-Urdu 16 20.5 35.5+; n = 1972Acehnese 10 15 50+; n = c. 3500*Jarai (Chamic,  Vietnam)

12 16 25+; n = c. 2200

Tagalog 9.5 13 20+; n = 2000Chamorro 29 33 62+; n = 6200Yapese 19 19.5 20+; n = 2900Ifijira-Mele 21 26 34+ (discrete

morphemes nearly 50 percent); n = 3228

Ngandi 32 34 35+; n = 1150Pipil 5 (+3 Spanish/

Pipil doublets)10 At least 24; n = 2853

*Mexicanero  (Durango Nahuatl)

27 28 30+; n = c. 1200

Garifuna 19 21 25+; n = c. 4500Cuz. Quechua 22 27 33+; n = c. 3000Berb. (Kabyle) 18 28 30+ (estimate);

n = c. 6000Berber (Ayt  Seghrouchen)

15 16.5 52; n = 3085

Berber (Siwi) 24 33 56; n = 1496Standard Sw. 20 21 27.8 +; n = 1450

By comparison with these fijigures, percentages of loan items on longer Swadesh lists (using the 200 item lists) for other Germanic languages (Embleton 1986: 100–101) are as follows: German 2%, Low Saxon 2, Dutch 1.5, Afrikaans 0.5, Flemish 2.5, Icelandic 0.5, Norwegian 6, Danish 7, Swed-ish 5.5, Yiddish 6, Pennsylvania German 2.5, West Frisian 9, Faroese 14, Tok Pisin 4.5, against English’s 16% borrowed lexicon on the longer Swadesh list.

99-122_JOHANSON-ROBBEETS_F6.indd 10399-122_JOHANSON-ROBBEETS_F6.indd 103 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

104 anthony grant

3 Pattern, Fabric and Diachronically Retroactive Transfer

As previously stated, those features from one speech tradition which fijind their way into another speech tradition can be classifijied as being transfers of fabric or transfers of pattern (Grant 2002). They can be examples of both kinds of transfer. Concentration here is on cases of transfer of fabric, which involves an actual phonological form plus meaning being incorpo-rated from one language to another, though pattern transfers abound in some of the languages (examples of pattern transfer are English phrasal verbs, a pattern which was borrowed from Norse, or English of-genitives, which were calqued from the French pattern).

Bound inflectional morphemes in these languages are rarely borrowed and they spread to use with pre-borrowed items even more rarely. Recon-struction of proto-forms is not equally advanced for all these languages, so that tracing the history of the bound morphemes is not equally possible for all the languages in the sample. It is therefore not always easy to distin-guish inherited morphological forms from innovated (but not borrowed) forms. However, such evidence as we have suggests that the bulk of the inflectional morphemes in all these languages (none of which are isolates) are inherited from some level of proto-language rather than being bor-rowed or internally innovated.

In general, most of their inflections are inherited from a proto-language. For example, English plural –s, comparative –er and superlative –est go back to Proto-Indo-European, past participial –en goes back at least to Proto-Germanic, as do most strong verbs which are not of Indo-European age and the few ablauting plurals left in modern English. Or else they have changed their senses over time: Present-Day English present participial –ing was formerly an abstract noun afffijix. On occasion they have evolved within the language: English 3sg present indicative –s is a Northern Middle English innovation replacing earlier English and Southern Middle English –eth, which itself traces back to Proto-Indo-European (Williams 1975).

Some of these languages have borrowed inflectional morphemes and have used them with some non-borrowed items (or with stems which are taken from a subsequent model language: Kalderash Romani in Boretzky 1994),5 and have freely combined morphemes of various origins in a single

5 To this end I have not further discussed most of the bound inflectional morphemes taken from Greek, Romanian or Slavic, which are used to integrate most loans of European origin (including most of those taken from Greek) into Romani morphology, the majority of which is of Indic origin. This system of ‘athematic morphology’, which involves bor-

99-122_JOHANSON-ROBBEETS_F6.indd 10499-122_JOHANSON-ROBBEETS_F6.indd 104 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

bound morphology in english (and beyond) 105

word. English uses a word such as washable (combining a French afffijix with an Old English stem). Yiddish, though it has a proportion of Swadesh list loans coming to less than 10%, has words such as šlimezalnitse (‘unlucky woman’: šlim < German schlimm ‘bad’, mazzāl < Hebrew ‘constellation’, -nitse < West Slavic -nica ‘feminine afffijix’) and a very few non-Hebrew nouns used with Hebrew afffijixes (poyer-im ‘farmer-s’ from the Middle High German analogue to German Bauer plus the Hebrew masculine plu-ral -im).

Williams (1975, 125–131) shows that more afffijixes in Present-Day Stan-dard English (specifijically items of derivational morphology) are borrowed than inherited, and that this is especially the case with productively-employed inflectional and derivational afffijixes which have survived into Modern English. Even so, Williams also listed Old English afffijixes which are not now used. The tables which Williams gives show that from a data-set of 79 derivational prefijixes in his collection 17 of those 38 items which are of Old English vintage are still in use, while the 21 others are extinct or nearly so (being restricted to use with one or two stems). Williams lists 91 derivational sufffijixes, of which 35 are at least of Old English vintage; 8 of these are no longer in use while another 5 are used with only one or two stems.

But the kind of diachronically retroactive transfer which these exam-ples instantiate is rather infrequent in the languages sampled. Normally, the borrowed afffijixes are not applied to elements which were present in the language before the afffijixes were borrowed or copied. As we shall see in section 5, Ngandi, Garifuna, Siwi Berber and KiUnguja (and at a pinch Acehnese6 and Kalderash) controvert this rule, as they have all absorbed some morphology from external sources and they use it with inherited morphemes as well (and also with loans from languages with which the speakers came into contact after encountering the languages from which they took the borrowed bound morphemes). But this is very much the exception.

rowed morphs used with nouns, verbs and adjectives, is discussed in Bakker (1997) for four Romani varieties, and is used to integrate even modern loanwords into Romani, but is confijined to certain (often very extensive) diachronic strata of the language. The ordinal morpheme –to- is an exception.

