Phonology and morphology of modern Sinhala

ABSTRACT

This dissertation organizes the Modern Sinhala lexicon within the fram ew ork proposed by Lieber in The Organization o f The Lexicon, 1981. Following Lieber, Sinhala verb, noun and affix morphemes are analysed descriptively and th eir lexical processes, both derivational and compounding, are examined within the th e o re tic al param eters of th a t work.

This dissertation argues th a t tense marking in verb morphology and number in noun morphology (which also includes definite and indefinite marking) are best accounted for as derivational processes in co n trast to the traditionally accep ted inflectional processes. It is claim ed th a t in noun and verb categories, allomorphy is a fa c to r in the underlying representation of the lexicon. On the o th e r hand, allomorphy in noun num ber marking affixes is due to a morphological rule sensitive to class membership of nouns. Going beyond The Organization o f The Lexicon which is lim ited to morphological rules, this study includes phonological rules operating in the lexicon.

Finally, in accord with the results of this d issertatio n two suggestions are made; F irst, it is essen tial th a t a com plete study o f Modern Sinhala phonology include an extensive investigation of the syllable stru c tu re ; Second, the concept of level ordering in the lexicon should be utilized to acco u n t for the h ierarch ical distribution of affixes. Support for the firs t suggestion re sts on th e

A b s tr a c t... ü

C o n te n ts ... iv

Acknowledgements ... vi

D e d ic a tio n ... viii

C hapter I: Background ... 1

1.1 Brief History of Sinhala ... 1

1.2 Early G ram m atical S tu d i e s ...2

1.3 R ecent Studies about the Language and This D is s e r ta tio n ... 3

1.4 T heoretical A p p ro a c h ... 3 1.5 C hapters to f o llo w ... 3 C hapter H: Verb M o rp h o lo g y ... 6 2.1 Prelim inaries ... 6 2.2 The Three T e n s e s ...8 2.3 V e rb -N o u n s... 11

2.4 More alio morphs ...12

2.5 Compounds ... 14

2.6 The Word Form ation Component according to O L ... 16

2.7 Tense Form ation and Stem B u ild in g... 23

2.8 Subcategarization R e s tr ic tio n s ... 47

2.9 Nominal D eriv a tio n s...50

2.10 E x c e p tio n s... 53

2.11 C o n clu sio n ... 56

C hapter IE: Noun M o rp h o lo g y ... 59

3.1 P re lim in a rie s ...59

3.2 Nominal C la ss e s ... 60

3.3 Underlying Forms of the No m in a is... 64

3.4 Morphological Rule A p p lic a tio n s...68

3.5 Phonological R u le s ... 72

3.6 D erivation of Feminine Gender N o u n s ... 72

3.7 Definite and Indefinite M ark ers... 76

3.8 D efinite- Indefiniteness in the P l u r a l ... 79

3.9 C onclusion... 81

-4.1 P re lim in a rie s ... 82

4.2 N ature of the C o m p o u n d s... 84

4.3 T heoretical A p p licatio n s...86

4.4 Indefinite Plural Marking of Nouns ... 87

4.5 C onclusion... 90

C hapter V: Phonology and Morphology of Modern S in h a la ... 91

5.1 Stem V a ria tio n ...91

5.2 Phonological R u le s ... 92

5.3 Rule Ordering ...95

5.4 Level Ordering and Cyclic Application of R u l e s ...101

5.5 Syllable S tructure... ... 104 5.6 C onclusion... 113 B ib lio g ra p h y ... 114 Appendix A: V e r b s ... 122 Appendix B: N o u n s ... 130 V

-My sincere thanks go to Dr. B. F. Carlson, my supervisor who insightfully and metaculously guided me throughout my graduate programme.

I thank very sincerely the members of my dissertation com m ittee, Dr. T. M. Hess, Dr. H. J. W arkentyne, Dr. S. Gunasinghe, and the external exam iner Dr. G. D. Prideaux who are all resposible for raising the questions th a t led to many improvements in this work.

Dr. G. N. O'Grady, Dr. T, E. Hukari, and Dr. D. B ates discussed p arts of this dissertation with me. To them all, I am grateful.

I wish to thank other members of the faculty of the D epartm ent of Linguistics and in particular, Dr. J. F. Kess, who as chairman and te ach er always encouraged me; Dr. B. P. H arris for her concern and ready cooperation as G raduate Adivsor.

Thanks to Ms. Darlene Wallace and Mrs. G retehen Moyer of the D epartm ental office fo r th eir consistant support.

A special word of thanks is owed the Gunasinghe fam ily. Their company and friendship supported me g reatly through these years. In th e ir kitchen, Siri, Hemi and I discussed and argued the many points of Sinhala gram m ar as this dissertation took shape. I will never fo rg et those tim es which always included delicious Sri Lankan meals which Hemi prepared!

I also wish to acknowledge the moral support given me by ray fellow graduate students in th e D epartm ent: Andrea, Suzanne, Michael, Chiharu, Sandra, Erica,

Kimary, Lydia, Nadia & Daniel, Lucie, and B arbara. They have been a close group of friends who added immeasurably to the pleasure of my graduate studies. Some of these people have continued to support and encourage me even a fte r their leaving the University.

My friends in Victoria have always made me feel a t home in Canada. I owe them a word of appreciation, especially Doubravka (Dee), Lorna & Auntie M argaret, Frank, H eather and Robert Gray.

Final thanks go to the Canadian Commonwealth Association for its continued financial assistance throughout my graduate programme.

-With due respect to my teach er Ven. Mampe W imalakeerthi There

and with Love to my m other and fath er

and

to the memory of my grandm other

BACKGROUND

1.1 B rief History of Sinhala

Sinhala is spoken by over 12 million people in Sri Lanka. Investigations done in the past have a tte ste d to the fa c t th a t Sinhala belongs to th e Indo-European Family of Languages. Geographically cut off from the mainland and confined to an island off the southern tip of India, Sinhala, however, has always been in touch with its sister languages - such as Sanskrit and Pali - in the northern India. In the course of its evolution, Sinhala has come in co ntact with not only Sanskrit and Pali but also w ith Tamil, Portuguese, D utch and English. From the la tte r group, Tamil is the only Non-Indo-European language while the re st rep resen t two branches - Germanic and Romance - of th e Indo-European language family.

The history of the Sinhala Language has been tra c ed back to the 3rd century B.C.. According to the classification of the Sinhala D ictionary, the evolution of the language has been divided into th e following stages.

1. Sinhala P rakrit (3c. B.C. - 4c. A.D.) 2. Early Sinhala (4c. A.D. - 8c. A.D.) 3. Middle Sinhala (8c. A.D. - Mid 13c. A.D.) 4. Modern Sinhala (Mid 13c. A.D. - the present)

Sinhala gram m atical concepts are inherited from those which began m illennia ago for describing Sanskrit and Pali. These ancient gram m atical traditions characterize the most valued description of Sinhala y et w ritten, Sidat Sangarawa (literally manual of rules), which was compiled in the 13th century. Essentially, all subsequent Sinhala gram m ars use the Sidat Sangarawa as th eir descriptive model, a model th a t is not as adequate for conveying the stru c tu re of Sinhala as are several 20th century theories of gram m ar. Furtherm ore, these Sinhala gram m ars a tte m p t to describe L iterary Sinhala, an archaic form of the language which no one speaks although it is still taught in schools and is the w ritten medium of expression for many purposes. The living spoken language of the Sinhala people has generally been ignored by gram m arians. As for example, the column B in the following shows the spoken Modern Sinhala as opposed to its w ritten v ariety in column A.

