Update: 2008-08-05 09:46 AM -0400

TIL

The Grammaticalization of Nominalizers in Burmese

Simpson-nominalize.htm

by Andrew Simpson, Professor of Linguistics & East Asian Languages and Cultures, University of Southern California,   http://victoria.linguistlist.org/~lapolla/nw/Simpson.doc    andrew.simpson@usc.edu [in 17 pages of MS-Word] Also available in PDF.

Downloaded on 080622 and edited by U Kyaw Tun (UKT), M.S. (I.P.S.T., U.S.A.), and staff of TIL (Tun Institute of Learning, http://www.tuninst.net ), from various sources. Prepared for students of TIL Computing and Language Center, Yangon, Myanmar. Not for sale.

indx-Gloss | Top
Contents of this page

Contents of this page
01. Introduction
02. Nominalizers and their common and divergent properties: a brief overview.
02.01. What are nominalizers and what sources do they develop from?
02.02. Is nominalization a lexical process or syntactic?
02.03. Nominalizers and verbalizers
02.04. Nominalization and/vs. compounding
03. Nominalization in Burmese
03.01. Sentential nominalizers in Literary Burmese: the elements thii and mii {th} and {m}
03.02. Sentential nominalizers in Colloquial Burmese te {t} and me {t}
4.0 Summary of conclusions and consequences for further work.

Simpson's footnotes
UKT's footnotes
Simpson's references

This paper is concerned with the grammaticalization of clausal nominalizers in two different but closely-related forms of Burmese, Colloquial Burmese and Literary Burmese. A contrastive overview of the morphosyntactic properties of the nominalizers thii {th} and mii {m} of Literary Burmese and their Colloquial Burmese counterparts te {t} and me {t}, together with the application of a number of tests for the identification of nominalized constructions, reveal that grammaticalization is more advanced in the colloquial language than in the literary variety: te and me have lost their nominal specifications and been reanalysed as grammatical elements of a different categorial type, instantiating verb-related mood and realisirrealis distinctions. The comparison of the system of nominalization in the two complementary varieties of Burmese allows for insights into the evolution, spread and reinterpretation of nominalization structures within a language. --
http://www.benjamins.nl/cgi-bin/t_articles.cgi?bookid=TSL%2076&artid=186116180 080627

Contents of this page

01. Introduction

The cross-linguistic phenomenon of nominalization is an area of morpho-syntax that has attracted growing interest in both formal and functional linguistics in recent years, as data from an increasing number of languages becomes available and adds important insights into the grammaticalization paths and syntactic structure underlying the emergence of nominalizing elements.  The present paper sets out to probe this area of morpho-syntactic development further with a study focused on Burmese, a language which is particularly rich in nominalization structures and where a highly informative picture of the results of the grammaticalization of nominalizers can be found through a comparison of two different though closely-related forms of the language: Colloquial Burmese and Literary Burmese. A careful examination of synchronic patterns in Colloquial and Literary Burmese provides evidence of the source and complex structure of clausal nominalizers in the language, and indicates that in certain cases earlier nominalizers have undergone re-grammaticalization as functional morphemes of a different formal type, no longer instantiating nominal categories but being reanalyzed as modal and complementizer-type elements.  Quite generally, the patterning found in Burmese nominalization raises questions about the assumed synchronic status of nominalizers in other languages and how the occurrence of cross-categorial reanalysis of (already) grammaticalized morphemes can effectively be detected. The structure of the paper is as follows. Section 2 first considers how nominalizers may be identified in a language and what morpho-syntactic properties and sources nominalizers seem to exhibit in general. Section 3 then turns to Burmese and attempts to establish how the current, productive clausal nominalizers in Literary Burmese have undergone grammaticalization producing complex, fused forms occurring in a range of syntactic environments. This leads on to a comparison with similar patterns in Colloquial Burmese, where the conclusion is reached that nominalization in the colloquial language has progressed a significant stage further, and resulted in the reanalysis of erstwhile nominalizers as grammatical elements of a rather different clausal type. Section 4 then closes the paper with a brief summary of its findings and questions that are raised by the patterning observed. 

Contents of this page

02. Nominalizers and their common and divergent properties: a brief overview.

02.01. What are nominalizers and what sources do they develop from?

In the broad, developing literature concerned with nominalization, it is not uncommon to find the term nominalizer being made use of rather loosely, to refer to elements whose status as nominalizers has not been particularly clearly established. As we begin this investigation it is therefore useful to attempt to clarify what kinds of morphemes should be genuinely labeled as nominalizers, and how one might attempt to identify a particular grammatical element as a nominalizer rather than as some other kind of syntactic category. 

The term nominalizer is a purely functional label which is appropriately used to refer to all those morphemes/words which have the specific function of creating a nominal morpho-syntactic form as the result of their combination with other kinds of non-nominal input, as indicated in (1):

(1) A nominalizer: a morpheme whose primary function is to convert a non-nominal input form into a nominal category.

Nominal categories, and hence the presence of functional elements which may be nominalizers, can in turn be identified in two basic ways: (a) through the occurrence of noun-like/nominal morphological patterns, and/or (b) via syntactic privileges otherwise commonly associated with nouns and their syntactic projections.

Concerning the first type of morphological evidence for the presence of a nominal (rather than a verbal, adjectival or other) category, the potential occurrence of a range of common nominal attachments/inflections on a syntactic phrase/XP or combination of certain other syntactic categories with the phrase can be taken as indication that a phrase is nominal, patterning in a morphological and combinatorial way like other simple nouns and their expansions into noun-rooted phrases. A number of such elements which frequently occur with specifically noun-like phrases are listed in (2):

(2) Morphological indications that a syntactic constituent is nominal:
i.  the occurrence of case inflections on a constituent
ii. possible pluralization/plural-marking of the constituent
iii. possible enumeration of the constituent (combination of the constituent with numerals)
iv. the potential occurrence of demonstratives and adjectives with the constituent, rather than complementizers and adverbs
v. use of case-marking strategies associated specifically with nouns in the  marking of arguments of the noun (e.g. use of possessive/Genitive case to mark the nouns arguments rather than Nominative/Accusative case)

UKT: In this paper, Simpson prefers the use of the term "genitive" over "possessive". However, the term "possessive" or {peing-hseing-mhu.} seems to make more sense to an ethnic Burmese like me. Similarly, of the two terms Nominative/Accusative, "accusative" literally implying an "accusation" {swup-sw:hkak} gives a wrong impression.

Syntactically, a complex constituent may be identified as a nominal phrase if it shows the distribution of other simplex phrases that are clearly nominal, for example, the ability to occur in subject position, or the ability to be co-ordinated with other clearly nominal categories. If other, non-nominal categories such as verbal/adjectival phrases are regularly excluded from such positions, but a verbal/adjectival phrase in combination with some additional morpheme is found to allow for occurrence in subject position/co-ordination with other noun-phrases, this may be taken as reasonable evidence for the nominalized status of the complex constituent, and for the nominalizing function of the morpheme combined with the verb/adjective and their dependents.