6 This is the case if the Acehnese causative prefijix peu-, taken from Mon-Khmer sources and used with verbs of any origin (for instance peubloë ‘to sell’ from bloë ‘to buy’, from Proto-Austronesian *beli), counts as an inflectional rather than a derivational prefijix.

99-122_JOHANSON-ROBBEETS_F6.indd 10599-122_JOHANSON-ROBBEETS_F6.indd 105 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

106 anthony grant

4 Relative Borrowability Among Different Kinds of Elements: Some First Observations

There seems to be a borrowability cline of the kinds of structural ele-ments above, with only a few (mostly readily explicable) exceptions to the pattern, but proportions of borrowed lexicon and morphology do not readily correlate. The study suggests that bound inflectional morphology is the least easily borrowed kind of morphology. Many languages which borrowed much of their basic lexicon have borrowed no structural mor-phemes at all (eg. Yapese). KiUnguja, which is the basis of Standard Swa-hili, has a Tense-Aspect-Mood afffijixal system, much of which derives from Northern Swahili varieties which have not provided the bulk of the basic vocabulary (Nurse & Hinnebusch, 1993). Ngandi, without inherited free inflectional morphology, did not introduce it as a fresh category through borrowing. Garifuna lacks a layer of borrowed derivational morphology, at least none has so far been identifijied. As its major loan sources are Antil-lean Creole French and Pidgin Kari’na (the putative source of Kari’na ele-ments in Garifuna), neither of them with much derivational morphology,

Table 3: Overview of general kinds of structural borrowings in the language sample

Language Borrows some function words?

Borrows some bound derivational

morphology?

Borrows some free inflectional

morphology?

Borrows bound inflectional

morphology?

English Yes Yes No NoKalderash Yes Yes* Yes NoHindi-Urdu Yes Yes* No NoAcehnese Yes Yes No NoTagalog Yes Yes (slight) No NoChamorro Yes Yes* Yes NoYapese No No# No# NoIfijira-Mele Yes Yes Yes NoNgandi Yes (few) Yes No# YesPipil Yes No? No NoGarifuna Yes No? Yes YesCuz. Quechua No No No NoBerber: Kabyle Yes Yes* No? NoBerber:  Tamazight

Yes Yes* Yes No

Berber: Siwi Yes Yes Yes YesStandard Sw. Yes Yes* (slight) No Yes

# the category does not exist in the language; * borrowed material is almost exclusively confijined to borrowed items.

99-122_JOHANSON-ROBBEETS_F6.indd 10699-122_JOHANSON-ROBBEETS_F6.indd 106 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

bound morphology in english (and beyond) 107

this lack of borrowing is unsurprising. Otherwise the implicational hier-archy is robust. Diffferent patterns of borrowing morphology in sampled Berber languages are striking.

5 A Comparative Examination of Structural Borrowings

The table below, and the rubric following it, demonstrate how far the lan-guages adopt the same kinds of structural morphs from various model languages. We see that the languages have borrowed widely diffferent features, and widely diffferent numbers of such features. Correlation with the proportion of Swadesh list borrowings is weak, though the language with the greatest proportion of borrowed morphs (Chamorro with 17) has almost a third of its Swadesh list items expressed by loans; Kalderash, Acehnese and Garifuna also score highly. ‘Other’ lists other languages which have borrowed structural morphs with the same role.

The items which were examined for analysis here cover most aspects of morphosyntax. The list of functional items in Creole languages given in Smith (1987) was the source of half of these; the others were added on the basis of my reading of Glendening (1965). These forms were selected because of relatively good availability of data or coverage on them for the language sample. It is not claimed that all the features are universal in human language or that they can be found expressed overtly in every language in the sample.

  1. indefijinite article; 1a: defijinite article; 2. usual noun plural markers; 3. ‘of X’ (possessor of a possessum, whether expressed by afffijixation or by

a free grammatical morph); 4. ‘by X’; 5. ‘to X’; 6. ‘for X’; 7. ‘with X’; 8. (some) nominal case markers; 9. at least one of ‘here’ or ‘there’;10. ‘so/thus’; 11. at least one demonstrative determiner;12. mode of expressing comparison of adjectives;13. mode of expressing the superlative;14. some productively used TAM-marking material;

99-122_JOHANSON-ROBBEETS_F6.indd 10799-122_JOHANSON-ROBBEETS_F6.indd 107 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

108 anthony grant

15. ‘to be’ in at least some of its senses (existential, presentative, locative, equative);

 16. ‘to have’; 17. ‘to become’; 18. ‘to want to’; 19. ‘to have to, must’;20. ‘can, to be able to’; 21. ‘should’;22. usual predicate negator ‘not’;23. prohibitive: ‘don’t (do) X!’;24. ‘and’ linking either NPs or VPs or both kinds of phrases;25. ‘but’;26. ‘or’;27. complementiser ‘that’ or its equivalent;28. ‘if ’;29. ‘because’;30. ‘in order to do X’;31. ‘when’ (temporal dependent clause marker);32. ‘although’;33. at least one personal pronoun;34. at least one indefijinite pronoun;35. a reflexive pronoun;36. a reciprocal pronoun (where these are present);37. general purpose adposition;38. at least some numerals below ‘ten’;39. at least one interrogative word (eg. ‘where?);40. question particle, if it possesses one;41. relative clause marker or relativisation particle, if one is used;42. at least some productively-used person-number marking on verbs;43. the major means of indicating the passive or passivisation;44. the major means of indicating causatives in the verb.

There are examples of borrowing of exponents or forms in at least one of the sampled languages for most of these categories, which reflect most functional categories and membership of form classes, and which have been drawn from those categories found in models of grammatical and typological sketches encountered over the past decades. I have included the handful of categories in which none of the sampled languages have borrowed a form because there are borrowed exponents for these forms in at least one other language. Because of constraints of space I have also

99-122_JOHANSON-ROBBEETS_F6.indd 10899-122_JOHANSON-ROBBEETS_F6.indd 108 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

bound morphology in english (and beyond) 109

taken the high-borrowing Siwi as the representative of the Berber lan-guages which I have examined. Had my selection of sample languages been slightly diffferent because of greater availability of textual, structural and lexical material on these languages, these ‘unborrowed’ categories would have been marked with borrowings too.