A B

minisa yay miniha yanawa

"The man goes" "The man goes" minissu yati minissu yanawa "The men go" "The men go"

oba yannehi oba yanawa

"You(sg.) go" "You(sg.) go" obala yannahu obala yanawa "You(pl.) go" "You(pl.) go"

mama yami mama yanawa

"I go" "I go"

api yamu api yanawa

1.3 R ecent Studies about th e Language and This D issertation

In recen t decades, however, there have appeared a few p artial descriptions of this or th a t aspect of modern, spoken Sinhala. See, for exam ple, the work of Abhayasinghe, (1972/73); De Silva, (1972); Dissanayake, (1969); Gair, (1970); Gunasinghe, (1978, 1985). This dissertation a tte m p ts to fu rth er this work on Modern Spoken Sinhala (h ereafter Modern Sinhala) by describing in detail the phonological and morphological stru c tu re of its vocabulary. As it will be noticed, the dem arcation for the Modern Sinhala era, according to th e Sinhala Dictionary, is too vast a span. By the term Modern Sinhala, this study refers to th e language of th e past five decades. Furtherm ore, considering the d ialectal variations in Modern Sinhala, th e d a ta provided in this belong to the standard dialect in Colombo, the cap ital city of the country.

1.4 T heoretical Approach

The th eo retical fram ew ork adopted here is th a t se t fo rth in The Organzation o f the Lexicon (h ereafter OL) by R. Lieber, 1981. This fram ew ork proposes a lexicon consisting of th ree subcomponents including morpholexical rules, binary branching lexical tre e s and conventions to perco late fe atu res up to the branching tre e s. (These th e o re tic al concepts are described in C hapter II.)

1.5 C hapters to follow

Chapters II, III and IV respectively discuss the topics of verb morphology, noun morphology and word-compounding. Each of these ch ap ters firs t introduces its topic according to the traditio n al method before proceeding to analyse the sam e

in the adopted fram ework. This is expected to give the prospective readers who are fam iliar with the traditional methods a little understanding in distinguishing the difference of the approach th a t this study has taken from th a t of the traditional. In accord with the premises presented in the OL theory, derivation and inflection, two well-known param eters of morphology, have been tre a te d as a unitary lexical process. The major focus of C hapter II is on the four-w ay division of the verb. It is argued here th a t this division is a result of a derivational process ra th e r than stem allomorphy {although such allomorphy does seem to be the major fa c to r in stem variation of a small group of verbs discussed a t the end of this chapter). Allomorphy also accounts for the stem variations of some nominals discussed in C hapter III. Where justified by the th e o re tic al c rite ria , a ltern atin g stem shapes are tre a te d as an existing stock in the perm anent lexicon and morpholexical rules express the relations w ithin se ts of such a ltern atin g stem shapes.

In Chapter 111 where stem allomorphy is commonly observed, nominals are divided into classes with each class consisting of a morpholexical rule. The existence of nominal classes is, in turn, held to be the reason for having several alio morphs for the number marking affix. These a lte rn ativ e affix morphemes are, however, argued to be derived by morphological rules. It is shown th a t nominal class membership plays a m ajor role in the form ation of morphological rules.

C hapter IV addresses the controversial issue of word-compounding in Modern Sinhala. Despite the absence of strong phonological argum ents to support the notion of Modern Sinhala word-compounding, Lieber’s mechanism of fe a tu re percolation convention continues to bear the only morphological evidence th a t is

found in its favour. It is also shown briefly th a t trad itio n al descriptions of compounding have relied on sem antic derivations.

The last ch ap ter co n trasts the results of this study with o th er contem porary theories such as Lexical Phonology and Morphology. Here too are presented some of the problems encountered with Lieber's fram ew ork and some solutions are suggested. Morpholexical rules are indicated as being a strong argum ent for underlying syllable stru c tu re of language, while derivational processes such as verb stem form ation are taken to suggest the existence of a level ordering in the lexicon. However, the interactions of phonology and morphology in the lexicon s till need fu rth er studies.

The fram ework provided in the OL theory has, overall, proved to be a simple and effectiv e mechanism fo r organizing the Modern Sinhala lexicon. It is believed th a t this description of Sinhala provides a useful contribution to the growing understanding of the word form ation in human language.

VERB MORPHOLOGY

In a study of the verb morphology in Modern Sinhala, one main point th a t has to be addressed is the division of a verb into four stem shapes. W hether or not this stem variation is a phenomenon of stem allomorphy in the lexicon, as pointed out for some modern Indo-European languages by Lieber, has y et to be investigated. In what follows, the four-w ay division of the verb stem will be investigated as it appears in the th re e tenses to g ether with the verb-noui. derivations.

2.1 Prelim inaries

The following are the present tense forms of the verb /b a la / "to see" in Modern Sinhala.^

2

.

1.1

a) bale - -ne (-wa) "look/read" b) b a le - -n e (-wa) "see"

e) b a le -we -n e (-wa) "make (someone) look" d) b a le -we -n e (-wa) "make (someone) see"

^ Unless otherw ise noted or se t off by slanted v e rtic al lines (/), th e exam ples given throughout this discussion are su rface form s.

-According to trad itio n al gram m ars, the difference betw een (a) and (b) in 2.1.1 is th a t the form er is a volitional a c t whereas the la tte r is involitional. In some ways this is sim ilar to the difference betw een tran sitiv e and intransitive form s of the verb. For example,

2

.

1.2

(a) a ri-n a (-w a ) "open" (as in 'He opens the door') (b) ære-na(-wa) "open" (as in 'The door opens')

However, fu rth er studies reveal th a t th ere is something more than the transitive and the intransitive difference operating betw een these two types of verbs. L iterary Sinhala utilizes these two types of verbs to denote the active and passive voices. Spoken Modern Sinhala, having lost the passive construction, now uses the stem of type (b) in 2.1.2 which denotes process to g eth er with lack of volition. Verbs of type (a) in 2.1.1 and 2.1.2 have been named A ctive, whereas those in (b) 2.1.1 and 2.1.2 are called Processive (cf. Gunasinghe 1976). The other two verb types shown in (c) and (d) of 2.1.1 are the Causative co u n terp arts of Active and Processive verbs respectively. In this study, type (c) is named A ctive Causative and (d) is Processive C ausative. Thus, a verb in Modern Sinhala consists of four stru ctu ral shapes which rep resen t four d ifferen t aspects. These are Active (=A), Processive (=P), A ctive Causative (=AC), and Processive C ausative (=PC).

2.2.1 Present 2

1 2 3 4

A= bala - -na {-wa) "Icok/read” P= b a le - -n a (-wa) "see"

AC= b a la -wa -na (-wa) "make (someone) look" PC= b a la -we -n a (-wa) "make (someone) see"

The column 1 of 2,2,1 shows the stem shapes of the verb root /b a la /. As can be observed, there are two main stem variations-bala " b a l e . It is assumed here th a t the two suffixes in column 2 are stem forming morphemes and thus th ere are two more stem variants balawa and balawe , making four the number of stem alio morphs a verb can have.