Because it is the outward morpho-syntactic behaviour of a constituent that identifies it as nominal, and because there are various ways in which nominal morpho-syntax can be exhibited (case-marking, pluralization, combination with adjectives etc), nominalizers may in fact originate from a range of different lexical sources. Consider a formal syntactic representation of the internal structure of a referential nominal phrase, a DP, in a head-initial language with classifiers, as given in (3).

If it is assumed that the occurrence of any lower syntactic head position such as N, Cl, or Num will necessarily result in the projection of other, higher portions of the structure in (3), and that different parts of the structure in (3) may identify a constituent as nominal in different ways (the occurrence of numerals in Num, or demonstratives in D), it is possible to see that the use of any of the head constituents (D, Num, Cl or N) in (3) might in principle be used to build up a phrase with certain overt nominal properties when combined with a non-nominal complement, nominalizing the latter. For example, were an element of syntactic/lexical category N to be combined with a verbal constituent (if complex, then a verb-phrase/VP), the result would be a structure such as (4), in which a full array of nominal properties might be expected to characterize the new nominalization, given the presence of the full set of syntactic heads and projections above the N position. In the representation in (4), NZR is intended to indicate a morpheme that has the function of a nominalizer [UKT: NorminaliZeR ?], and the original syntactic category of a noun (i.e. a nominalizer derived from a noun, perhaps via bleaching of the lexical content of the noun):

Alternatively, however, a verbal (or other non-nominal) constituent might allow for conversion into a category with certain nominal properties if combined with a nominalizer sourced from one of the higher head categories in (3), such as an element of type D. Such a constituent would be expected to allow for the syntactic distribution of other nominal expressions (e.g. unrestricted occurrence in subject position), but might not allow for the full array of properties commonly associated with NPs and their expansions into DPs due to the lack of lower portions of the nominal structure in (3), notably the Num, Cl and N heads and their projections.  Such a possibility is schematized in (5) (where the nominalized constituent is indicated as being possibly a verb-phrase/VP, a tense-phrase/TP or a fully-clausal complementizer phrase/CP):

Two other intermediate possibilities are also anticipated to be available and exist, first, the use of a classifier as a nominalizing element combining with a non-nominal complement, as in (6), and second, the use of a numeral as a nominalizer, as in (7).  In both instances the structures produced are expected to have some, but not necessarily all typical properties of nominal projections. For example, if adjectival modification is assumed to occur via the adjunction of an adjectival phrase/AdjP to a noun-phrase/NP, the absence of an NP in structures (6) and (7) may be expected to correspond to a lack of adjectival modification with such nominalizations.

 

Interestingly, it would seem that all the possible ways that a nominalizer might theoretically be instantiated, as outlined above, do indeed seem to occur in different languages, and nominalizers grammaticalize from a variety of sources.  Nominalizers ultimately derived from nouns are most probably quite common in occurrence (e.g. Korean kes as discussed in Simpson and Wu 2001), but it is also possible to identify classifier, numeral and demonstrative/determiner sources of other nominalizers too. Burmese, for example, makes use of the numeral one (ta-) {tic}/ {ta.} (UKT-fn01) in a range of nominalizations, Thai and Bengali show evidence of classifier use in a nominalizing function with certain clausal constituents, and Chinese, Japanese and Lakhota can be argued to have derived nominalizers from earlier elements of type D (see Simpson 2001, 2003a/b, and Simpson and Wu 2001 for much discussion).

Contents of this page

02.02. Is nominalization a lexical process or syntactic?

In addition to variation in the source category of nominalizing elements, a second important parameter of variation in the (synchronic) realization of nominalizers is whether the attachment of such elements occurs as a lexical process, or is effected as part of a syntactic derivation. 

Certain nominalizers seem to be very clearly phrasal attachments, combined with a constituent that is a full syntactic phrase, as in the case of sentential/clausal nominalizers found in a significant number of languages, where a full clause is converted into a nominal argument of some other predicate by the use of an appropriate nominalizer (e.g. Japanese no, Korean kes, Simpson and Wu 2001).  If sentential and other phrasal nominalizers convert a full syntactic constituent into a nominal output, such nominalizers need to be considered syntactically independent words, combining with a complement during the syntactic derivation of a sentence, much in the way that determiners, complementizers and other similar functional element are understood to be discrete grammatical words rather than parts of other words (i.e. bound morphemes).

Other nominalizers, however, may appear to be affixes attached to word-level  elements such as verbs, adjectives etc, rather than to verb-phrases, adjective-phrases etc.  Such nominalization may have many more of the unpredictable properties of lexical processes (perhaps being restricted and unproductive and giving rise to allomorphic variation), and so be assumed to be purely morphological attachments, combining with non-nominal roots/bases during the pre-syntactic creation of words. (Simpson-fn01)  An example of such morphological attachment would be nominalizations produced with English ant, as in servant, defendant etc.

Contents of this page

02.03. Nominalizers and verbalizers  

Given the very widespread nature of nominalizers in the worlds languages, being even present in languages not considered to have morphological attachments such as Vietnamese, it is natural to wonder whether there might be equivalents to nominalizers functioning in the verbal domain, what might be called verbalizers, and, if such elements exist, what they might in turn indicate about the nature of nominalizers.  In other words, if nominalizers combine with non-nominal input to create words and constituents which can be utilized as noun(-phrase)-like arguments, are there also functional elements which combine with non-verbal input to create constituents which can then be embedded as verbal units?  And if not, why do such elements not exist?

Although there has been little discussion of such potential counterparts to nominalizers in the literature, certain reflection suggests that there are indeed verbal elements which correspond to nominalizers in their function of converting non-verbal input to a syntactically utilizable verbal form, and that such verbalizers occur both as syntactically independent words and as purely morphological attachments in a way very similar to the occurrence of nominalizers.  An example of the former lexical/morphological verbalization would be the application of affixes such as English -ize to adjectival input to form new verbs, e.g. grammatical grammaticalize.  Examples of syntactic verbalizers can be given in two quite common types.  A first, cross-linguistically well attested type is the class of light verbs such as (prototypically) make or do, which are used to combine with nominal input to produce verbal forms, as, for example, in Hindi kaam karna work do = to work, or Japanese benkyoo suru studying do = to study. A second potential candidate for consideration as a syntactic verbalizer would be the simple occurrence of copulas, used to embed non-verbal input and create a verbal structure that allows the application of tense and aspect, e.g. John was a great help = John helped a lot.

Supposing that there are indeed functional elements in the verbal domain equivalent to nominalizers, the study of light verbs and copulas (as well as morphological verbalizers) may be used in a comparative way to further open up and inform our understanding of the patterning and morpho-syntax of nominalization.  A comparison of nominalizers and verbalizers also has the potential to lead to interesting typological questions and the issue of whether there really is the full cross-categorial equivalence that one might pre-theoretically expect in such a domain, and if not, why full equivalence does not exist. (Simpson-fn02)

Contents of this page

02.04. Nominalization and/vs. compounding

Finally, in connection with the source of the grammaticalization of nominalizers, in certain cases (at least) it may be interesting to consider the possible connection of nominalization with compounding in a language.  It is quite plausible that in various instances nominalizers grammaticalize from nouns which are frequently used to create nominal compounds forming a connected class of items, e.g. man in doorman, chairman, tax man etc.  With the occurrence of bleaching of the meaning of such nouns and a concomitant extension of the way they can be used to combine with other words in compounding, a simple process of nominalization may well evolve.  In this regard, it is intriguing to find a correspondence between the direction of headedness in compounds in certain languages and the linear position of nominalizers relative to nominalized material.  For example, languages such as English, Japanese and Burmese which have right-headed compounds also have nominalizers occurring to the right of the constituent/word nominalized, whereas languages which have left-headed compounds, such as Thai and Vietnamese, have nominalizers which occur to the left of the constituent/word nominalized.  If such a patterning is non-coincidental and can be found to occur consistently in a wider sampling of languages, it may establish an interesting diachronic link resulting from processes of grammaticalization between two types of morphological operation regularly treated as being unrelated. 