The reader will note that the bulk of borrowed functional morphemes in the languages sampled are free rather than bound. The bulk of bor-rowed morphs in Ifijira-Mele, for instance, are free morphs.

6 Structural Borrowings Viewed Crosslinguistically: Comments on the Table

In the table above, borrowed elements are in the minority and types of borrowed elements of which exemplars are difffused language-wide are a minority of these. Examples of borrowed bound inflectional morphemes which can be freely used with inherited and older items as well as with items borrowed from the same stratum or from later strata (rather than only being used with stems which are of the same source as the afffijixes attached to them) are rare in the sample of languages, so that the set of examples here is pretty comprehensive. Ngandi always uses –ku ‘dative/genitive’ and –t:u ‘ergative’, and the prefijix rray-X ‘possessing X’ from Ritharngu on nouns of any origin. Garifuna’s Carib-derived noun plural –gu (lumadagu ‘his friends’ from lumada ‘his friend’) is similarly versatile as nouns of various origins which can be pluralised; it is not confijined to nouns of Kari’na origin.7 In the area of verbal morphology, KiUnguja continuous hu-, anterior –(a)li-, perfect –me- and present tense –ta- and the two conditional afffijixes –ke- and –ngali-, are all taken from Northern Swahili and are used with verbs of all origins, including those of South-ern Swahili origin. The paper by Gardani in this volume can be read with profijit in connection with this.

As to the spread of features involving derivation, Siwi deadjectival abstract nouns of any origin use Arabic morphological templates with Siwi stems: aməllāl ‘white’ > ləmlālat ‘whiteness’, as do comparatives: agzāl ‘short’, gzəl ‘shorter’ and agent nouns: əčč ’to eat’, aččēwī ‘glutton’ (Souag 2009); both Berber and Arabic use templatic morphology. Greek-derived –to- forms all ordinal numerals in Kalderash, including those of Indic

7 The Garifuna irrealis verbal morpheme me, another borrowing from (maybe pid-ginised) Kari’na, is a free morpheme.

99-122_JOHANSON-ROBBEETS_F6.indd 10999-122_JOHANSON-ROBBEETS_F6.indd 109 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

110 anthony grantT

able

4: 

Som

e ‘g

ram

mat

ical

bo

rro

win

gs’ i

n t

he

sam

ple

lan

guag

es. (

Rm

= R

om

ani)

En

Kal

Hi

Ac

TgC

hYp

IMN

gP

iG

aC

QSi

SwO

ther

1(q

)X

XX

YX

XX

XX

XX

Ab

ruzz

i Rm

?1a

XX

XX

XX

XX

XX

Gu

aran

í2

**(

y)*

**

*Y*

**

Cat

alan

Rm

3Q

**

*D

ura

ngo

Nah

ua

4*Y

*N

ort

her

n T

ajik

5Y

*Y

*N

ort

her

n T

ajik

6Y

YY

*YY

*N

ort

her

n T

ajik

7Y

YY

YM

any

Mex

lgs

8X

XX

XX

XY

XX

Wis

hra

m9

Y+

YB

alin

ese;

Sar

amac

can

10K

ild

in S

aam

i11

YY

YSa

lar

12Q

;*Y

YY

?Y

Cah

uil

la13

Q;*

YY

Y?

YB

alka

n R

m lg

s14

**

**

Y*

**Y

**

*YD

ole

njs

ki R

m15

Y+

YX

YM

okl

en16

XX

YX

XX

XSa

ngo

17Y

Swed

ish

18Y

YY

*Y

YD

om

ari

19Y

YY

*Y*

YW

elsh

Rm

20Y

Y*

YM

apu

du

ngu

n21

YY

YY

YT

urk

ish

22

YY

No

rth

Fri

sian

8

23Y

+C

hu

vash

24Y

YY

?Y

+Fa

rsi

25Y

YY

YY

YY

YT

urk

ish

8 A

par

t fr

om

th

e va

riet

y sp

oke

n o

n H

elgo

lan

d w

hic

h u

ses

the

ori

gin

al F

risi

an f

orm

, No

rth

Fri

sian

var

ieti

es h

ave

bo

rro

wed

th

eir

gen

eral

pre

dic

ate

neg

ato

r fr

om

Dan

ish

(o

r in

so

me

case

s Ju

tish

).

99-122_JOHANSON-ROBBEETS_F6.indd 11099-122_JOHANSON-ROBBEETS_F6.indd 110 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

bound morphology in english (and beyond) 111E

nK

alH

iA

cTg

Ch

YpIM

Ng

Pi

Ga

CQ

SiSw

Oth

er

26Y

Y+

YY

Kap

amp

anga

n27

YY

XX

XT

urk

ish

28Y

YY

Y*

YY

*YT

urk

ish

29Y

YY

YY

Q*Y

YY

*T

urk

ish

30Y

YY

*Q

*YY

idd

ish

31*

Y*

No

rth

Ru

ss R

m32

YY

Y?

Yid

dis

h33

YY

YY

YM

any

lgs

34Y

Span

ish

35Y

Du

tch

36Sr

anan

To

ngo

37Y

*Sa

ram

acca

n?

