The Qualificative in Tswana E.S. M%to (Dept. of Mrican Languages VBS) The current definition of a qualificative as a word which qualifies a Substantive is problematic. It immediately raises questions such as: i) Must the Substantive necessarily be there too, already identified, standing qualified, for us to identify a qualificative? In NATURAL speech, when do we identify our words, before or after uttering them, before or after structuring them syntagmaticallY? iii) If before, then a choice has already been made by the time the words are uttered in such relationship that one qualifies and the other is qualified. On what basis was such advance selection made? We see an object in the distance. We tell its colour - white. We still don't know what it is. Coming closer we name it - an elephant. It appears here that our choice of colour (Qua- lificative) came before and independently of our naming the object we saw, before supplying the Substantive. We may even identify and name the object (the Substantive) before determining and utteri.g its colour. 89 The question is: on what basis does a baby choose Tswana mma (mother) and weeks or months after, -~ < monate (sweet)? There seems no relationship between the chaise of the naming word and that of the qualifying word, nor between the circumstances of their choice. Since, however, we teach language from books, it is possible to place words deliberately in certain rela- tionships in order to make a pre-conceived point. Let us therefore examine the following paradigm of qualificatives in terms of our definition: Substantive Qualifica ti ve a.. kgomo e tshwana a cow - black b. pitse e tilotsana a mare - black and white-spotted c. kgomo e e gangwang e a cow that is milked se na namane e (JUilk cow) this having no calf d. e e gangwanE e e tladitseng which is milked, it e se ns kgamelo having no calf, that namane e has filled the pail e. kgomo eo e gangwang the cow there that is milked f. kgomo ele e gangwang the cow yonder that jaanong is being milked now Attention must be invited to a few important points: 1. The qualificatives above qUalify Substantives and we knew a Qualificative and a Substantive before hand so to place them. 2. SaDIe contain one word, some more , up to three words even by conjunctiVe standards. c, d, e, t are c1ause5~ The Qualificative clause of (c} is identical with the substantival clause of (d ) • 90 5. The first lexical item ~!-7 in the Qualifica- tive of (c) is identical with the last ite$ of the same Qualificative. Conjunctivists would render the first one conjunctively, referring to it as the demonstrative element of the relative concord and the last one disjunctively recognising it as a Demonstrative, thus: eeRangwang esenanamane e 6. A very ticklish problem arises here, why one Demonstrative is a formative and the other a word. These are the points on which consensUs is hard to reach regarding the word. This kind of irregularity occurs currently in Zulu too, for example, in the word-group lomuntu (this person) and by transposition umuntu 10. The first 10 (this) is treated as a prefixal for- mative and the next one as a word. This goes to show that the near boundary of a word is blurred by a lack of clarity as to what a prefix is. Hence, when a word is placed before another it may be treated as a prefix, even if in the next breath it appears again as a word, and a separate unit at that. What we have seen above, however, is that a Qua- lificative is not always a word although a qualifying word, if we can identify it, will always be a Qua1ifica- tive. This places a high premium on the proviso in " . t" into Qualifica- current grammars that the c1ass1f1ca 10n tives, Substantives and other '-ives' (six in all) is . t' hip and that based on function and grammat1cal rela 10ns , there is yet another classification, into word-classes, The classification into based on the form of the wor d s • • d t b an exercise the six '-ives' is therefore adm1tte 0 e 91 in syntax. It is highly doubtful that it can pass as an exercise in word-identification. For as long, however, aa the two exercises run concurrently a string of words can always drop into the same slot as a word, and so long __ y we refer to both as a word. D. T. 001e1 'Warns agains~ this, saying, It ••• all relatives, strictly speaking, are relative clauses". This means that currently, because a functional concept ea Qualifieative) and a lexical concept (a Relative) are both defined as a word, a clauae is a word and a word is a clause in Tswana, in Southern Bantu. The idea of de Sllusllurethat, \lIn language there are only differences" ia relevant here. If there is no difference between a Relative and a clause there are no two linguistic issues. One must disappear. How then do we identify the qualifi- aative word? Or any other word? The late Prof. D. Ziervogel of the University of South Africa gives a reaum~ of the views of various authorities on the word generally and we may learn from hilll. D. Ziervogel sees a turning-point with the pub- lication or O. M. DOke's Textbook of Zulu Grammar in 1927, where O. M. Doke, rollowing Daniel Jones's (London) idea or phonetics, determines the Zulu word and by impli- cation the Bantu word phonetically; the criterion being stress on the penultimate syllable as signal of its far boundary. We might add here that A. G. Nkabinde2 sees "Ooke'a identification of the word according to its capability of being pronounced alone as well as the presence of a lIlainstress ••• as a conjunctive technique of word-identification". D. Ziervogel goes on to state that this lIlethodwas critieised by G. P. Lestrade and others. but tbat the conjunctiVe word has come to stay. He adde. "Tortay there ia no linguist of note who regards the word as diSjunctive".' He warns, however, that the 92 problem is the