Contents of this page

03. Nominalization in Burmese

Having considered some of the general issues involved in the study of nominalization and the grammaticalization of nominalizers in a language, we now turn to an investigation of nominalization phenomena in Burmese. The discussion here will focus in particular on the sentential/clausal nominalizers present in the language, as these can be shown to reveal much about the way reanalysis applies to create complex new grammaticalized morphemes/words, and give rise to shifts between categorical types. (Simpson-fn03).  As briefly mentioned in the introduction, Burmese is commonly described as having two complementary forms: Colloquial Burmese and Literary Burmese.

The latter is used orally in formal announcements, news broadcasting and is the most common written form of Burmese, whereas the former occurs in most spoken communication and is also sometimes found in more informal writing.  The primary and most obvious difference between the two varieties of Burmese is in the instantiation of their functional-grammatical morphemes, including the elements employed as nominalizers, and it is here that a comparison of patterns in Colloquial and Literary Burmese is often interesting. (Simpson-fn04)  The present examination of nominalization in Burmese will begin with patterns in Literary Burmese in section 3.1 and then move on to Colloquial Burmese in 3.2

Contents of this page

3.1. Sentential nominalizers in Literary Burmese: the elements thii and mii

UKT: Pronunciation:
{th} IPA [i] pronunciation similar to English <the>
{m} IPA [mji]

In Literary Burmese, the morpheme thii  {th} occurs in clause-final position, both in main clauses (as a sentence-final morpheme), and when clauses are embedded as arguments of other predicates:

(8)
U-Win-Win manee-ga        yauq-laa     thii

U-Win-Win yesterday-PAST arrive-come THII
U Win Win arrived yesterday.

UKT (08):

{U:wing:wing: ma.n.ka. rauk-la-th}

When thii  {th} is used to embed clauses as the arguments of a predicate, it is naturally accompanied by a case-marker. (Simpson-fn 05). Examples (9) and (10) show this with the embedding of clauses as the object of a verb, and (11) and (12) with the embedding of a clause in subject position. (Simpson-fn06). It should also be noted that the use of thii in all of (8-12) is obligatory and clauses may not occur as the arguments of verbs without this morpheme:

(9)
canaw [U-Win-Win manee-ga    yauq-laa   thii]-kou  caa  ya  thii.

I      U-Win-Win yesterday-PAST arrive-come THII  ACC hear get THII
I heard that U Win Win arrived yesterday.

UKT (09):

{kya.nau [U:wing:wing: ma.n.ka. rauk-la-th] ko kra:ra. th}

 

(10)
hsain-shin-hmaa [hkalee-twee hseileiq   hkou thii]-kou myin-ya thii

shopkeeper-NOM child-PL       cigarette  steal THII   ACC see-get   THII
The shopkeeper saw the children stealing the cigarettes.

UKT (10) (this example by Simpson is culturally sensitive):

{hseing-rhin-mya:  ka.l:tw hs:laip hko:th ko  mring-ra. th}

 

(11)
[thuu bamaa-sagaa      pyaw nee thii]-hmaa   theiq kaun hla thii.
he  Burma language speak ASP THII NOM very  good INT THII 
He speaks Burmese very well.
(Lit. His speaking Burmese is very good).

UKT (11):

{[thu ba.ma-sa.ka: prau:n-th] mha  thaip-kaung:lha.th

 

(12)
[leezeiq-twin   daq-poun  yaiq thii]-hmaa tayaa-m-win pa.
airport-within photograp take THII NOM    not-be-legal POL
It is illegal to take photographs inside the airport.
(Lit. Taking photographs in the airport is illegal.)

UKT (12) (this example by Simpson is culturally sensitive):

{l-hsaip twing Daat-pon reik th mha  ta.ra:ma.wing-pa}

Thii-suffixed clauses consequently have both the syntactic distribution (occurring in subject/object position) and marking of nominal phrases (nominative/accusative case).  The element thii therefore exhibits key aspects of the patterning of typical clausal nominalizers. 

Thii is also found to occur in relative clause structures such as (13): (Simpson-fn07):

(13)
canaw we thii saouq
I    buy THII book
the/a book I bought

UKT (13):

{kya.nau w-th. sa-oap}

In such an environment thii does not maintain its "mid-level tone", but instead occurs in a high creaky tone, one of the four tones that regularly occur on Burmese words.  (See UKT-fn02) As a suprasegmental morpheme [UKT: IPA [ĭ] ], such high creaky tone frequently signals Genitive case, which is otherwise realized as a discrete morpheme ye pronounced with high creaky tone:

(14)
canaw  ye  ein
I  GEN house
 my house

UKT (14):

{kya.nau r. aim}

Sequences such as that in (14) may in fact have two pronunciations. If the possessor (here canaw {kya.nau}) normally occurs with a mid-level tone, it can maintain this tone as a Genitive-marked possessor, or it is possible for (14) to be pronounced with high creaky tone on both ye and the possessor canaw  [ {kya.naau.}], the tone associated with Genitive case spreading leftwards from ye [ {r.}] to the nominal element it marks to its left.  A third pattern commonly found is for a (normally) mid-level tone possessor to occur pronounced with high creaky tone even in the full absence of Genitive ye (canawcreaked ein my house {kya.naau.aim}).  The high creaky tone of Genitive ye [ {r.}] can therefore become disassociated from its regular host and simply occur marking an appropriate nominal element.  In relative clauses such as (13), clause-final creaky tone thii  {th.} is consequently most naturally analyzed as the combination of a clausal nominalizer (mid-level tone) thii  {th} and Genitive case, so that relative clauses in Burmese are instances of the modification of a noun by a nominalized clause, linked via Genitive case. (Simpson-fn08).

UKT: Since {kya.naau.} is already showing possession, inclusion of {r.} (which also show possession) is redundant, and the following sentence is not right.

* {kya.naau. r. aim} -- incorrect sentence

This position is supported by U Tun Tint of MLC (Myanmar Language Commission -- personal communication 080623).

If we now consider what the source of the nominalizer thii  {th} might be, it can be observed that mid-level tone thii  {th} occurs elsewhere in Literary Burmese, regularly functioning as a pre-nominal demonstrative :

(15)
thii saaouq
this book
this book

UKT (15):
{th sa-oap}.
There are two ways in which this can appear: in formal writing {I-sa-oap}, and in casual speech {di-sa-oap}. Simpson's (15) is not strictly correct: it should be written as {I-sa-oap}.
(I am waiting response from my peers.)