The -wa /-w e suffix in column 2 has been described as a causative m arker. A ltogether, columns 1 and 2 provide enough inform ation as to th e stem variations among the four verb groups in question. Between A and P, th e major difference is th a t the la tte r has fro n t vowels as opposed to the back vowels in the fo rm er. The difference between A and AC is th a t the la tte r has the suffix -w a, and it is the fro n t vowels again th a t make PC d ifferen t from P. Traditional Sinhala gram m ars claim th a t the causative construction is built by affix atio n . However, nothing is said about the form ation of involitive verbs. It is not explicitly m entioned in trad itio n al gram m ars w hether this affixation is inflectional or derivational, nor A

This division of the verbs into columns are not based on any th e o re tic al assumption. It was done to achieve two purposes: one, to clarify th e cu rren t verbal system , and two, to help the read er grasp th e Modern Sinhala verb variations by traditional descriptive methods.

N evertheless, it is evident from the cu rren t study th a t the stem variation in Modern Sinhala verbal system is a productive process. A few more examples are in 2.2.2.

2

.

2.2

1 2 3 4

A= bale - -na (-wa) "look/read"

P= b a le - -na (-wa) "see"*^ AC— b a le -wa -na (-wa) "CAUS. look" PC= b a la -we -ne (-wa) "CAUS. see"*

A= wapure - -na (-wa) "sow" P= wæpire - -n a (-wa) "sow"* AC= wapure -wa -n a (-wa) "CAUS. sow" PC= w apire -we -n a (-wa) "CAUS. sow"*

A= hade - -na (-wa) "make" P= hade - -n a (-wa) "make"* AC= hade -wa -na (-wa) "Caus. make" PC= hade -we -na (-wa) "Caus. make"*

A= hapa - -na (-wa) "bite/chew " P= hape - -na (-wa) "bite/chew "* AC= hapa -wa -n a (-wa) "CAUS. bite/chew

^ The asterisk (*) serves to warn th e read er th a t the glosses so marked are p articu larly poor sem antic equivalents. N evertheless, they are the nearest approxim ations th a t English affords in a single word.

PC= hapa -we -n a (-wa) "CAUS. b ite /c h e\

A= patura - -na (-wa) "spread" P= p a t i r e - -n a (-wa) "spread"*

AC p a tu ra -wa -n a (-wa) "CAUS. spread" PC= p a b ira -we -na (-wa) "CAUS. spread"*

A= wika - na (-wa) "bite" P= wike - -n a (-wa) "bite"*

AC= wika -wa -n a (-wa) "CAUS. bite" PC= wika -we -n a (-wa) "CAUS. bite"*

2.2.3 P ast Tense

The following are the past tense stem shapes of the verb /b a la / "to see"

1 2 3

A= b alu - -wa "looked/read" P= b a lu - -n a "saw"*

AC= b a le -wu -wa "CAUS. looked" PC= b a le -wu -n a "CAUS. saw"*

All th e four stem shapes in 2.2.3 have fro n t vowels plus the c h a ra c te ristic vowel u a t th e end of each stem ; The vowel fronting in p ast tense form s rem ains unexplained in traditional gram m ars. Vowel fronting of P and PC in th e ir present tense form s could be argued to have been phonologic ally conditioned by th e stem final front vowel e; but th ere is no such phonological evidence to support a claim of vowel fronting in past tense stem v arian ts such as in 2.2.3. It should also be

noted th a t in column 3 of 2.2.3, the -wa suffix associates with A and AC stem s whereas the na suffix follows P and PC stem s. All these variations require fu rth er investigation which is undertaken in section 2.6 a fte r describing the re st of Sinhala stem variation.

2.2.4 Future Tense

The future tense form s of the verb /b a la / are illu strated below.

1 2 3

A= b a la a - -w i "will look/read" P= bælee - -w i "will see"*

AC= b a la -waa -w i "will CAUS. look" PC= bæla -wee -w i "will CAUS. see"*

In the future tense form s, the suffix is added to stem finals which have a lengthened vowel. T raditional Sinhala gram m ar does not account for these form ations in the language.

2.3 Verb-Nouns

Each of the four verb groups- A, P, AC and PC- has a corresponding nominal form as illu strated for the stem variants of the verb root /b a la / in 2.3.1.

2.3.1

A) b a la -> b a l- iim P) b a le -> bal-um AC) balaw a -> balaw -iim PC) balaw e -> balaw-um

Besides these, there are two more nominal form s th a t cannot be re la te d to an exact stem form in term s of sem antics. They are b æ l - i l i and b æ la w - lli.

2.3.2

b«Bl-ili and bælaw-ili.

Traditional gram m ars describe these nominals as gerunds and have tre a te d their form ation as an inflectional process. As can be noticed from the exam ples above, th ere are three suffixes -iim ,-um and 4 ^ occuring w ith the stem . In 2.3.1, the -urn suffix is associated with P or PC stem s. This distributional difference for the most p art is identical with th e sem antic referen ces of the p articu lar stem . However, this is not always the case as there are instances where eith er of the verb-noun pairs - A, AC and P, PC - is used in the sam e context.

2.4 More allomorphs

The verb root /b a la / has still more stem shapes as illu strated in 2.4.1.

2.4.1

b a lo b a la "see or look continuously (volitive)" b a lib æ li "see continuously (involitive)" b a la la "having seen or looked (volitive)" baluu "seen or looked (volitive P ast P arti.)" b alapu "seen or looked a t (volitive P ast Parti.)" balanna "see or look (volit. Infinitive)"

balawanna "CAUS.see or look (volit.Infinitive)" bælenna "see (involit. Infinitive)"

So far, various stem variants of the verb root /b a la / were illu strated. It is not only for this particu lar verb root th a t th ere are so many d ifferen t stem shapes, but also for th e re st of th e verb roots in Modern Sinhala with only a few exceptions. C urrent Sinhala gram m ar books, most of which have followed the traditional gram m ars, do not p resen t a detailed explanation about the nature of all these stem shapes th a t exist in the verbal system of Modern Spoken Sinhala. One im portant point which is evident from overall discussions in traditional gram m ars is th a t Sinhala has undergone a g re a t many changes especially in phonology and morphology since the 13th century; however, none of these gram m ars seem s to adm it this reality . The discussion about verbs in Sidat-Sangaraw a is in fa c t a good discriptive analysis of th a t e ra th a t would also be helpful in understanding the nature of the cu rre n t Sinhala verbal system . The discription in Sidat-Sangarawa, about a vast array of affixes occuring with several d ifferen t verb-form s, speaks to the fa c t th a t even a t th a t tim e the inflectional morphology of Sinhala had already sta rte d a process of change which sim plified the declensional and conjugational paradigms to the c u rren t s ta te . As a resu lt of this, what one can find today is th a t a lim ited s e t of stem form s from the old inflectional system perform s some very basic gram m atical functions in the language. For exam ple, the infinitive form s, such as in 2.4.1, w ere in fa c t dative form s in the old system of verb-noun declensions (Gunasekara, 1962). The p ast tense form bælu was in fa c t past tense plural stem form w hereas b a l l was its singular co u n terp art. When the conjugational and declensional paradigm s s ta rte d to disappear from Sinhala, only a few rem ained taking over th e d ifferen t basic gram m atical functions in Modern Sinhala. It is, th e re fo re , im portant to pursue a descriptive study of th e verbal

system of Modern Sinhala to understand its organization in the lexicon. This study, as mentioned a t the beginning, a tte m p ts to investigate the concept of stem allomorphy in Modern Sinhala within the fram ework of OL and to study how fa r such a theory would be helpful in analyzing and accounting for the phonology and morphology of Modern Sinhala.

2.5 Compounds

It was observed in section 2.1 th a t each verb root has basically four stem shapes representing the four aspectual classes - A, P, AC, and PC. In section 2.3, it was mentioned th a t each of these stem shapes has a nominal form in addition to its finite verb form. It was also pointed out th a t according to the way the trad itio n al gram m ars tre a te d these stem variants, they w ere more likely to be the resu lt of an inflectional process.