38Y

YY

YY

YY

Man

y lg

s39

YY

YB

ulg

aria

n R

m

40Y

idd

ish

41*Y

YSi

wi;

Do

mar

i42

Do

len

jski

Rm

, 43

YSw

edis

h44

YC

ham

ic lg

s#

412

712

1117

08

86

132

1010

Ru

bri

c: Y

: fo

rm i

s b

orr

ow

ed; X

:- f

eatu

re a

bse

nt,

Y+

: mo

re t

han

on

e w

ay o

f ex

pre

ssin

g th

is c

on

stru

ctio

n i

n t

he

lan

guag

e, a

nd

a b

orr

ow

ed a

nd

a n

on

-b

orr

ow

ed m

ean

s co

exis

t; Q

: th

e p

atte

rn f

or

the

con

stru

ctio

n i

s b

orr

ow

ed b

ut

the

form

use

d i

s n

ot

bo

rro

wed

. A

bla

nk

cell

mea

ns

the

feat

ure

exi

sts

bu

t is

no

t b

orr

ow

ed; *

: th

is i

s ex

pre

ssed

by

a b

ou

nd

mo

rph

eme

bu

t is

no

t b

orr

ow

ed; *

Y: t

his

is

a b

ou

nd

bo

rro

wed

mo

rph

eme;

?: n

o d

ata

is a

vail

able

o

n t

he

pre

sen

ce o

f th

is f

eatu

re i

n t

his

lan

guag

e. #

: to

tal

nu

mb

er o

f b

orr

ow

ed m

orp

hs

in t

his

sel

ecti

on

.

Tab

le 4

(co

nt.

)

99-122_JOHANSON-ROBBEETS_F6.indd 11199-122_JOHANSON-ROBBEETS_F6.indd 111 3/30/2012 9:21:31 AM3/30/2012 9:21:31 AM

112 anthony grant

origin: duj ‘two’, dujto ‘second’ (Boretzky 1994, Bakker 1997), but one can see it as a derivational afffijix, and Kalderash has Greek-derived –mo(s) and plural –mata form abstract nouns from stems from any source (thus pimo ‘drink’, pimata ‘drinks’ from Indic-derived pijel ‘s/he drinks’). Some fur-ther widely borrowed structural features are less universal, but examples can be found in languages throughout the world. One such case is that of numeral classifijiers. Seifart (2007) shows the Arawak language Resígaro of Colombia borrowed all its numeral classifijiers, both fabric and pattern, from the unrelated Witotoan language Bora, and it has used them with nouns of all origins, while at the same time only borrowing about 5% of its total recorded lexicon from Bora or other languages such as Spanish.

Of the 45 features examined in the table, 40 are borrowed (sometimes coexisting with native equivalents) in at least one of the sample languages. The other fijive features (a defijinite article, ‘so/thus’, a reciprocal pronoun, a particle marking polar questions and some bound morphology marking person and number in verbs) are expressed by loans in some other lan-guage. Our sources are silent on the realisation of a few forms in the table in some languages, and certain other features (such as articles) are areal rather than globally distributed features. Features borrowed in at least fijive of the sample languages are: ‘for NP’, ‘to want to’, ‘to have to’, ‘should’, ‘but’, ‘if ’, ‘because’, at a pinch some personal pronouns (not always the same: close matches are forms for ‘they’ in English and Acehnese and forms for ‘I’ in the other three borrowers) and at least ‘7–9’ among lower numerals. These borrowed forms (all have V/CV syllable structure) rarely appear as bound forms; only 9 categories using bound borrowed morphemes appear on the table from a possible 630. At least 29 further slots contain forms showing transfer of pattern without transfer of fabric.

Even if we take into account the fact that some languages in the sample use agglutinative morphology (Quechua, Ngandi, Garifuna, Swahili, argu-ably also Hindi/Urdu, Pipil, Tagalog, and Chamorro though to a slight degree in all these four), while others are broadly isolating (Ifijira-Mele, Acehnese, Yapese), fusional to a greater or lesser degree (English, Kal-derash), or use templatic morphology (Siwi and other Berber languages), it would still be very difffijicult for us to factor it into a grander implicational hierarchy of kinds of structural transfer, as the numerous borrowed items which coexist with inherited items on this table show.

Data on the spread and borrowing of productively-used or diachronic-ally less restricted items of bound derivational morphology is omitted. Such data are less easy to gather and tabulate, partly because of the shortage of data on this topic for some languages (especially because comprehensive

99-122_JOHANSON-ROBBEETS_F6.indd 11299-122_JOHANSON-ROBBEETS_F6.indd 112 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

bound morphology in english (and beyond) 113

Table 5: How diffferent kinds of contact compare across languages. Contact-induced influence in various strata in the languages of the sample, presented on a set of scales from 0 to 5 (information on contact influences upon lexicon is presented on a longer scale calculated from 0 upwards; loans into categories

which are not cumulative are indicated item by item)

Phonetics Phonology Infl Morph. Syntax Lexicon

a b cEnglish 1 <1 1 4 4 4 10Kalderash 2 1 1–2 3 4 6 11Urdu 4 3 1 3 5 5 10Acehnese 1 1 0 3 5 4 10+Tagalog 1 3 1 3 3 4 11+Chamorro 1 1 2–3 3 5 8 13+Yapese 2 <1 0 0 4 6 7Ifijira-Mele 1 1, 2 1–2 1 4 6 8Ngandi <1 0 2 1 7 7 8Pipil 2 2 1 1, 2, 3 0 1 5Garifuna 1 1 2 1, 2, 3 4 6 8+ Cuzco Q. 1 1 0 3 4 6 7Siwi Berber 2 0 3 1? 4 7 12Swahili 1–2 0 3 1–4 4 4 8

Rubrics for the table above run as follows:Phonetics and segmental phonology:1: up to 5 phones are borrowed; 2: 6–10 phones are borrowed; 3: 11–15 phones are borrowed; 4: 16–20 phones are borrowed; 5: over 20 phones are borrowed.Canonical Phonology (the scores are not cumulative in nature):1: complex onsets have been introduced through language contact; 2: simple consonantal codas introduced; 3: complex consonantal codas introduced; 4: lexical tone introduced.Inflectional Morphology (again the total score is not entirely cumulative):0: none borrowed; 1: only derivational morphology is borrowed; 2: nominal inflectional morphology is borrowed; 3: verbal inflectional morphology is borrowedSyntax (these numbers are also not cumulative in nature):0: no syntactic features are transferred; 1: addition of new orders of constituents or adver-bial discourse particles; 2: coordinating clause markers; 3: complementisers; 4: subordinat-ing clause markers; 5: relativisers.Lexicon (in three columns: a-: items on Swadesh 100-item list, b-: items on 223-item [= 100 + 215-item] list, c-: items in the whole lexical sample for the language concerned):1: under 5% of this lexical stratum is borrowed; 2: 6–10% is borrowed; 3: 11–15%; 4: 16–20%; 5: 21–25%; 6: 26–30%; 7: 31–35%; 8: 36–40%; 9: 41–45%; 10: 46–50%; 11: 51–55%; 12: 56–50%; 13: 61–65%; 14: 66–70%; 15: 71–75%; 16: 76–80%.