Assuming this demonstrative element to be the most likely source of the nominalizer thii then suggests that clausal nominalizations in Literary Burmese have an internal structure such as that in (5) (parametrized in a head-final way), representing the combination of a D(eterminer)-type element with a clause to produce a DP nominal projection. (Simpson-fn09).

Interestingly, mid-level thii {th} also occurs in a third, rather different function in Literary Burmese, as a common instantiation of Nominative case marking a subject.  Example (16) thus shows four different uses of thii: as a demonstrative, Nominative case, linking a relative clause to a following nominal (with creaky tone), and in simple (non-relativized) clause-final position (Simpson-fn10).

(16)
thii pyiqsii-myaa-thii Daw-Hla-Mee htaa  hke thii-myaa hpiq pa thii
this thing-PL-NOM         Daw-Hla-Mee  put  ASP NZR-PL      be POL   NZR
These are the things that Daw Hla Mee left behind.' (Simpson-fn11).

UKT (16):
-- objectionable
{th pic~s:mya:th dau-lha.m hta:hk.th-mya: hpric-pa-th}

UKT (16a)
-- acceptable
{I pic~s:mya:th dau-lha.m hta:hk.th-mya: hpric-pa-th}

Assuming a demonstrative source of nominalizer thii to be plausible and likely, the grammaticalization of thii in its nominalizer function brings with it a further question relating to word order.  As demonstrative thii occurs preceding the nominal complement it combines with, why might it be positioned following a clausal complement when nominalizing the latter?  Although no definitive answer can be provided in the absence of data on the early development of nominalizing thii, two speculations can be offered here.  First, as Burmese is a head-final language, and thii  as a demonstrative is likely to be positioned in a phrase-initial specifier position (SpecDP), when such an element grammaticalized as a nominalizer, it can be hypothesized that it came to occupy the head-position of its (DP) phrase in an occurrence of the Spec-head reduction process argued to frequently characterize instances of phrasal grammaticalization (see Simpson & Wu 2002a and van Gelderen 2004). Given the linear organization of (elements in) specifier and phrasal head positions in a head-final, specifier-initial language such as Burmese, Spec-head reduction and grammaticalization would be expected to relocate a demonstrative from a phrase-initial specifier position to a phrase-final head-position, and so result in nominalizer thii coming to follow its clausal complement.

A second possibility might be to hypothesize that thii came to be used as a demonstrative in a resumptive position following a clausal subject, as occurs in certain Indic languages, and schematized in (18):

(18)
[John likes Mary]i, thisi is true.
It is true that John likes Mary/That John likes Mary is true.

Such a linear sequence might then lead to grammaticalization of the demonstrative as a nominalizer following the clause it introduces:

(19)
[[John likes Mary]-this] is true.

In addition to the question of the linear order of thii relative to its complement clause, an interesting complication for the analysis of thii and both its synchronic status and grammaticalization comes from the fact that thii is actually not just a simplex nominalizer derived from a demonstrative, but also encodes realis mood, being used to combine with clauses that represent past or present actions, states and habits.  Where future actions and hypothetical situations are referred to, a fully parallel set of nominalized constructions are built with the nominalizer mii, which differs from thii only in encoding irrealis mood (i.e. non-realized actions/states) (Simpson-fn12). This is illustrated in (20) and (21) with clauses in subject position and Nominative case, and in (22) with an Accusative-marked clause in object position:

(20)
[than shi mii]-hmaa-le    ahman hpyiq ii.
lice  be  NZRIRR-NOM-too   true    be    II
That there will be lice too is true. (Okell and Allott 2001:158)

UKT (20) (this example by Simpson is culturally sensitive):

{than: rhi.m mha-l: a.mhan-hpric-th}

 

(21)
[thati pyu ya mii]-hmaa      htaransitsataa-amyouasaa-pin hpyiq thii
attention  do  must NZRIRR-NOM  transistor-type-EMPH       be     NZR REAL
What we will have to pay attention to is the transistor type. (Okell and Allott 2001:158)

UKT (21):

{tha.ti.pru. ra.m-mha tran-sic-sa-ta a.myo:a.sa:ping hpric-th}

 

(22)
[naneq ngaa naayii khan shi mii]-kou      aloulou-hman-mi thii.
morning 5 oclock about be NZRIRR-ACC  estimate   NZL REAL
(He) estimated that it must be about 5 oclock.
(Adapted from Okell and Allott 2001:158)

UKT (22):

{nn-nak nga:na-ri hkan. rhi. m ko  a.lo-lo mhan:mi.th}

Example (23) shows the occurrence of irrealis mii in sentence-final position, paralleling thii in example (8), and (24) shows its occurrence embedding a relative clause. As with thii when it links a relative clause to a noun, mii carries a high creaked tone in (24) corresponding to Genitive case, and is elsewhere pronounced with mid-level tone.

(23)
maneqhpan   pyan   laa     mii
tomorrow      return come NZLIRR
He will come back tomorrow.

UKT (23):

{ma.nak-hpan pran-la m

 

(24)
[thuu yuu laa   mii]       pyiqsii-myaa
he   take come NZLIRR.GEN thing-PL
the things he will bring

UKT:

{thu yu-la-m. pic~s:mya:

A further interesting parallel in morpho-syntactic patterning between thii and mii is that mii also occurs in a pre-nominal interrogative specifier function, which corresponds closely to the demonstrative function of thii, differing essentially in the hypothetical irrealis-type property of mii: whereas demonstrative thii is realis in picking out a referent with definite reference, interrogative mii is used for hypothetical reference in questioning the reference of a noun:

(25)
mii-thuu 
which-person/one
who

UKT:

{m-thu}

The clausal nominalizers thii and mii in Literary Burmese are therefore complex elements incorporating two parts/functions:

(a) an ir/realis component: th-/m-   / {tha./ma.}
(b) a nominalizer component, the common/shared vowel coda: -ii  {}

UKT: It is interesting to note that the word for <visitor> {.th} (guest --MEDict625) is pronounced {.th}. However, the vowel {} which according to Simpson is the nominalizer in the words {th} and  {m} is pronounced {i}.

Of further potential relevance here is the observation that in addition to pre-nominal demonstrative/interrogative thii and mii, Literary Burmese also contains a simplex demonstrative element unspecified for ir/realis distinctions whose form corresponds exactly to the nominalizing component ii in thii and mii:

(26)
ii    saaouq
this book

UKT:

{I-sa-oap}
In casual speech and writing, this can become {di-sa-oap}.

Quite possibly then, the complex demonstrative-nominalizers thii and mii are grammaticalizations of the combination of ir/realis together with this simplex demonstrative ii, and the creaky tone nominalizers occurring in relative clauses will in turn be composed of three distinct components (Simpson-fn13):

ir/realis + demonstrative-nominalizer + genitive case
th/m +  ii + creaky tone

UKT:
th-/m-   / {tha./ma.} + {} + ( {wuc-sa.} - or "dot below")

Contents of this page

03.02. Sentential nominalizers in Colloquial Burmese

Turning now to patterns in Colloquial Burmese, here it is found that there are elements which occur in the same clause/sentence-final and relative clause positions positions as the nominalizers thii and mii in Literary Burmese.  Similar to Literary Burmese, the linking elements in relative clause structures carry a creaky tone. Also parallel to the situation in Literary Burmese, there are two sets of elements which occur in these positions, one set encoding realis mood and the other irrealis:

(27)
realis clause/S-final:    te     realis relative clause linker:   te+creak
irrealis clause/S-final: me    irrealis relative clause linker: me+creak a

UKT:
{t}  {t.}
{m}  {m.}

Examples (28) (32) illustrate the use of these morphemes in clause-final, sentence-final and relative clause environments:

(28)
thuu maneqphan laa  me htin   te
he     tomorrow come ME think TE
I think he will come tomorrow.