Following the th e o re tic al background adopted in this study,^ it is argued to be the case in Modern Sinhala th a t stem variation is a process th a t takes place within the lexicon. This assumption can be fu rth er supported by the fa c t th a t th e re are compound words as illu strated in 2.5.1 (b) of which one mem ber is a verb-noun.

2.5.1 (a) (1) minis human (2) balu dog (3) pot book + gati nature + p æ ti sm all + saappu shop > mlnisgati human nature > b alu p æ ti puppy > potsaappu book shop (b) (1) b®l-um seeing (2) kiyow -iira reading (3) w®w-ili plantation + gal rock + saala hall > baluragal w atch tow er > k iy o w iim saala reading hall + karraaanta > wæwilikarmaanta industry plantation industry

The compounds in 2.5.1 (a) and (b) have a noun + noun com bination. In 2.5.1 (a), compounding has taken place betw een regular nouns whereas in (b) the leftm o st s iste r is a derived noun from a verb stem . The verbal roots of the three verb-nouns balum, klyaw ilm , and w aw ill are /b a la /, /k iy aw a/ and /w aw a/ respectively. If the verb stem s w ere not present in the lexicon, th e re would not have been reasonable grounds to account for th e process such as nominal

derivation and compounding. This kind of evidence in the w ord-form ation of Modern Sinhala supports the claim th a t the stem variation in Sinhala is a process to be tre a te d in the lexicon. In what follows, this will be discussed in detail within the fram ew ork of OL.

2.6 The Word Form ation Component according to OL

The Word Form ation Component described in OL is composed of three subcomponents - Perm anent Lexicon, Lexical Structure, and String Dependent Rule Component. The Perm anent Lexicon contains lexical en tries, morpholexical rules, and redundancy relations; the Lexical S tructure Component consists of binary branching unlabelled trees, and general node labelling conventions; and the String Dependent Rule Component contains productive morphological rules sensitive to the segm ental nature of the string on which they o p erate.

In organizing the Modern Sinhala verb morphology within the OL fram ew ork, one has to first consider the lexical item s th a t need to be included in the perm anent lexicon and then to account for the re s t of the word form ation processes w ith the help of inform ation stored in the tw o o th er subcomponents. 2.6,1 illu strates the four classes of stem variants fo r each of a sm all se t of illustrative verbs including their present, past and fu tu re tenses.

2.6.1

Infinitive P resent Past Future Nominal

1. "see"

A b a la -n n a balana-w a balu-w a balaa-M i b a l- iim P b a le -n n a balena-w a b a lu -n a b a lee -w i bal-um AC balaw a-nna balawana-wa balewu-wa balawaa-wi balaw -iim PC bælawe-nna balawena-wa balaw u-na balaw ee-w i balaw-um

2. "kill"

raara-nna marana-wa maru-wa maraa-wl m ar-lira

mære-nna marena-wa maru-na m aree-wi mar-um

marawa-nna marawana-wa marewu-wa marawaa-wi maraw-iim

raærawe-nna marawena-wa marawu-na marawee-wi maraw-um

3. "sow"

wapura-nna wapurana-wa wapiru-wa wapuraa-wl w a p ir-iim w ap lre-n n a w apirena-w a w aplru-na w apiree-w i wapir-um wapurawa-nna wapurawana-wa wapirewu-wa wapurawaa-wi w apiraw -iira w apiraw e-nna wapirawena-wa wapirawu-na wapirawee-wi wapiraw-um

4."strike/hit"

happanna happanawa happuwa happaawi happlim

happenna happenawa happuna happeewi happum

happawanna happawanawa happewuwa happawaawi happawiim

happawenna happawenawa happawuna happaweewl happawun

hapanne hæpenna hapawanna hæpawenna hapanawa hæpenawa hapawanawa hæpawenawa hspuwa hæpuna hæpewuwa hæpawuna hapaawi hæpeewi hapawaawi hæpaweewi hæpiim hæpum hæpawiim hæpæwura

(See appendix for more listings.)

In 2.6.1, it can be observed th a t the stem variations, eith er among the four verb groups or among present past and fu tu re tenses, follow a coherent morphological p a tte rn . For exam ple, th e stem variation among A, P, AP, and PC in the present tense is roughly as follows;

2.6.2

CqVCqV ~ C oV C oe ' C oV C oV w a ~ C oV C oV w e

Naturally, stem variations resulting from productive processes are excluded from the perm anent lexicon in the OL model. Only th e root from which each se t of variants derives is en tered here. One must, th e re fo re , make explicit the relationship betw een fro n t and back vowels fo r all stem s derived from the sam e root.

In the past tense all four stem s have fron t vowels, but in the present tense fro n t vowels in P and PC stem s oppose back vowels in the A and AC stem s. Front vowels in the present tense can be argued to be conditioned by the e vowel in th eir accompanying suffixes; however, nothing in the phonological environm ent accounts for fro n t vowels in stem s of the past tense.

Similarly, consistant phonological evidence is lacking to p red ict the occurrence of front vowels in th e stem s of verbal nouns. There are th re e nominal

suffixes -iîm , -îli, and -um . Observing the nominal stem form s, one could argue th a t the distribution of these suffixes is com plem entary in the sense th a t -iim and -ill suffixes are to be found with A and AC stem from s whereas the -um suffix always associates with the P and PC stem forms. According to this argum ent- one, back vowels in A and AC stem forms are fronted a fte r the affixation of -iim or -ili, both of which have high fro n t vowels; two, front vowels in -um nnminals have already been th ere before th e nominal suffixation. In the la tte r case, back vowels wer;- fronted by the accompaying e in the ^e and -we suffixation which precedes the nominal suffixation. The justification for such argum ent is, however, not based on phonological or morphological grounds, ra th e r on sem antic considerations.

J u st as in Germ an^, so too in Sinhala the um lauting rule of the earlier language no longer operates®, in spite of the fa c t th a t the results of its influence are so pervasive th a t a superficial look a t Sinhala suggests th a t umlaut is still a p redictab le and productive process.

The next a tte m p t to account for vowel fronting in Modern Sinhala is to consider morphological evidence. In the search of a morphological solution to the vowel fronting, the first step is to isolate the morphemes found in cu rren t Sinhala.

As s ta te d in 2.1.1, a verb divides into four groups named A ctive, Processive, A ctive C ausative, and Processive C ausative. The n atu re of these four divisions can be expressed with two sets of binary value fe a tu re s [+/-Agent] and [+/-Volitive] (2.6.3). The fe a tu re [+/-Agent] indicates w hether the action originated from the doer or not while [+/-Volitive] indicates w hether th a t action

® Lieber, 1981. ® G eiger, 1968, 19-23

does or does not involve the doer's will.

2.6.3

Active [+Agent], [+Volit]

Processive [+Agent],[-Volit] A ct. Causative [-Agent],[+Volit] P roc.C ausative [-Agent], [-Volit]

y

It is argued th a t each of these four stem p a tte rn s consists of a root plus a suffix in the underlying form , and the suffixes c arry the binary value fe a tu re s. See 2.6.4 and 2.6.5. 2.6.4 A /b a la / + / a / P /b a la / + / e / AC = /b a la / + /w a / PC = /b a la / + /w e /

Within the th e o re tic al definitions "The le a st elem ent in a p a rtia l ordering of lexical term inals will be called the ROOT" (Lieber 1981, pg.43). In this analysis the term ro o t is used to distinguish the underlying stem from its derived variants. Except for this deviation, the term s STEM and AFFIX will hold the sam e definitions as given within the OL fram ew ork:

1. STEM : a morpheme whose lexical en try does not subcategorize another morpheme.

2. AFFIX : a morpheme whose lexical entry specifies some so rt of lexical term in al to which it can a tta c h (Lieber 1981, pg.37).