lists of such morphemes are hard to gather), partly because several of the languages borrow little if any derivational morphology to begin with. It seems there is a continuum of borrowing, from languages such as Yapese, which have borrowed no morphology, to English, where some borrowed derivational morphemes occur with some elements of inherited lexicon too, and the even freer Siwi with copied templates from Arabic, which it

99-122_JOHANSON-ROBBEETS_F6.indd 11399-122_JOHANSON-ROBBEETS_F6.indd 113 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

114 anthony grant

has been able to incorporate into its own morphological system because Berber and Semitic languages both make massive and essential use of templatic morphology.

7 Theorising from Case Studies in a Sprachbund: Jeffrey Heath on Borrowing

Drawing on his fijield studies of Aboriginal languages of Northeastern Arn-hem Land (Ngandi, Nunggubuyu, Warndarrang, Mara, Ritharngu, Dhu-wal) Heath (1978b: 105) lists those kinds of structural items which are transferred from at least one language to another unrelated one in this speech area, and just as importantly, he lists those kinds of items which resist such transfer in this area. He discusses various kinds of borrowing, including cases of pattern transfer (a term he coins; his usual term for this kind of transfer in Heath 1978b is ‘indirect difffusion’, whereas ‘direct difffu-sion’ corresponds to the transfer of actual morphs possessing phonologi-cal forms). Additionally, Heath’s fijive criteria for the linguistically-based factors which enable the transfer of such afffijixes from the productive morphological system of one language to another have not received the attention they should have, although I fijind them valid, insightful, well-motivated and I cite them later. Heath’s list of the morphological catego-ries of difffusible and indifffusible items in Arnhem Land languages has received more attention in the linguistic literature; I provide it below. All of these items reflect instances of the borrowing of fabric rather than of pattern (in which latter division I include borrowing of category).

The material listed here reads like an outline for a sketch grammar of a language, but of course there are other internally-classifijiable categories of non-contentive morphemes which can also be surveyed and compared for degrees of borrowability. Furthermore, the strictures here against bor-rowing certain classes of items are contravened by some other languages spoken outside Arnhem Land. For instance, quite a few of the world’s lan-guages borrow some independent pronouns either from linguistic rela-

Yapese > Swahili > Ifijira-Mele > Chamorro > English > SiwiCuzco Quechua Tagalog KalderashPipil Acehnese NgandiGarifuna Urdu

Figure 1: Scale of retroactively transferred borrowed bound morphology in the sample languages to the amount of bound morphology as a whole, from least to greatest proportion borrowed.

99-122_JOHANSON-ROBBEETS_F6.indd 11499-122_JOHANSON-ROBBEETS_F6.indd 114 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

bound morphology in english (and beyond) 115

tives or from unrelated languages (English they, them, their from Norse is just one set of examples among many, see also Thomason and Ever-ett 2000, and Tagalog tayo ‘you and I’ from Kapampangan, Chamorro yo [dzʊ] ‘I’ from Spanish yo). Standard Swahili, a Southern Swahili variety, has borrowed several unbound personal pronouns (in fact all except the 3rd person ones), a demonstrative pronoun –le ‘that’ and, as shown above, several TAM verbal afffijixes from Northern Swahili (Nurse and Hinnebusch 1993).

Heath (1978b) is very much still a valid contribution to the fijield of the study of contact-induced change. Heath’s fijive structural criteria relating to the smoothness of borrowing, which are intended to be widely valid rather than to be specifijic to Eastern Arnhem Land languages in their application, and which determine the ease with which morphology can be reinstalled into a new language (at least an agglutinative one), are as follows: syllabicity (the principle that the morpheme constitutes a sylla-ble), sharpness of boundaries (this means the morpheme is easy to identify and detach from surrounding morphemes), unifunctionality (which dem-onstrates that the role of the morpheme is clear and unambiguous), and the two less readily apprehensible ones, namely categorical clarity (which means that the morpheme can be labelled or identifijied clearly and its role can be apprised independently), and analogical freedom (the absence of unidirectional analogical pressure from another morpheme). Heath’s list of features presented above suggests that those forms which consti-tute distinct and easily perceptible syllables (especially CV syllables), and which have a single major functional role in the structure of the language which can be easily recognised and apprehended by listeners, and which are equipped with shapes that are not easily confused with those of other

Table 6: Difffusible and undifffusible structural features in languages of North-eastern Arnhem Land (Heath 1978b)

Difffusible Undifffusible

Case afffijixes independent pronounsNumber afffijix bound pronominalsNoun-class afffijixes (several) verbal inflectional afffijixesDiminutive afffijix demonstrative stemsDerivational verbal afffijix demonstrative adverbsNegative afffijixOther postpositions (several)Inchoative verbalizerThematizing augmentSpecial compound initials

99-122_JOHANSON-ROBBEETS_F6.indd 11599-122_JOHANSON-ROBBEETS_F6.indd 115 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

116 anthony grant

forms which appear in the same slot, are the kinds of morphemes which are most easily transferred from one language to another.

8 Borrowed Bound Inflectional Morphology which is Used Productively in the Sample Languages

One conclusion arising from an examination of Heath’s work is that cer-tain kinds of items which are generally assumed to be unborrowable may in fact be borrowed and integrated into a language system if the system of the basic language (or the will of the speakers of these languages) is open to such borrowings. The more isomorphic the typologies of the source and basic languages are, the greater the ease with which borrowing may occur.