UKT:

{thu nak-hpan la-m hting-t}

 

(29)
thuu manee-ka saaouq we te
he   yesterday   book    buy TE
He bought a book yesterday.

UKT:

{thu ma.n.ka. sa-oap w-t}

 

(30)
thuu manee-ka  we  te+creak   saaouq
he  yesterday     buy TE          book
the book he bought yesterday

UKT:

{thu ma.n.ka. w-t. sa-oap}

 

(31)
thuu maneqphan thamain-saaouq we me
he    tomorrow   history-book   buy ME
He will buy a history book tomorrow.

UKT:

{thu ma.nak-hpan tha.meing:sa-oap w-m

 

(32)
thuu maneqphan we  me+creak thamain-saaouq
he   tomorrow   buy ME   history-book
the history book he will buy tomorrow.

UKT:

{thu ma.nak-hpan w.m. sa-oap}

Such neat parallels between Literary and Colloquial Burmese diverge however in instances where clauses occur as the clearly nominalized arguments of verbs of perception and cognition in subject and object position in Colloquial Burmese.  In such environments, where the overt occurrence of case-markers confirms the nominalized status of the embedded clauses, the elements which embed clauses in Colloquial Burmese are not in fact homophonous with the clause/sentence-final and relative clause morphemes indicated in (27), but instead have the forms taa for realis clauses, and hmaa for irrealis clauses. Examples (33) (37) show the occurrence of these elements in a range of subject, object and object of postposition and ir/realis combinations:

(33)
[thuu thoun-nayii-hmaa hote-ka htweq]-taa-kou myin-ya  te
he   3 oclock-at     hotel-from exit   - TAA-ACC see-get TE
(I) saw him leaving the hotel at 3 oclock.

UKT:

{thu thon:na-ri mha hau-t-ka. htwak-ta-ko mring-ra.t}

 

(34)
[thin ne]-taa-ga      hkeq-th-la?
learn ASP - TAA-NOM difficult-Q
And how about learning it, is it difficult?

UKT:

{thing-n-ta-ka. hkak-tha.la:}

 

(35)
[thuu baa-hma m pyaw]-taa-ne apyin htweq thwaa te
he    whatever not say   TAA with outside exit go TE
(I) left as/when he didnt say saying anything.

UKT:

{thu Ba-mha.ma.prau:ta-n. a.pring htwak-thwa:t

 

(36)
[ale pyan]-hmaa-kou pyaw th-laa?
visit return HMAA ACC say Q
Did he say he was going to come for a visit?

UKT:

{a.l-pran mha ko prau:tha.la:}

 

(37)
[thuu dii-lou louq]-hmaa-ga     theiq m kaun pu,   htin  te
he    this way do    HMAA NOM very not good not think TE
I think it is not good at all that he is going to do that.

UKT:

{thu di-lo-loap-mha-ka. thaip-ma.kaung:Bu: hting-t

The most plausible analysis of the elements taa and hmaa is that they result from the collapse of te/me together with the light noun haa meaning one/thing (similar to English this one/that one/the blue one):

(38)
te + haa --> taa
me + haa --> hmaa

UKT:
{t} + {ha} --> {ta}
{m} + {ha} --> {mha}

Elsewhere in Burmese the element haa  {ha} occurs with this meaning one/thing and can also optionally fuse with the demonstrative dii  {di}  resulting in the form daa  {da}.

(39)
dii  haa   --> daa
this one/thing    this/that

UKT:
{di-ha} --> {da}

The clause-final elements taa  {ta} and hmaa  {mha} are also found to occur in headless relative clauses, as in (40) - (42):

(40)
[canaw hmaa]-taa-ga     asein-caw          pa
I    order-TAA-NOM vegetable-fried POL
What I ordered was fried vegetables. (lit. the one/thing I ordered..)

UKT:

{kya.nau mha ta ka. a.sain:kyau pa}

 

(41)
[thuu Yangoun-hmaa we]-hmaa-ga   seqbein pa
he  Yangoun-in   buy HMAA-NOM bicycle POL
What/the thing he is going to buy in Rangoon is a bicycle.

UKT:

{thu ran-koan-mha w-mha-ka sak-bi:pa}

 

(42)
[canaw yee-htaa]-taa-kou   thuu  theiq m caiq pu
I    write-ASP-TAA-ACC   he   very not like not
What I wrote he really does not like.

UKT:

{kya.nau r:hta:ta-ko thu thaip ma.kraik-Bu:}

In such environments taa and hmaa here again arguably result from the collapse of te/me and the dummy/light noun haa {ha}, resulting in a complex fused morpheme with the meaning the one/thing that..

Finally, taa and hmaa are additionally found in alternation with te and me in sentence-final position, as in examples (43) and (44).   In such instances, the use of taa/hmaa rather than te/me results in a clear difference in meaning from te/meand a cleft-like interpretation with focus on one part of the sentence as new information set off against a presupposed background (similar to the Chinese shi-de construction and Japanese no-desu forms, see Simpson 2003a, Simpson and Wu 2002b).

(43)
canaw zee-hmaa we taa
I  market-in  buy TAA
I bought it in the market./Its in the market that I bought it.

UKT:

{kya.nau Z: mha w ta}

 

(44)
thuu maneqphan yauq hmaa
he    tomorrow   arrive HMAA
He will arrive tomorrow. / Its tomorrow that he will arrive.

UKT:

{thu ma.nak-hpan rauk mha}

From such a distribution and patterning, one can hypothesize a likely three-step route of grammaticalization and development of taa and hmaa from te and me.  From an earlier stage in which te/me occurred in all clause-final (including relative clause-final) positions, similar to the distribution of Literary Burmese thii/mii, it can be supposed that the combination of te and me with haa one in headless relative clauses resulted in two collapsed/fused relative clause nominalizers taa and hmaa.  Where such taa/hmaa-final relative clauses might have occurred as the natural object of a copula in a main clause (e.g. This is the one/thing I bought yesterday.), significantly this would then have resulted in taa/hmaa occurring in sentence-final position, as copulas are regularly null in Colloquial Burmese.  Finally, from such occurrence as (relative) clausal nominalizers in sentence-final position, it can be hypothesized that taa/hmaa may have allowed for an expansion of use as clause-final nominalizers in clauses which were not only sentence-final but alternatively embedded as the clausal arguments of verbs of perception and cognition (and as also subjects of one-place predicates such as (to be) easy/difficult/interesting etc). 