2.6.5

+[a] = [+Agent], [+Volit] +[e] = [+Agent],[-Volit] +[wa] = [-Agent], [+VoIit] +[we] = [-Agent],[-Volit]

In this way, only th e respective root verbs of any stem variants will be listed in the perm anent lexicon to g eth er with the suffixes. Following the perm anent lexicon is the lexical stru c tu re subcomponent which consists of binary branching unlabelled tre e s and fe a tu re percolation conventions as shown in 2.6.6.

2

.

6.6

F eatu re P ercolation Conventions:

1. Convention 1: All featu res of a stem morpheme including category fe a tu re s p erco late to the first non-branching node dominating th at morpheme.

2. Convention 11: All featu res of an affix morpheme including category fe a tu re s p erco late to the first branching node dominating th a t morpheme. 3. Convention 111; If a branching node fails to obtain featu res by convention

11, fe a tu re s from the next low est labelled node are autom atically perco lated up to th e unlabelled branching node. (Lieber, 1981. 49-51.) When affix atio n tak es place in the lexical stru c tu re subcomponent, the resulting stem shapes receiv e th e ir fe a tu re specifications from the suffixes. In 2.6.7, th e derivation of the four stem shapes a re illu strated along w ith the fe a tu re percolation principles which account for the fe a tu re specifications of each of the four stem shapes.

2.6.7 a) bala]v [+agt] [+vol] b) bala^ [+agt] [-voi] [-agt] [+vol] c) bala] y d) [+voI]

Y [-agt]

[-vol] balaly we]Y ' [-agt] [-vol]

In 2.6.7, arrow I shows th a t the category fe a tu re of th e ro o t morpheme is perco lated up to the first non-branching node don m ating th a t morpheme by convention I. Arrow II dem onstrates convention II whereby all fe a tu re s of the affix morpheme including the category fe a tu re are p erco lated up to the first branching node dom inating th a t morpheme. The derived stem s in each of the four exam ples have received all fe a tu re values of th eir outerm ost morphemes including the category fe a tu re . Significantly, both morphemes in each derivation have the sam e categ o ry fe a tu re . In this case, one could always argue th a t the suffixes

should carry no category featu res, for they could be percolated up to the higher node from the node dominating the root morpheme by convention III. However, so far, th e re is no evidence for or against such an argum ent. Therefore, in this description o f Sinhala lexem es, the suffixes are marked for category featu res too.

If th e re were no fu rth er affixation, the derivations illu strated in 2.6.7 would end with the following surface form s.

2 .6 .8 A= b a l e P= b a l e A C= b a la w a PC= b a la w e

These stem shapes show th a t th ere is a difference in the quality and the quantity of vowels betw een the underlying form s and the derived surface forms in

2.6.8. M oreover, some vowels have been deleted. These m atters are discussed la te r (see 2.8.5).

2 .7 T e n s e F o rm a tio n and S te m B u ild in g

In sections 2.2.1, 2.2.4 and 2.2.5, it was explained th a t each of the four stem shapes had a corresponding form fo r presen t, past and future tenses. This section presents th e morphological and phonological nature of these form s th a t distinguish th ese tenses.

2.7.1 (a) P resent A= b a la na P= bæle na AC= balawa na PC= bælawe na (b) Past A= bæl uwa P= b e l una

AC= bælew uwa

PC= bælaw una (c) Future A= b a la a wi P= bælee wi AC= balawaa wi PC= bælawee wi

In (a),(b) and (e) of 2.7.1, th e first column represents the stem shapes th a t have already been presented in 2.6.8. As can be observed in column two, suffixes -n a , -uwa/-un@ and -w i are associated w ith (a), (b) and (c) respectively. In the present tense, only the P and PC stem shapes have fro nt vowels, but in th e past ten se, all four stem shapes have front vowels. The final vowel of each fu tu re tense stem is lengthened. As m entioned earlier, these vowel changes will be addressed in d etail la te r; this section considers how these p resen t, p ast and future tense forms are derived.

E arlier in this discussion (2.6), it was mentioned th a t the stem shapes consist of Verb Root + Suffix. This argum ent was raised to account for the four-way division of the verb. Now this argum ent can be extended to say th a t a verb form marked for its tense has the stru ctu ral analysis of Verb Root + Suffixi + Suffix?;. In this stru ctu re, the Suffix^ stands for those affixes mentioned in 2.6.4 whereas Suffixg represents any of th e tense marking affixes. It is, th erefo re, argued on the basis of this assumption th a t the suffixes -n a , -uwa/ -una and -wl are tense

Q

marking affixes and should be listed in the perm anent lexicon with relevant inform ation such as the category and class to which these morphemes belong including d iacritic fe a tu re s as in 2.7.2.

2 .7 .2

-n a -uwa/ -wa -wl

[+pres] [4-past] [4-futr]

In the second subcomponent of th e gram m ar, these affixes will be inserted into lexical tre e s following the inform ation available for each lexical elem ent and the fe a tu re percolation principles will percolate this inform ation up to the higher node. What follows illu stra te s how the lexical stru c tu re subcomponent of the gram m ar would gen erate th e present, p ast and future stem shapes.

Suffixg derivation will be called a Tense Building Process and the relevant affixes will, th erefo re, be named as Tense Building Affixes. S uffixj which is a Stem Building Process consists of Stem Building Affixes.

2.7.3 Present (a)

y

[+agt] [+vol [+pps] a]V ' [+agt] [+vol] bala _ne]V [+prs] (b) _[+agt] [-vol i+prs] y [+agt] 1 [-vol] bala. e]V ' [+agt] [-vol] na]V [+prs]

t l (0 bD 9 h o o > ' b o o

2.7.4 Past [+agt] [+vol [+pst] (a) [+agt] [+vol] alV / [+agt] balaJV uwa l+pst] I+vol] (b) e]V bala [+agt] [-vol] [+pst]

(c)

y

[-agt] '' [+vol [+pst] [-agt] [+vol] bala]V wa]V ' [-agt] ï+vol] uwa]V [+pst] (d) [-agt] [-vol [+pst] we]V [-agt] [-vol] balai u n a jv [+pst]

2-7.5 Future [+agt] [+vol [+ftr] (a) y [+agt] [+vol] wi]V a]V I [+agt] bala]V [+ftr] E [+vol] [+agt] [-vol î+ftr] (b) [+agt] [-vol] wi]V e]V / [+agt] bala]V [+ftr] [-vol]

(c) [-agt] [+vol [+ftr] [-agt] [+vol] wi]V bala]V [+ftr] [+vol] [-agt] [-vol [+ftr] (d) [-agt] [-vol] wi]V bala]V we]V ' [-agt] [+ftr] [-vol]

As in 2.6.7, so too here, the arrows I, II and III show how the fe a tu re s are p ercolated to higher nodes by the fe a tu re percolation conventions I, II and III

respectively. Any verb form derived by these diagrams has all the fe a tu re specifications necessary for its proper placem ent in the gram m ar.