Another consideration, and an important one in this paper, is that of the degree of nativisation of the morphs in question. This means that a nativised morph is one which is used in its basic language not only with items borrowed from the language which provided this morph, but that it is also used with morphs of other origins, often including its application onto morphs which come from the genealogical stratum in the language in question. Latin-origin –able is regarded as a nativised morph in English as it can be used not only with items which share its Latin origin but also with morphs of other origins, including Old English origin, such as in the word washable. One of the characteristics of nativisation in this schema is that the nativised borrowed item in question can be used productively: it can be readily applied to new words of any origin, and not just to bor-rowed items.

Even when we can decide what we mean by the ever-ambiguous term ‘grammatical borrowing’, the task of comparing morphological transfer across languages is, however, difffijicult for a number of reasons. Some of these follow (and there are others):

1)   Some languages possess particular morphological categories which others lack. For instance marking of defijiniteness on noun phrases, tense-marking on verbs, and obligatory evidentiality markers in verb groups occur in some but by no means all languages, yet all languages have the means to negate the content of a predicate;

2)  Semantic categories at various levels (including the internal structures of semantic fijields) and also the types of form-classes may also difffer between languages. For instance adjectives or property items behave

99-122_JOHANSON-ROBBEETS_F6.indd 11699-122_JOHANSON-ROBBEETS_F6.indd 116 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

bound morphology in english (and beyond) 117

like stative or intransitive verbs in many languages, and use verbal morphology, but they act structurally like nouns in many respects in other languages. This has consequences for the kinds of morphology which they take. Meanwhile, Ngandi has a system of prefijixing noun-classes, semantic in origin, which Ritharngu lacks and into which bor-rowed Ritharngu nouns must fijit.

3)  Two languages may nonetheless use the same typological categories but one language may use bound morphology to express or encode (some of) them while the other language signals these features using free-standing morphs to fulfijil the same function.

In short, the typological diffferences between source and basic languages may sometimes be massive, even though the languages are spoken by neighbouring speech communities and may or may not be genealogically related. Yet although typological dissimilarity does not seem to impede the borrowing of lexicon, it may impede extensive transfer of morphemes which cannot easily be detached and isolated according to the criteria which Heath (1978a, 1978b) described for Arnhem Land languages includ-ing Ngandi. These can then be integrated into the language’s structure in such a way that subsequently acquired lexical items can have these bor-rowed morphs (which belong to earlier loan strata) applied to them.

As a result, the degree of borrowing of morphology overall in the lan-guages under consideration is not great. Function words, that slippery metacategory whose contents themselves vary from one language to another, are borrowed more easily than items of bound derivational mor-phology, which themselves are borrowed and integrated (one hesitates to say ‘nativised’) more easily than items of bound inflectional morphology. (Not all morphologists see a distinction between inflectional and deriva-tional morphology as ill founded.)

But bound inflectional morphology can become the dominant or sole mode of expression of a concept in certain languages, and we must not imagine that we will always be able to fijind an explanation for such a replacement; change through replacement is a crucial fact of diachrony, but the motivation for such change may be little more than guesswork. Our explanations may really be ex post facto rationalisations of the kind which states that some morphemes being added to a language’s morphological system because they thereby help reduce the number of tasks which a hitherto widely-used morpheme subsequently performs. This would con-fuse outcome with process (Babs Richardson, personal communication.)

99-122_JOHANSON-ROBBEETS_F6.indd 11799-122_JOHANSON-ROBBEETS_F6.indd 117 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

118 anthony grant

Let us assume that Language A receives a grammatical morpheme (say a noun pluraliser) from Language B, which in the course of time it incor-porates into its own grammatical system and which it renders increas-ingly productive, so that it is used with more and more nouns in the basic language. The stages leading up to this would go as follows:

Stage 0: this is the state found in most languages, whereby a gram-matical concept (for example the marking of plurality in NPs) is always expressed by a morph (or a set of morphs) which derives from language A and not from another language—this is the usual state of afffairs in lan-guages which have not absorbed any instances of borrowed morphology;

Stage 1: a lexical form from language B is borrowed, together with the morph encoding this concept (e.g. a noun borrowed with its plural marker; many examples exist, for instance Kalderash kambana—kambani ‘bell—bells’ into Romani from Greek, which itself borrowed the stem from Latin);

Stage 2: the use of this marker spreads to more, then most, then prob-ably all of the other nouns (or whatever) from language B found in lan-guage A, whether or not this is historically justifijied (this is the case with the use of –s as a marker of pluralisation with Spanish nouns in some varieties of Quechua: Gómez Rendón 2009;

Stage 3: by now the usage has spread to appropriate forms deriving from later contact languages, namely languages C, D and E, which are subsequently taken into language A after the borrowings from Language B have begun. It thus becomes the customary form to be used with bor-rowed items of any loan origin (eg. –uri, the noun plural marker deriv-ing from Romanian, itself deriving from a Latin neuter plural –ora, which occurs in some Vlax Romani dialects, and -i plural noun markers from Greek used in some other Romani dialects; an anonymous referee has sug-gested that this stage may not be found if the afffijixes which are involved in it are no longer productive in the basic language).

Stage 4: the use of the borrowed marker spreads (fijirst of all optionally, then compulsorily) to forms from Language A (we may call this supple-mentation). There are very few examples of this, although the beginnings of this stage can be seen in certain instances in Romani varieties, for instance the spread of a Greek-derived past participial ending in Finnish Romani to inherited verbs of Indic origin (Valtonen 1971);

Stage 5: using the borrowed marker becomes the general case with forms in language A which were part of the language before the adop-

99-122_JOHANSON-ROBBEETS_F6.indd 11899-122_JOHANSON-ROBBEETS_F6.indd 118 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

bound morphology in english (and beyond) 119

tion of this form from language B fijirst took place (‘supplanting’; the use of a Greek-derived ordinal morpheme with Indic numerals in Kalderash, stated above, is one case);

Stage 6: by now the use of the original morphological forms in language A is now held to be irregular and anomalous within the more regular and revised system of the language, thus bringing about ‘partial suppletion’. A partial parallel to this are relic ablaut plurals in English, such as man—men, foot—feet, and the invariable sheep, remnants of a more widespread system of noun pluralisation;

Stage 7: supplementation of the old form by the new has reached such a stage that philological or other evidence from earlier documentation of the language is now the only source of evidence remaining which can show that the original morphological form existed and that it has been completely supplanted in the modern form of the language. The earlier use of –eth to mark the 3rd singular present indicative, which although it is of Indo-European vintage is not found in Present Day Standard English texts but is recorded as being in regular use into the 17th century, is an instance of this.