If the above can be assumed to be a reasonable path of development for taa and hmaa as general clausal nominalizers not just restricted to relative clauses, this now raises an important question about the synchronic status of te and me.  If te and me are the original clause-final morphemes in Colloquial Burmese (and partial inputs to the later creation of taa and hmaa), and equivalents to Literary Burmese thii and mii, as seems most likely, and if the latter elements thii/mii in Literary Burmese can be shown to be nominalizers, a fairly natural conclusion is that te and me may have shared this function as nominalizers too, at least in their earlier stages of development. Such an assumption is supported by the observation that a Genitive case creaky tone is present with te and me in their occurrence as clause-final elements in relative clauses. If (Genitive) case is naturally taken to be marked only on nominal constituents, then te/me-final relative clauses must be assumed to have been nominal(ized) clauses at least when the Genitive creak was first added to te/me.  Consequently, like Literary Burmese thii and mii, te and me can be concluded to have had a nominalizing function when added to clauses, in addition to expressing ir/realis mood, at least at some point in their development. The question to be considered now is synchronically whether these elements are (still) nominalizers in the way that their frequent Literary Burmese equivalents thii and mii can be shown to be.

UKT: What Simpson has meant by "Genitive case creaky tone" is understood to be "Possessive case pronounced similar to pitch-register #1 of spoken Burmese. The "creaky tone" has an extra-short duration (IPA [ă]).  See Modal voice in The Human Voice by UKT http://www.tuninst.net/Romabama/Human-Voice/HV4/hv4.htm#Modal-voice 080627

The answer here would seem to be clearly and interestingly no.  If simple tests for the nominal status of te/me-marked clauses are employed, it is found that te and me do not in fact show signs of nominalizing the clauses they are combined with.  First of all, it is found that case-markers cannot occur on clauses ending in te/me, unlike clauses ending in the clear nominalizers taa and hmaa (compare (45) and (46) with earlier (33), (34), (36) and (37)).

(45)
[U-Win-Win manee-ga yauq   te](*-kou) pyaw te
U-Win-Win yesterday arrive TE ACC      say TE
(He) said that U Win Win arrived yesterday.'

UKT:

{U:wing:wing: ma.nak-ka. rauk t (-ko) prau:t}

 

(46)
[canaw maneqhpan thwaa me](*-kou) htin te
I    tomorrow  go  ME ACC   think TE
'I think I'll go tomorrow.' (Simpson-fn14).

UKT:

{kya.nau ma.nak-hpan thwa: m ko htin t

Second, postpositions such as ne  with {n.} can only occur with clearly nominalized clauses ending in either taa or hmaa and not te /me :

(47)
sagaa sa   pyaw laa   taa/*te   ne   ta-pain-neq,
word begin say come TAA/TE with at-the-same-time
At the same time that he began speaking,

UKT:

{sa.ka: sa. prau:la-ta n. ta.preing-nak ... }
*
* {sa.ka: sa. prau:la-t n. ta.preing-nak ... }

Third, other category-sensitive patterns involving elements introducing rationale clauses similarly indicate that while taa and hmaa do create nominal categories, te and me do not.  Specifically, the Burmese words mouq and caun (meaning because) {kaung.} are elements which only allow for combination with nominal/noun phrases, and are found to naturally occur with taa/hmaa-final (nominalized) clauses but may not occur with te/me-final clauses, indicating clearly that the latter elements (synchronically) do not nominalize the clauses they combine with (Simpson-fn15).

(48)
[clause + taa/hmaa]-mouq/caun because of [clause]
*[clause + te/me]-mouq/caun because of [clause]

 

(49)
ngwee lou    taa/*te caun,..
money need TAA/TE because
Because he needed money,

UKT:
/
{ngw lo-ta-kraung. ... } / {ngw lo-t-kraung. ... }

 

(50)
maneqhpan ethe-twee laa   hmaa/*me mouq,..
tomorrow   guest-PL come HMAA/ME because,..
Because there are guests coming tomorrow,..

UKT:

{ma.nak-hpan .th-tw la-mha / (la-m

UKT note: <visitor> {.th} (guest --MEDict625) is pronounced {.th}.

The conclusion to result from the above is that the elements te and me in modern Burmese do not create nominal categories and therefore synchronically are not nominalizers. As the available evidence and patterns surveyed from elsewhere in the language and connected with te/me clearly suggests that te/me were at one time part of a nominalizer paradigm similar to Literary Burmese thii and mii, it would seem that Colloquial Burmese te and me can be seen as cases of nominalizers that have undergone a further development in their path of grammaticalization, significantly losing any nominal-related categorical specification that would support case-marking and other properties of nominal(ized) constituents, and undergoing reanalysis as grammatical markers more strictly associated with verbal syntax and the representation of ir/realis mood.  If it is supposed that nominal syntactic patterns (such as case-marking) result from the presence of a DP (section 2.1), or some other lower constituent such as a NumP, ClP or NP which might project a DP level of structure, the loss of expected nominal syntactic patterns can be interpreted as a reanalysis in syntactic category and the change in the categorial identity of an element from one associated with nouns D, Num, Cl or N to one associated with verbs, such as T(ense), Asp(ect), Mood, or C(omplementizer).  In the case of main clause te and me, it might seem reasonable that their categorical identity is now strictly Mood and hence that their hypothesized reanalysis would be from an occurrence and function as a D-type nominalizer to a plain Mood marker combined with verbs an instance of cross-categorial re-grammaticalization from one major functional domain to another (lateral re-grammaticalization Simpson 2003a) (Simpson-fn16).

A further conclusion following from the above is that if te /me are no longer nominalizers in Burmese, then the occurrences of te and me with creaky tone in relative clauses can no longer synchronically be the combination of a nominalizer with Genitive case. (Simpson-fn17). Here again it would seem that categorial reanalysis must have applied converting a nominalizing morpheme + Genitive case into a new, composite form that is (a) non-nominal/not a category which projects a nominal constituent, and (b) specified for occurrence with verbs and their projections in the environment of relative clauses. If the occurrence of creaky tone on te and me in relative clauses suggests that te and me in relative clauses are not identical to te and me elsewhere in clause-final position (hypothesized to be instantiations of just Mood), then the dedicated embedding function of creaked te and me in relative clauses can be suggested to represent a reanalysis of te and me as new (relative clause) Complementizer elements, formally of type C and so distinct from pure Mood te and me, though carrying a specification for ir/realis mood in a way that recalls the complex occurrence of subject agreement on Complementizers in languages such as Flemish and Bavarian German (Fuss 2005) (Simpson-fn18).

The syntactic patterning observed with Colloquial Burmese te and me and the way this is distinguished from the nominalizer forms thii and mii in Literary Burmese therefore leads to the hypothesis that a pair of erstwhile nominalizers have undergone a further, important reanalysis as elements lacking nominal specifications, and have come to instantiate new verb-related categories. Generally, then, it can be argued that the grammaticalization of elements as nominalizers does not necessarily represent a final stage of functional development and that nominalizers may in fact develop further into verb-related grammatical categories with formally different syntactic properties.  To the extent that the distinction between clause-final te/me and relative clause embedding creaked te/me seems to require the assumption of a categorial identity for the latter that is formally different from the former, the paradigms examined here also provide support for the existence of a special C(omplementizer) category.  Such a category has often been assumed for elements such as English that, taken to be reanalyzed as a C(omplementizer) from an earlier source as a D(emonstrative), but clear evidence in English for such a categorical switch/reanalysis is not easy to identify.  The patterns examined here in Burmese helpfully seem to suggest that the occurrence of D-to-C re-grammaticalization and the occurrence of a formally distinct category C is indeed quite plausible. 