The third subcomponent of OL is the string dependent rule component. Basically, this is where morphological rules are applied. The verbal derivations in Sinhala need furth er modifications in th eir segm ental fe a tu re s so as to be able to appear in co rre ct surface form s. As briefly pointed out earlier, most of these changes have to do with vowel quality or quantity of the verb root. In general verb roots have back vowels. The [+back] vowels in the verb ro o t become [-back] on two occasions; (a) in the present tense of P and PC verb forms and (b) in the past tense forms of all the four verb groups. With regard to the form er, the possibility of treatin g the vowel fronting as a phonological rule was considered above. The triggering facto r in this rule would be the front vowel e, which in itself is a morpheme. However, th ere are problems with such a phonological tre a tm e n t on two grounds. F irst, in the verbal system itself th ere is a case in the past tense forms where vowel fronting occurs with no phonological environm ent to account for it; this fronting has to be tre a te d as a morphological process. T herefore, within the verbal system itself, two kinds of vowel fronting rules would be required, one which is phonological and the o th e r morphological. Second, Modern Sinhala has no other known instances outside verb morphology where a phonological rule of vowel fronting results from the addition of For exam ple, the homonymous locative m arker ^e added to no minais as seen in 2.7.6 causes no vowel fronting.

2.7.6 /h a ta ra / "four" /g a s/ "tree" /m e e sa / "table" h a ta r - e "at four" gas-e

"in the tree" mees-e "on the table"

In the examples in 2.7.6, the second column represents the surface nominal form s with the suffix added. However, the back vowels rem ain the same despite the front vowel In the last example, the vowels are underlyingly [-back]. Considering the evidence thus available in Modern Sinhala in general and within the verb morphology in particular, the preferable solution seem s to be a morphological one. In th a t case, the OL model requires th a t such a morphological rule operation be perform ed in the string dependent rule subcomponent of the gram m ar.

In order to arrive a t a morphological solution fo r the vowel fronting in the Modern Sinhala verbal system , the fe atu re [+umlaut] will be introduced. The affix es discussed so fa r in 2.6.5 and in 2.7.2 are rep eated in 2.7.7.

2.7.7

-a = [+ag] [+vo]

-e = [+ag] [-vo]

-w a = [-ag] [+vo]

-we = [-ag] [-vo]

-n a = [+prs]

-uw a/-u n a = [+pst] [+um]

-wi = [+fut]

The fe atu re [+um] invokes the following morphological rule:

2.7.8

[+syl] —> [-bk] / — [+um]

2 .7 .9 ( a l ) e]v na]v balajv [+agt] [-vol] [+um] [+prs]

V

/ (a2) e ]v balæ ]v n a ] v [+agt] [-vol] [+um] [+prs]

(bl) we]v [-agt] [-vol] bala]v [+ura]

\y

(b2) n s]v b a læ ]v [+prs] [-vol] i+um]

(cl) a]v b a la jv uwa]v [+agt] [+vol] I+pst] [+um] (c2) æ]v uwa]v b a la ]v [+agt] [+vol] [+pst] ï+um]

(d l) una]v e]v balajv [+agt] [-vol] i+um] [+pst] ium]

V

(d2) e ]v un@]v b a ls lv [+agt] i-vol] i+um] [+pst] i+um]

(el) wa]v [-agt] uwa]v [+pst] bala^ [+vol] [+um] (e2) b a la lv uwalv [+pst] [+vol] [+um]

(fl) b a la jv we]v [-agt] [-vol] una]v [+pst] [+um] [+um] b a ls jv wejv [-agt] [-vol] [+um] una]v [+pst] [+um]

In the examples of 2.7.9, vowel fronting is triggered by the fe atu re [+um] and the operatic e rule in it is a string dependent morphological rule. T herefore, string adjacency is required for this rule to operate. Thus, in examples (a) and (b) this requirem ent is satisfied, but in examples (c) and (e) the um laut triggering featu re is on the rightm ost morpheme. It is assumed in such cases th a t the segm ent in the adjacent string is um lauted, and this is then spread across the other syllabic segm ents from right to le ft. However, umlaut spreading is a phonological fa c to r which is specified la te r.

There is another way of looking a t this vowel fronting. It could be assumed th a t the featu re [+um] is percolated up to the higher node and any vocalic segm ents in a node th a t are dom inated by the higher node with the [+um] featu re will be um lauted. However, the validity of such a stru c tu re dependent rule has y et to be investigated in depth.

The form er assumption is adopted here on the grounds th a t even in an example like (a) of 2.7.9 where the string adjacency is satisfied , it is only the rightm ost vocalic segm ent th a t is um lauted first and the um lauting of the re st of the vocalic segm ents in the root is considered to be a result of an um laut spreading rule.

A change in vowel quantity is found in fu tu re tense form s. In balaa-w i, for exam ple, the stem final vowel is lengthened. This process takes place regularly in th e fu tu re tense form s of all four groups. This lengthening process is also assumed to resu lt from a morphological rule because th ere are no reasonably accountable grounds for a phonological solution. In Modern Sinhala, the only instance of phonologically conditioned vowel lengthening takes place betw een two [+low] vowels when th ere is a vowel coalescence as illu strated in 2.7.10.

2 .7 .1 0

/ a t a / + /arin aw a/ — > a ta a rin a w a

hand open

"give u p /let go"

In this example the coalescence takes place betw een the two vowels [a] - the final vowel of / a t a / and [a] the initial vowel of /arin aw a/.

Vowel lengthening in the fu tu re tense form s is, th erefo re, assumed to be triggered by the fe a tu re [+long] in the affix -w i. Thus, the affix bears th e featu re [+long] in addition to [+future], and th e morphological rule th a t lengthens the stem final vowel looks like the following;

2 .7 .1 1

[+syl] --> [+long]/— [+long]

This rule is also string dependent taking place in the string dependent rule subcomponent of the gram m ar. 2.7.12 below illu strates th e operation of this rule in deriving future tense forms.

2.7.12 (al) Wi]V a]V [+agt] bala [+ftr] [-vol] [+long] (a2)

V

w l]V aa]V bala]V [+agt] [-vol] [+ftr] [+long]

(bl) e]V wi]V bala]V (b2) [+agt] [-vol] [+agt] [-vol] [+ftr] [+long]

balajV ee]V Hi]V

[ + f t r ] [+long]

(cl) bala’ wi]V [+ftr] [+vol] [+long] waa]V [-agt] bala' wi]V [+ftr] [+vol] [+long]

( d l ) wi]V we]V [-agt] bala^ [+ftr] [-vol] [+long] (d2) wijV wee]V [-agt] bala [+ftr] [-vol] [+long]

Following (a) of 2.7.12, one could argue, as noted above, th a t vowel length results from the coalescence of two identical vowels which would mean th a t vowel lengthening is a phonological process. However, (b), (c) and (d) of 2.7.12 illu stra te cases of vowel lengthening where identical vowel sequences are lacking. (This

situation is rem iniscent of umlauting part which could have been accounted for by a phonological rule but required a morphological rule to cover all cases.)

2 .8 S u b c a te g a r iz a tio n R e s tr ic tio n s

Thus, stem variations in Sinhala verbal paradigms can be accounted for, if the historical inflectional endings are tre a te d as lexiealized affixes. These are brought to g eth er in 2.8.1, 2

.

8.1 I II - a -na - e -uw a/-una -wa -w i -we

The affixes in column 11 always follow those in 1 indicating a hierarchical relationship betw een the two sets which the gram m ar must recognize. In the OL fram ew ork, this is done by means of subcategorization re stric tio n s such th a t affixes of column I are added to verb root followed by those in column II. The output of both affixation processes is still a verb,

Stem s c reated by adding column I suffixes are said to be derivations while derivations are those form ed by th e fu rth er addition of column II suffixes, the tense morphemes. It is now possible to draw up the following subcategorization re stric tio n s. Affixes in column I a re subcategorized as, ]V— ]V^ and those in column II as, ]V^— ]V^, indicating th a t the column II affixes are a tta ch e d to and th e resu lt is V^.