9 Conclusions

Following the application of our methods, and looking comparatively at our data, we fijind that certain conclusions arise in varying degrees of clar-ity. Patterns of borrowing or copying, including copying of bound mor-phemes, have been presented and compared in the sample languages. Some correlation occurs between levels of heavy lexical borrowing (and importantly, heavy lexical replacement of older stems through borrowing new forms) and the extensive replacement of bound and other structural morphemes via borrowing, but this correlation is rather weak (Tables 2–5). Nor are the same kinds of structural morphemes borrowed in the same sequence (if at all) in all languages. The amount of gradience of fusion of morphological and other elements in heavily borrowing languages var-ies from case to case, and from one morph type to another, and it does not correlate too well either with the widespread replacement of Swadesh list and other basic vocabulary or with the widespread borrowing of pro-ductively-employed morphology (especially if it is bound, inflectional or both). General resistance to borrowing elements into the 100-item and to a lesser extent longer Swadesh lists (compared with longer lexical sam-ples) is, however, borne out strongly.

99-122_JOHANSON-ROBBEETS_F6.indd 11999-122_JOHANSON-ROBBEETS_F6.indd 119 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

120 anthony grant

Some languages borrow plenty of items of morphology, especially items of derivational (and they transfer many patterns) but rather little lexicon. Pipil borrowed fairly little very basic Swadesh-list lexicon but took most of its conjunctions from Spanish within the last 500 years and developed a new coordinating conjunction wan ‘and’ from a postposition meaning ‘with’ which still remains in use in Pipil (Campbell 1985). Siwi, in contact with Arabic for over 1000 years, still retains most of its Berber conjunctions (especially its coordinators) but borrowed (and has thereby replaced) much basic lexicon from Arabic. Urdu has replaced many of its original conjunctions (and about a sixth of its basic lexicon) with forms from Persian, many of these themselves originating in Arabic.

The reasons for borrowing are not wholly shaped by the prior structure of languages. In some languages certain borrowed bound morphemes also used with inherited stems and borrowed items acquired subsequent to the acquisition of the borrowed bound morphemes. In other languages (such as Hindi/Urdu) borrowed morphemes are only used with borrowed lexicon and they have not leaked into the pre-borrowed lexicon. Yet other languages (such as Yapese) have replaced much of their basic lexicon with loan elements but have borrowed no bound or free morphology for use with morphs of any origin.

The study suggests that the principles of borrowing that were devel-oped in Heath (1978b) are of much wider relevance as a general means of explaining what is likely to get borrowed and why, but it is easier for us to understand ex post facto why a particular feature was borrowed than to predict that it will be.

References

Abdel-Massih, Ernest T. 1971. A computerized lexicon of Tamazight (Central Atlas Berber). Ann Arbor: Center for Near Eastern Studies, University of Michigan.

Bakker, Peter. 1997. “Athematic morphology in Romani: The borrowing of a borrowing pattern”. In The typology and dialectology of Romani, edited by Y. Matras, P. Bakker & H. Kyuchukov, 1–21. Amsterdam and Philadelphia: John Benjamins.

Bakker, Peter and Maarten Mous, eds. 1994. Mixed languages: 15 case studies in language intertwining. Amsterdam: IFOTT.

Boretzky, Norbert. 1994. Romani: Grammatik des Kalderaš-Dialektes mit Texten und Glossar. Wiesbaden: Otto Harrassowitz.

Campbell, Lyle. 1985. The Pipil language of El Salvador. Berlin and New York: Mouton de Gruyter.

Canger, Una. 2001. Mexicanero de la Sierra Madre. Mexico City: El Colegio de México.Castro Medina, Margarita María. 2000. Descripción morfológica del sistema verbal del mexi-

canero de San Pedro Jícora. Variante dialectal del nahua en Durango, México. PhD diss., University of Cologne.

99-122_JOHANSON-ROBBEETS_F6.indd 12099-122_JOHANSON-ROBBEETS_F6.indd 120 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

bound morphology in english (and beyond) 121

Cayetano, E. Roy. 1993. The people’s Garifuna dictionary. Dangriga: National Garifuna Council of Belize.

Clark, Ross. 1993. “The Polynesian Outliers as a locus of language contact.” In: Language Contact and Change in the Austronesian World, edited by T. Dutton and D. Tryon, 109–139. Berlin: Mouton de Gruyter.

——. 1998. A dictionary of the Mele language (Atara Imere). Canberra: Pacifijic Linguistics C-149.

Comrie, Bernard and Norval S. H. Smith. 1977. “Lingua Descriptive Series Questionnaire.” Lingua 42: 1–77.

Cusihuamán, Antonio. 1976. Diccionario Quechua Cuzco-Collao. Lima: Ministerio de edu-cación del Perú.

Dallet, Jean-Marie. 1985. Dictionnaire français-kabyle. Paris: SELAF.Embleton, Sheila M. 1986. Statistics in historical linguistics. Bochum: N. Brockmeyer.Fochi, Camillo, and Giulio Soravia. 1995. Vocabolario sinottico delle lingue zingare parlate

in Italia. Rome: Centro Studi Zingari.Formoso, Bernard, and Georges Calvet. 1987. Lexique tsigane II: sinto piémontais. Paris:

POF-Etudes.Gardani, Francesco. 2011. Plural across inflection and derivation, fusion and agglutination.