extent to which the word may be so conjun- ctivised. We have hinted above that the problem of which prefix inheres to which stem remains to be solved, which is the crux of the problem of the near boundary of the word. We must now find that it is possible that from the extreme where, It ••• the divider (of words) is the meaning or shade of meaning with which words are invested, not concord ••• nor yet accent It, in which case morphemes being meaningful units could also claim their right as words, and are in fact doing so, a turning-point Occurred in 1927 to the opposite extreme where it was not always known how much not to prefix. Is this why for instance a Relative (a word) can, strictly speaking, that is, when its 'prefixes' are credited with their own standing and identity, be viewed as a clause? The question arises too, to what extent linguistic forms are going to be classified more than once as forms, that is, as distinct from classification of their functions. Back to D. Ziervogel, he sketches his own "new approach" to the wo!'d in his A Grammar of Swazi (1952). His criterion is mO!'phology and not quite phonetics as Doke had previously done. He rejects penultimate stress or length as a criterion. He finds that there is some stress on the root-syllable of a word and therefore a word .. ff' If there were 1S bas1cally a root plus pre- and su 1xes. Some indication of the limit to prefixes and suffixes, especially the former, this approach would have been very plausible. In 1959 D. Ziervogel takes the matter further (Northern Transvaal Ndebele) and adds to root_syllable stress the criteria of function, prosodic factors and t. al definition, meaning. We suppose that to the runc 10n • It he If ••• a word which qualifies a SUbl!ltant1ve• •• ( -be odic features maJ Would add morphological features, pros d the 11 ble_length), an also the self-same penultimate BY a db • • • ThilS waul • semantic features of a Qualit1cat1ve. h .• .l!e {indlS f.ul t wi t Very comprehensive definit10n. c. M. Doke's single, phonetic criterion of penultimate stress and prefers more than one criterion. Unluckily he still includes the criterion of function. In our view the criteria of form and meaning on the one hand, and form, meaning and function on the other, produce different results, so that there must be a limit to our criteria too. By the criteria of form and meaning the word motho (person) is classed as a noun on grounds of:- class-prefix ~ stem/root -tho meaning person By the criteria of form, meaning and function the classi- fication is different, that is:- motho 0 ja bogobe (a person eats porridge) motho is now a syntactical component, viz. subject or-Bentence, still referred to as a noun. bogobe bo jewa ke ~ (porridge is eaten by a person) motho is, once again, a syntactical component, viz. extension of predicate, now seen as adverb ke motho. D. Ziervogel also recounts E.B. van Wyk's idea of a word in Zulu and in Northern Sotho. E.B. van Wyk is stated as having declared himself opposed to C.M. Doke's conception of a word, and espousing Reichling's. He bases his word-identification on syntactical criteria such as separability, transposition, isolationability. Once again, D. Ziervogel is not satisfied with a single syntactic criterion. He urges that the three approaches of C.M. Doke, B.B. van Wyk and his own be merged, which in fact we think he tries, but without Success since he includes function among his criteria and therefore still identifies a Relative (which is a clause), and a Copula- tive (which is the pred~cative USE of words other than verbs). To us the very definition of a Copulative, "•••• non-verbal predicatives ••• formed in Tswana by direct prefixal inflection of sUbstantives, qualificatives and adverbs,,,4 is syntactical. It classifies a usage of words already classified. That these words are inflected with prefixes that do not inhere to them or diminish the properties on grounds of which these words were originally classified is abundantly apparent and intensifies the difficulty of determining the near boundary of a word. Malcolm Guthrie criticises the whole exercise of word-division in Bantu for this very reason. He finds that what was to have been word-division turned out to be an exercise in word-fusion, which means that instead of determining where to divide words, rather, what had to be affixed to the stem was sought. He also employs syntacti- cal criteria to identify his word, similar to those of B. B. van Wyk above, mainly isolationabi1ity, separability, replacability, transposition • . One simple example may illustrate this word- fusion allegation. Take the word-group:- dibaga tse dintle - fine beads It is generally agreed that the Adjective is tee dintle, conjunctively written tsedintle. The segment is divisible into concord ~ + 2i. and adjectival stem -ntle. The ~_ is described as the demonstrative element of the concord, and the di- as the class-prefixal element, the latter by A. J. Wookey too.5 We may extend tse dintle to tse dintle tse, conjunctively tsedintle tse. Currently the first Demonstrative is prefixed by conjunctivists and declared a formative, and the second one separated and declared a word. In addition to this inconsistency, the first Demonstrative is separable in a construction such as tse di leng dintle (which are fine) or tse tota di felang di Ie dintle (whieh indeed are absolutely fine). We see no ground for viewing the .. f~rst tse as anyth1ng else bu t a thoro ugh going Demonstra- . -- ... 