Two suffixes -uws and -u n a, have the fe a tu re [+past]. For them a fu rth er subeategorization is necessary such th a t -uwa is added only to [+volit] stem s and -u n s only to [-volit] stem s.

In summary, the following inform ation about affixes is sto red in the perm anent lexicon of the gram m ar.

2

.

8.2

(a) S tem B u ild in g A f f ix e s

Affix D iacritic Features

-a [+Agt],[+Vol] -e [+Agt],(-Vol],[+Um] -wa [-Agent], [+Volit] -we [-Agt],[-Vol],[+Um] C ategory V V V V Subcategorization

]y— ]y^

Jv— ]y— ]y— ]y^ (b) T e n se B uild in g A f f ix e s

Affix D iacritic Features

-no [+Pres] -uwe [+Past],[+Um] -una [+Past],[+Um] -wi [+Futr] C ategory V V V V Subcategorization ]y^— ]y^ ]y^— ]y^ ] y l ] y 2 ] y l ] y 2

Verb stem s are also stored in th e lexicon w ith the inform ation as given in the following example.

2 .8 .3

C ategory Class VERB

(bala) morpholexical rules (none) phonological represen tatio n

sem antic representation: 'see/read' insertion f r a m e :...

In the second subcomponent of the gram m ar, derivations take place according to the illustrations shown in 2.7.3 - 2.7.5. Morphological rule applications take place in the third subcomponent. The present, past and future verb forms discussed so fa r will have the following output a t the end of the third subcomponent of the gram m ar.

2 .8 .4

(a) Present

bala] a] na] [balana]

bale] e] na] [belena]

bala] wa] na] [balawana]

bale] we] na] [belawena]

(b) Past

bala] ®] uwa] [beluwa]

bale] e] una] [beluna]

bala] we] uwa] [belawuwa] bale] we] una] [belawuna]

(c) ]Future

bala] aa] wi] [balaawi]

bala] ee] wi] [beleewi]

bala] waa] wi] [balawaawi] bale] wee] wi] [beleweewi]

However, to g et the c o rre ct surface forms as illu strated on the right hand column of 2.8.4, a few more changes are necessary. These are phonological changes listed in 2.8.5. 2 .8 .5 1. Umlaut Spreading 2. Vowel Loss 3. Resyllabification 4. Vowel Reduction

No a tte m p t is here made to describe these four phonological rules. However, see C hapter V for an explanation about phonological rule applications in Modern Sinhala.

2 .9 N o m in a l D e r iv a tio n s

In 2.3.1 and 2.3.2, , re p e ate d in 2.9.1, it was m entioned th a t th ere a re nominal derivations th a t are re la ted to the fo u r verb groups.

2.9.1

A) bala -> bal-lim

P) bale -> bæl-um

AC) balawa -> balaw-iim PC) balawe -> _bælaw-um

-> bæl-ili -> b&law-lll.

Here it is argued th a t no minais are form ed by adding the suffixes -iim , -um and 41i to the derived stem s described in 2.6.8. These th ree suffixes are stored in the perm anent lexicon along with category featu res, any d iacritic features necessary and subcategorization restrictio n s. These suffixes belong to the nominal category and the the subcategorization restrictio n s must say th a t the environm ent for the nominal affixation follows V^-verb stem s. This inform ation is presented in summary in 2.9.2.

2.9.2

(c) N oun F o rm in g A f f ix e s

A ffix D iacritic Features

-iim [+Um] -um I+Um] -ili [+Um] Category N N N Subcategorization ] y ^— ]n ]v ^— In Iv ^ — In

The o th er significant issue in this nominal derivation is the fronting of back vo xels. Is this vowel change phonological or morphological? The fro n t vowel i in the two suffixes -iim and suggests th a t the vowel change is phonological.

However, with -um th ere is no such phonologically conditioned environm ent for the vowel change. Once again th ere are two conditions one of which supports a phonological solution fo r the vowel fronting and th e other which does not. In order to describe this fronting th a t is triggered by all three nominal suffixes by a single rule, this study suggests a morphological process. In doing so, it is assumed th a t the suffixes bear the diacritic featu re [+umlaut] which causes vowel fronting in the verb’s stem vowels. This morphological rule is sim ilar to the one described in 2.7.8 which is rep eated in 2.9.3.

2.9.3

[+syl] —> [-bk] / — [+um]

The derivational processes fo r no minais through the second and the third subcomponents of the gram m ar are shown in 2.9.4.

2.9.4 Y [+agt] / [+vol] !im]N [+um] bala] æ] iim] I+vol]

[-agt] [+vol] [-agt] [+vol] wa]v ' [-agt] iim]N [+um] bala! [+vol]

In 2.9.4 the arrows I, II and III re fe r to the featu re percolation conventions I, II and III respectively. The right directional arrow shows the output a fte r the morphological rule applications in the third subcom ponent. The featu re [+um] is a strin g dependent morphological rule which triggers th e back vowel to be fronted. (This procedure recalls th a t discussed regarding vowel fronting in p ast tense forms in 2.7.9.)

2 .1 0 E x c e p tio n s

Consider the verbal paradigm s of /a d i/ "pull" and /w a"d i/ "genuflect" in 2

.

10

.

1

:

2

.

10.1 (a) A P AC PC

Present Past Future

(b)

A P AC PC a d i-n a ad e-n a adde-na adawa-na addawa-na adawe-na addawe-na w a"dl-na wa’^de-na wende-na wa"dawa-na wandawa-na wa” dawe-na wandawe-na add-a ad-una add-una addew-uwa wand-a wa^d-una wand-una wandew-uwa wandaw-una a d i i - w i i adee-w i addee-w i addawaa-wi addaw-una addawee-wi H a"d li-H l wa^dee-wi wandee-wi wandawaa-wi wandawee-wi

There is a lim ited number of verb roots in the Modern Sinhala lexicon th a t behave like those in 2.10.1. The final root vowel of this type of verb usually is a high fro n t vowel, i. The past tense form of the A verb has only a (instead of -uwa) a t the end. The P verb has doublets in present, past and fu tu re form s. AC and PC verb form s have doublets in th e ir p resen t tense form s. The doublets are

form ed by copying the consonant th a t im m ediately precedes the high front vowel, i.

The crucial point here is how to account for the occurrence of these allomorphs of which the only difference is th a t one of them has a consonant clu ster as a resu lt of copying th e final consonant of the root. This consonant copying process is neither phonologically nor morphologically predictable. One way of addressing this situation is to list the variations in the perm anent lexicon and have th e ir relations expressed by morpholexical rules. The two verb roots /a d i/ and /w a^d i/ have the following CV shapes.

2

.

10.2

adi = VCV

wa^di = CVCV (initial C rep resen ts the semivowel w)

In 2.10,1, it is noted th a t stem form s have the shapes (C)VCV + Suffix or (C)VCCV + Suffix (i.e. ædena ' æddena). A morpholexical rule like the one in 2.10.3 can be drawn to re la te these two form s if they were to be listed in the perm anent lexicon.