Contents of this page

04. Summary of conclusions and consequences for further work.

This study of Burmese has attempted to explore and highlight a number of properties of nominalization and its grammaticalization which we can now summarize in brief here. In section 2.1 issues of the sources of nominalizers were discussed and it was suggested that nominalizers may in fact arise from a variety of sources, including elements of type D, Cl, Num and N.  In the course of the investigation of Literary Burmese in section 3, evidence was then presented indicating that clausal/sentential nominalizers in Burmese may be sourced from demonstratives of category D which have also grammaticalized a modal feature encoding ir/realis distinctions.  Section 2 also emphasized the need for specific morpho-syntactic evidence when attempting to classify elements as nominalizers and noted that the term nominalizer may sometimes be used without clear confirmation of the nominalizer status of an element. In section 3 it was then argued that while certain morphemes in Burmese can be concluded to be nominalizers, other elements with a superficially similar distribution have in fact undergone reanalysis as elements of a different categorial type and no longer function as nominalizers in the language.  Considering the syntactic realization of relative clauses in Burmese, it was shown that at least two functional morphemes are (historically) involved in the linking of a relative clause to a noun(-phrase): (a) a nominalizer, and (b) Genitive case. This raises an important question about the identity of linking morphemes in other similar languages (such as, for example, Chinese) where just a single element links a relative clause to a head-noun are such elements nominalizers or occurrences of Genitive case, or possibly even a grammaticalized composite of both?  The examination of Burmese finally also showed how nominalization structures are potentially unstable and may arguably undergo reanalysis as purely verb-related forms, with nominalizers being absorbed into the functional system projected in a clause.  Considerably rich in the information provided by its two systems of functional elements in the colloquial and literary forms of the language, Burmese consequently offers an interesting perspective on variation in processes of grammaticalization and the reanalysis of grammaticalized morphemes in new directions.

Contents of this page

Simpson's footnotes

Simpson-fn01. It is also possible that certain nominalizers attached to word-level elements might be syntactic attachments, combining with a word-level constituent within a syntactic structure, perhaps via cliticisation or some form of (syntactic) head-movement. Various approaches to English nominalizing ing assume such an analysis (and Lotha (Tibeto-Burman) a - would be another candidate for a similar treatment, see Herring 1991).
Go back Simpson-fn01b

Simpson-fn02. For example, while it may seem that nominalizers grammaticalize from a range of functional elements commonly associated with nouns demonstratives, classifiers, numerals it is not obvious that verbalizers are derived from higher functional elements associated with verbs, such as tense, aspect and mood morphemes, and the main source of verbalizers would seem to be verbs. Why this asymmetry between the source of nominalizers and verbalizers exists is not immediately clear.
Go back Simpson-fn02b

Simpson-fn03. In addition to the clausal/sentential nominalizers discussed in this work, Burmese also has a wide range of other nominalizers, many of which combine with sub-clausal constitutents. For simple reasons of space and focus, no attempt to describe these will be attempted here (see Okell 1969, Okell and Allott 2001, and Hopple 2003 for much useful information and a comprehensive listing of all putative nominalizers in Burmese).
Go back Simpson-fn03b

Simpson-fn04. A highly significant proportion of non-grammatical vocabulary is shared by Colloquial and Literary Burmese, though the latter also has available certain variant forms (frequently multi-syllabic) which are often made use of in writing (see Okell 1969, and Saw Tun 2005).
Go back  Simpson-fn04b

Simpson-fn05. The case-marking of nominal arguments is generally optional in Burmese, though heavily preferred in certain instances, both as parsing aids and sometimes also to signal contrastive emphasis, as in Korean (Schhtze 2001). Informants indicate that the use of case-particles with thii-embedded clauses is much preferred.  For further discussion of factors triggering the use of case-marking particles in Burmese see Kassevitch (2005).
Go back Simpson-fn05b

Simpson-fn06. ACC = Accusative, NOM = Nominative, ASP = Aspect, POL = Politeness marker.
Go back Simpson-fn06b

Simpson-fn07. In addition to thii, Literary Burmese also makes use of a morpheme thaw as an alternative linking element in relative clauses.  This element is particularly frequent with intransitive predicates and adjectives (see Okell 1969 and Hopple 2003 for further discussion of thaw).
Go back Simpson-fn07b

Simpson-fn08. Such an analysis is well-supported by the occurrence of Genitive case with putative nominalizations in many instances of relative-clause marking/linking in Tibeto-Burman languages (see, for example, Noonan 1997, Thurgood and LaPolla 2003).
Go back Simpson-fn08b

Simpson-fn09. See also Herring (1991) for evidence from other languages that clausal nominalizers may be sourced from demonstratives in Tibeto-Burman languages.
Go back Simpson-fn09b

Simpson-fn10. A teasing question is how the Nominative case marker thii might be linked to and possibly derived from the demonstrative element thii. One possibility here may be to consider that subject noun-phrases are more frequently definite and anaphoric in reference than objects are, and so might attract marking with a demonstrative, encoding a familiar, +definite referential value.
Go back Simpson-fn10b

Simpson-fn11. NZR = Nominalizer, ASP = Aspect, POL = Politeness particle, PL = Plural.
Go back Simpson-fn11b

Simpson-fn12. As well as thii and mii, Literary Burmese has an element ii (pronounced with a high creaky tone) which occurs as a stylistic variant of thii/mii in sentence-final position or embedded quotations with no specification for ir/realis (see Okell and Allott 2001:271-2, and footnote 17 below).  An example of the sentence-final use of ii occurs in example (20). 
Go back Simpson-fn12b

Simpson-fn13. The simplex demonstrative ii is furthermore found in early Burmese inscriptions before thii and mii are attested, supporting the feeding relation hypothesized here with thii and mii being formed from ii together with an encoding of ir/realis meaning.  Thanks are due to a reviewer of the chapter for drawing attention to the early occurrence of ii in Burmese inscriptions. 
Go back Simpson-fn13b

Simpson-fn14. Matisoff (1972) offers an example of the attachment of Accusative case -kou  {ko} to a clause ending in te  {t} (* below).  However, this sentence is actually not well formed and is rejected by native speakers of Burmese.  First of all the occurrence of negation should automatically displace the occurrence of te as in all other negative sentences.  Secondly, -kou cannot co-occur with te in other, non-negative environments, hence even if the negation is removed, the sentence remains ungrammatical with -kou.

*[[hkinpya shi te m-shi   te]-kou   be-hne thi m-le?
     you        be TE not be    TE ACC   how     know Q
     How will I know whether youre there or not?