In this volume, XX.Gjerdman, Olof, and Erik Ljungberg. 1963. The Language of the Swedish Coppersmith Gipsy

Johan Dimitri Taikon. Uppsala: A.-B Lundequist.Glendening, P. J. T. 1965. Teach yourself to learn a language. London: English Universities

Press.Gómez Rendón, Jorge. 2009. ‘Loanwords in Imbabura Quechua.’ In Loanwords in the

World’s Languages: a comparative handbook, edited by Martin Haspelmath and Uri Tad-mor, 944–967. Berlin and New York: Mouton de Gruyter.

Grant, Anthony P. 1999. Mixed languages: a conspectus, April 1999. Unpublished manu-script, University of Southampton.

——. 2002. “Fabric, Pattern, Shift and Difffusion: What Change in Oregon Penutian Lan-guages Can Tell Historical Linguists.” In Proceedings of the Meeting of the Hokan-Penutian Workshop, edited by Laura Buszard-Welcher, June 17–18, 2000, University of California at Berkeley. Report 11, Survey of California and Other Indian Languages, 33–56. Depart-ment of Linguistics, University of California at Berkeley.

——. 2006. “Correlations in some heavy borrowing languages: the Swadesh list and beyond.” Paper presented at the Max Planck Institute for Evolutionary Anthropology, Leipzig, 26 September 2006.

Grant, Anthony P., and P. Baker. 2007. “Comparative Creole typology and the search for the sources of Mauritian Creole features.” In The making of Mauritian Creole: analyses diachroniques à partir des textes anciens, edited by Philip Baker and Guillaume Fon Sing, 197–220. London: Battlebridge.

Headley, Robert K. 1965. Dictionary of the Jarai language. Washington DC: the author.Heath, Jefffrey. 1978a. Ngandi grammar, texts and dictionary. Canberra: AIAS.——. 1978b. Linguistic difffusion in Arnhem Land. Canberra: AIAS.ILCAA (Institute for the Study of Languages and Cultures of Asia and Africa). 1968. Linguis-

tic questionnaire for Asia and Africa. Tokyo: Tokyo University for Foreign Studies.Kouwenberg, Silvia. 1994. A grammar of Berbice Dutch Creole. Berlin and New York: Mou-

ton de Gruyter.Lafont, Pierre-Bernard. 1968. Lexique Français-Jarai-Viêtnamien: (parler de la Province de

Plei Ku). Paris: École Française D’extrême-Orient.Laoust, Emile. 1931. Siwa I: son parler. Paris: Ernest Leroux.McFarland, Curtis D. 1989. A frequency count of Pilipino. Manila: Linguistic Society of the

Philippines.McGregor, R. S. 1993. The Oxford Hindi-English dictionary. Oxford: Clarendon Press.

99-122_JOHANSON-ROBBEETS_F6.indd 12199-122_JOHANSON-ROBBEETS_F6.indd 121 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM

122 anthony grant

Nurse, Derek, and Thomas J. Hinnebusch. 1993. Swahili and Sabaki: a linguistic his-tory. Berkeley and Los Angeles: University of California Press (UCPL 121). OED online. Accessed December 2010.

Penchoen, Thomas. 1973. Tamazight of the Ayt Ndhir (Afro-Asiatic Dialects I). Los Angeles: Undena.

Perrott, Daisy Valerie, revised by Joan Russell. 2002. Teach Yourself Swahili dictionary. Rick-mansworth: Teach Yourself Books.

Ross, Malcolm. 1996. “Contact-induced change and the comparative method: Cases from Papua New Guinea.” In The comparative method reviewed: Regularity and irregularity in language change, edited by Mark Durie, M. D. Ross, 180–217. New York: OUP.

——. 2009. “Loanwords in Takia, an Oceanic language of Papua New Guinea.” In Loan-words in the World’s Languages: a comparative handbook, edited by Martin Haspelmath and Uri Tadmor, 717–746. Berlin and New York: Mouton de Gruyter.

Rubino, Carl R. Galvez. 1998. Tagalog-English, English-Tagalog Concise Dictionary. New York: Hippocrene.

Saxena, Anju. 2010. “Loanword Typology wordlist for Hindi.” Accessed 12 December 2010. https://svn.spraakdata.gu.se/sb/IDS/pub/lwt-hin.html.

Seifart, Frank. 2007. “Studying language contact in language documentations: Bora influ-ence on Resígaro (North West Amazon).” From searches to fijindings: using language documentation resources. DoBeS Workshop, Nijmegen, 14–15 June 2007.

Siu, Lap Minh. 2009. “Developing the First Preliminary Dictionary of North American Jarai.” MA Thesis, Texas Tech.

Smith, Norval Selby Hunter. 1987. The genesis of the creole languages of Surinam. Unpub-lished PhD dissertation, Universiteit van Amsterdam.

Souag, Lameen. 2009. “Siwi and its signifijicance for Arabic dialectology.” Zeitschrift für ara-bische Linguistik 51: 51–75.

Thomason, Sarah Grey, and Daniel Everett. 2000. “Pronoun borrowing.” In Proceedings of the 27th meeting of the Berkeley Linguistics Society, edited by Charles Chang, Michael Houser, Yuni Kim, David Mortensen, Mischa Park-Doob & Maziar Toosarvandani, 301–315. Berkeley: UC Department of Linguistics. (Appeared 2005.)

Thurgood, Graham. 1999. From Ancient Cham to modern dialects: two thousand years of language contact and change. Oceanic Linguistics Special Publication 28. Honolulu: Uni-versity of Hawai’i Press.

Valtonen, Pertti. 1971. Suomen mustalaiskielen etymologinen sanakirja. Helsinki: Tietoli-pas.

Weinreich, Max. 1973. Geshikhte fun der Yidishn Shprakh: bagrifn, faktn, metodn. New York: YIVO.

West, Michael P. 1948. The General Service List of English Words. Oxford: Oxford University Press.

Williams, Joseph M. 1975. Origins of the English language. New York: Free Press.

99-122_JOHANSON-ROBBEETS_F6.indd 12299-122_JOHANSON-ROBBEETS_F6.indd 122 3/30/2012 9:21:32 AM3/30/2012 9:21:32 AM