2 .1 0 .3

(Ci)VC2V

- (Ci)VC2C2a

However, th e re is still an im portant point to be addressed. In the paradigm of 2.10.1, the C2 is a prenasalized stop. It is always the case in such situations th a t when copying a consonant segm ent which is a prenasalized stop the nasal sound

and the stop sound appear as two sep arate segm ents. According to the morpholexical rule in 2.10.3, it will be implied th a t the copied Cg is again a prenasalized stop. Therefore in addition to 2.10.3, it is necessary to have another morpholexical rule th a t co rre ctly predicts form s such as those in (b) of 2.10.1. The morpholexical rule in 2.10.4 captures the relatio n of form s th a t have a prenasalized stop consonant in Cg position.

2 .1 0 .4

(CjV^CgV - (Ci)VN2Cga

The 2.10.4 rule s ta te s th a t a "Cg {which is prenasalized consonant) has a re la ted form with a com plete nasal segm ent preceding the Cg which is now a non-prenasalized consonant. The capital N stands for any nasal sound which is always homorganic with the following Cg stop sound.

Given these two additional morpholexical rules, the c o rre ct verbal variants are derived by the sam e procedure as described for the four verb stem groups and the present, past and fu tu re tense form ations.

2 .1 1 C o n clu sio n

It is clear from this investigation th a t stem variation in Modern Sinhala verbal system is the result of a stem building process th a t each verb undergoes. This stem building process is followed e ith e r by a tense building process or by a noun form ation process. Compounds are form ed from verb-nouns th a t derived by the la tte r process.

During th e discussion, however, th e need of a com ponent in the gram m ar for relev an t phonological rule applications was pointed o u t. Vowel reduction, which is

fed by resyllabification rule and special verb form s like those in section 2.10 support the claim of underlying syllable stru ctu re as argued by Clem ent and Keyser (1983). In addition, the fa c t th a t stem building suffixes always precede the re st of the affixation process seem s to support th e claim of having level ordering in morphology as discussed in Lexical Phonology and Morphology (Klparsky, 1982b. ; Mohanan,1986). Except fo r these few points, Lieber's OL theory provides the essential mechanism for th e organization of the lexicon, and it addresses the basic nature of the Modern Sinhala verbal system which has lexiealized a g reat deal of inflectional processes which need to be addressed synchronically as morphological processes.

Two points need to be clarified here. F irst, Sinhala speakers alte rn ate verb-nouns A with ? on the one hand and AC with PC on the other. Thus one may use b a liim a ltern atin g w ith bælum and baelawiim altern atin g with belawum. Second, th e argum ent for the existence of PC verb from s was based on sentences and clauses like the following:

1. sinduwa ehena k o ta mate nikaitiraa natawena-wa / nætena-wa. so n g -hear-w hen-m e-w ithout e f f o r t - d a n c e

In me it s ta rts to dance when I h ear the song.

2. (a) gahe Idena k e s e lg e d i

on th e tr e e - r i p e n - p l a n t a i n s P lantains th a t ripen on th e tre e .

(b) dura g a h a la idawena k e s e lg e d i s ra o k e d -re p e n e d -p la n ta in s

(c) dura g a h ala id en a k e se lg e d i sra o k e d -rip e n -p la n ta in s

Plantains th a t ripen because of smoking

3. ra ja y e n hadena / hædawena paalaraa by th e governm ent-being b u i l t - b r i d g e The bridge th a t is being built by the governm ent

The examples show th a t PC verb form s exist in Sinhala. Frequently PC verbs alte rn ate with P verb forms. However, the sem antic differences are evident in examples like 2(b) and (e).

C hapter

ni

NOUN MORPHOLOGY

3.1 Prelim inaries

The Sidat Sangarawa and derivative gram m ars describe Sinhala nouns as having nine eases, Nominative, A ccusative, Instrum ental, Auxiliary®, Dative, Ablative, G enitive, Locative and Vocative. These are illustrated in 3.1.1 with minis 'man'.

3.1.1

Singular Plural

1. Nom. minis-a miniss-u

2. Ace. minis-a minis-un

3. Instr. minis-a-wisin minis-un-wisin

4. Aux. m inis-a-karanakote m in is-u n -k ara n a k o ta

5. D at. m in is-a-ta m in is -u n -ta

6. A blat. m inis-a-gen minis-un-gen

7. Gen. minis-a-ge minis-un-ge

8. Loc. m inis-a-kerehi m inis-un-kerehi

9. Voc. minis-a minis-uni

This is sim ilar to Instrum ental case in function. 59

-Traditional gram m ars claim th a t these endings are still inflectional case m arkers in the modern language- However, it is dem onstrated in this chapter th a t these suffixes have evolved into a p art of the derivational system . Their function in contem porary Sinhala is described first (3.2), followed by consideration of the n atu re of underlying representation (3.3). These two sections in turn provide the background necessary to describe stem final consonant gem ination in nouns. (An example of such gem ination is the double ss of the nom inative plural of minis in 3.1.1.)

3 .2 N om in a l C la s s e s

It is assumed here th a t any investigation of the nominal system in Modern Sinhala must first co n cen trate on the form ation of the singular and plural form s of the noun. It can be seen in the paradigm of 3.1.1 th a t all oblique cases are built on the nominative suffix in the singular and on the accusative suffix in the plural. In fa c t, in modern, spoken Sinhala, the nom inative and accusative are not marked and the suffixes %u and ^un indicate number ra th e r than case. For the entire inventory of nouns one of two suffixes and ^ designates the singular, while e ith e r or ^o marks the plural.

3.2.1

Singular Plural

-a,-9 -u,-o

Of the singular suffixes, is used with masculine nouns and with feminine and n euter nouns. The n eu ter nouns have no o vert marking in their plural form s

It should be noted th a t the g ram m atical gender in Modern Sinhala is identical with the natural gender. Nouns th a t stand for inanim ate objects are marked

while anim ate (=masculine and feminine) nouns take or ^ or no suffix a t ail. In this last case, the suffix is said to be segm entally null. Some nouns in the anim ate category may have e ith e r the or the ^o suffix to mark the plural. Examples below illu strate the distribution of these singular and plural suffixes.

3 .2 .2 (a) a o nari-y-a "fox" kaw i-y-a "poet" nari-y-o "foxes" kawi-y-o "poets" I I . maalu-w-a "fish" panu-w-a "worm" maalu-w-o "fish" panu-w-o

as n eu ter and the nouns of anim ate objects are divided into masculine and fem inine gender on the sam e criterio n as it would apply in biology to discrim inate male from fem ale. Apparently, masculine and feminine nouns are subsets of an anim ate noun class.

(b) (c) a o ib-b-a ib-b-o "tortoise" "tortoises" b al-l-a bal-l-o "dog" "dogs" kol-l-a kol-l-o "lad" "lads" a u gon-a gon-n-u "ox" "oxen" p ut-a p u t-t-u "son" "sons" kok-a kok-k~u

(d) (e) 9 0 gas-9 "tree" putU-W-9 "chair" adi-y-9 "foot" .11 9 0 kikili-y-9 "hen" bael9li-y-9 "tabby cat" gas "trees" putu "chairs" adi "feet" kikili-y-o "hens" b æ l9 li-y -o "tabby cats" (More examples are to be found in Appendix B.)

From these exam ples it is seen th a t nominal stem s ending in a vowel require glides before the number suffixes. This glide form ation is totally predictable in phonological term s. Stem s with a final high front vowel, i, or a mid c en tral vowel, 3, take y before the suffix while those with a final high back vowel require w.

Number marking suffixation of consonant final stem s is more involved. Those in class (d), the inanim ates, tak e -9 in the singular but remain unmarked in the plural. This syllabifies with the final consonant of the stem .