UKT:

{hking-bya: rhi.t ma.rhi.t ko B.n. thi. ma.l}

[UKT: The contention by Simpson that "the above sentence is actually not well formed and is rejected by native speakers of Burmese" is debatable. As a native speaker of Burmese myself, I can not concur with Simpson.]
Go back Simpson-fn14b

Simpson-fn15. This restriction of mouq and caun on the categorical status of their complements permitting only nouns/noun-phrases would not seem to be reducible to the meaning of mouq/caun as because (of).  Other lexical items in Burmese with a similar meaning, such as lou because, impose fully opposite selectional restrictions, combining only with verbal/clausal categories and not allowing noun/nominal complements.  It is significant to note that lou does allow for combination with a clause ending in te/me, supporting the assumption shortly put forward here that such clauses are not nominalizations.
Go back Simpson-fn15b

Simpson-fn16. Such a putative reanalysis here might seem to represent a simplification and reduction of the featural specification of te/me from being [+nominal, +mood], by hypothesis, te /me become simply [+mood].
Go back Simpson-fn16b

Simpson-fn17. And indeed, creaked te and me do not allow for separation of the Genitive case from te/me via the use of independent Genitive case-marker (creaked) ye: *[..te/me ye], suggesting that creaked te and me are fully fused, new forms and are not decomposable into a nominalizer and Genitive case sub-parts. Also note that if te/me do not induce nominal syntactic constituents (i.e. nominalize other phrases), then the creaky tone suprasegmental that is elsewhere an indication of Genitive case cannot/can no longer be Genitive case when present with te/me (in relative clauses), as case-marking only marks nominal constituents. Finally, the conclusion that Genitive case in relative clauses has been reanalysed as a part of a different type of morpheme is supported by other cases in Burmese where Genitive case-marking seems to have clearly undergone reanalysis.  In both Colloquial and Literary Burmese there are sentence-final morphemes which have developed from Genitive case-markers (creaked ye in Colloquial Burmese and creaked ii in Literary Burmese) and which now function as markers of (a) surprise in Colloquial Burmese, and (b) the end of the sentence in an ir/realis neutral way in Literary Burmese. Given the complementary distribution these elements have with other markers of the category of Mood, it is most natural to see them as reanalysed into this category and no longer performing any (Genitive) case function.
Go back Simpson-fn17b

Simpson-fn18. For a view on how a single morpheme might relate to two syntactic featural specifications such as C and Mood, or Tense and Mood, see Simpson (2003a).
Go back Simpson-fn18b

Contents of this page

UKT's footnotes

UKT-fn01. At one time when I was young (I am now an old man of 75), words which are equal to the English <at once> or <immediately> were spelled {ta.hka-t:}. However, since about 1980, MLC (Myanmar Language Commission) has changed the prefix {ta.} to {tic} to show the numeral "one", and {ta.hka-t:} is now spelled {tic-hka-t:}. (See MEDict (Myanmar-English-Dictionary by MLC) p181)). However, I cannot accept the reason given because there is no word such as {nhic-hka-t:} where {nhic} would show the numeral "two".
Go back UKT-fn01b

UKT-fn02. The "mid-level tone" mentioned by Simpson is the "Modal Voice" mentioned in The Human Voice, by UKT http://www.tuninst.net/Romabama/Human-Voice/HV4/hv4.htm#Modal-voice 080627 

The Burmese-Myanmar language is a pitch-register language : with 3 basic vowels {a. a a:}, and vowels followed by "killed" consonants -- {a.tht}. Because of this the language is described by some as having 4 tones.  The first is a creaky register (creaky tone), and the second a modal register or modal voice.  The word {kya.naau.} is derived from the modal and a {wuc~sa}. Though similar in pitch-register to {a.}, it is derived from {a} to show possession. It is a suprasegmental and is represented by IPA [ă]. To bring this out more clearly, I will rewrite (14) substituting the more polite {kya.nau} with more basic {nga}:

(14a)  {nga r. aim}

The possessive case {ngaa.} is derived from {nga} IPA [ŋa] and a {wuc~sa.} resulting in a suprasegmental IPA [ŋă]. 
Go back UKT-fn02b

Contents of this page

Simpson's References

Fuss, Eric. 2005. The Rise of Agreement: a Formal Approach to the Syntax and  Grammaticalization of Verbal Agreement. Amsterdam: John Benjamins.

Herring, Susan.  1991.  Nominalization, Relativization, and Attribution in Lotha, Angami, and Burmese.  Linguistics of the Tibeto-Burman Area, 14:55-72

Hopple, Paulette M.  2003.  The Structure of Nominalization in Burmese.   Ph.D. diss, University of Texas, Arlington.

Kassevitch, Vadim.  2005.  Syntactic and Morphological Markers in Burmese: are they really Optional?  In Justin Watkins (ed.), Studies in Burmese Linguistics.  Canberra: Pacific Linguistics: 87-96.

Matisoff, James.  1972.  Lahu Nominalization, Relativization, and Genitivization. In John Kimball (ed.), Syntax and Semantics Volume 1. Seminar Press, New York: 237-57.

Noonan, Michael. 1997. Versatile Nominalizations. In Joan Bybee, John Haiman and Sandra Thompson (eds.), Essays on Language Function and Language Type. Amsterdam: John Benjamins: 373-394.

Okell, John. 1969.  A Reference Grammar of Colloquial Burmese.  Oxford: Oxford University Press.
Okell, John and Anna Allott. 2001.  Burmese/Myanmar Dictionary of Grammatical Forms. London: Curzon.

Saw Tun.  2005.  Modern Burmese Writing: an Examination of the Status of Colloquial Burmese.  In Justin Watkins (ed.), Studies in Burmese Linguistics.  Canberra: Pacific Linguistics: 201-220.

Schhtze, Carson.  2001.  On Korean Case stacking: the Varied Functions of the Particles ka and lul. The Linguistic Review 18:3:193-232.

Simpson, Andrew.  2001.  Definiteness Agreement and the Chinese DP.  Language and Linguistics 2:1:125-156.
Simpson, Andrew.  2003a.  Empty Determiners and Nominalization in Chinese, Korean and Japanese.  In Audrey Li and Andrew Simpson (eds.), Functional Structure(s), Form and Interpretation: Perspectives from Asian languages. London: CurzonRoutledge: 131-161.
Simpson, Andrew.  2003b.  On the Status of Modifying DE and the Structure of the Chinese DP.  In Sze-Wing Tang and Luther Liu Chen-Sheng (eds.), On the Formal Way to Chinese Languages. Stanford: CSLI: 74-101.
Simpson, Andrew and Zoe Wu.  2001.  The Grammaticalization of Formal Nouns and Nominalizers in Chinese, Japanese and Korean.  In T. E. McCauley (ed.), Language Change in East Asia. London: Curzon:250-283.
Simpson, Andrew and Zoe Wu.  2002a.  Agreement, Shells and Focus.   Language 78:2:287-313. 
Simpson, Andrew and Zoe Wu.  2002b.  From D-to-T: Determiner Incorporation and the Creation of Tense. Journal of East Asian Linguistics 11:2: 45-90.

Thurgood, Graham and Randy LaPolla (eds.).  2003.  The Sino-Tibetan Languages.  London: Routledge.

van Gelderen, Elly. 2004. Grammaticalization as Economy. Amsterdam: John Benjamins.

[End of MS Word paper]

Contents of this page

End of TIL file