1 d independent a word as t~ve funct~on~ng concord~a ly an as the ~ at the end of:- 95 tse di leng dintle ~ (conjunctively tsedileng- dintle tse) • There is an abundantly clear difference between the two elements of the concordial structure, tse di, 'namely that only the di inheres to the adjective stem as is true of a noun-class-prefix. Finally, the first ele- ment ~, asserts its features as Demonstrative in that it may be uttered and replaced by tseno, tseo, tsele, (these here hardby, those there, those yonder) in the same con- struction. We would therefore view an Adjective, like a Noun, as class-prefix plus adjectival stem (form) meaning some quality of a Substantive, irrespective of function. Any other word meaning quality of a Substantive would not have an adjectival stem. There is no end to differences of opinion about a word. A. C. Nkabinde6 cites J. A. Louw's approach which he regards as geared to the conjunctive word too. Perhaps the trouble arises from seeking the conjunctive word instead of the word, resulting in 'word-fusion' instead of 'word division' or at the other extreme seeking the disjunctive word and ending up with formatives as words. In any case, J. A. Louw is said to demarcate words by the criterion of word juncture, that is, when two juxtaposed VOwels do not coalesce as in ubaba uya- bingelela (father greets). The kind of elision between two such words - ubab' uyabin~elela - also points to the tar boundary of the first and the near boundary of the second word. The merit of J. A. Louw's approach is that it tries to determine the near and the far boundary of the word. But it does not provide an answer to the pro- blem of the fusion of n~a + umkhonto (with + the spear) into ngornkhonto which yet leaves mkhonto with its proper- ties as a noun intact. Is ngomkhonto now an adverb even though mkhonto i8 qualified and further inflected as a 96 noun as if nothing has happened? The further merit of J. A. Louw's approach is that it seeks word-boundaries in NATURAL speech and not in grammar. The problem of word-identification still besets languages like English. An authority says, "In any language, some lexical units seem to be more word-like than others, and between languages there is no necessary parallelism. For example, because of the way 'the' behaves in English it is hard to call it a 'word' in the same clear sense as 'children' is a word. On the other hand, it is certainly more word-like than the '-ren' of 'Children'''? Without suggesting that African languages do as English, it seems there are similar features of degrees of word-likeness and somewhere both conjunctivisM and disjunctivisM could take account of this. The idea is held that a Bantu word must be capable of being spoken alone, a very perfect degree of isolationability. Is there a language in which every word is isolationable to this extent? To our mind the opposite pole to declaring formatives words indiscriminately, is declaring a clause a word. It appears that a search lor a middle way can be fruitfUlly attempted by reference to NATURAL SPEECH. Currently the Tswana Qualificative is analysed as follows, the Qualificative given between square brackets:- e.g. with it by an (8 fine child), disjunctively : ngwana conjunctively : n~ana concord '$.0 mo / ,omo- i1) the Enumerative _ 8 word which qualifies a substantive and is brought into agreement with it by an enumerative concord, e.g. 9? disjunctively letsatsi Lie len~wi7(another day) conjunctively letsatsi Lielengwi7 concord Ie le- /lele- i11) the Quantitative - a word which qualifies a substantive and is brought into agreement with it by a quantitative concord, e.g. disJ"unctivelyb w~a~s~e~ts~a=n=a~~_b~a~b_o_t_l_~~ (all the - girls) conjunctively basetsana Lbabotl~ concord ba bo- /babo- iv) the Possessive - a word which qualifies a substantive and is brought into agreement with it by a possessive concord, e.•g. disjunctively selepe LBa~a~(your axe, sin~) selepe Lfla lo~(plural) conjunctively selepe Lflaga~(singular) sele~e L;alo~(plural) concord v) the Relative - a word which qualifies a subs- tantive and is brought into agreement with it by a relative concord, e.g. disjunctively kgomo La e gangwani7(a cow that is milked/a milk cow) conjunctively kgomo /jegangwanil concord e e /~- When used in isolation these qualificatives are declared qualificative pronouns. Many problems arise: whether each of the above qualificatives is in fact a word, having heard already that the Relative is really a clause? Whether when spoken in isolation and each is now reclassified as a qualificative pronoun, we are at this stage classifying a form that means quality of a substantive or declaring their substantival function, in which case whether we are not doing syntax in the name of word-classification? Whether concord as such is a morphological or syntactical 98 requirement, therefore whether a form identifiable by a concord is a morphological or syntactical unit? Whether qualificatives can only be identified when in the functio- nal company of substantives and concords, that is, when not spoken alone, in which case they would not qualify as words if they cannot be spoken alone? Whether these con- cords are determined by the qualificative or the substan- tive and therefore whether they belong to either or both and to what extent? What actually constitute these con- cords in point of form and meaning, irrespective of the concordial function? Whether or not these concords are classifiable differently on their own steam as forms with meaning? We do not think that it is absolutely necessary to answer all the questions above in detail in respect of each of the Qualificatives. Since we have already heard that the Relative is in fact a clause we make it the guinea-pig. In the sentence ntate 0 rekile kgomo e e gangwang (my father has bought a cow that is milked/a milk cow), the form e e gangwang (conjunctively eegangwang):(that is milked) is the Relative Qualifica- tive, in isolation the Relative Qualificative Pronoun, otherwise the Relative Qualificative Clause. Sentence analysis would be something of this sort: a) DETAIL~D ANALYSIS r-----.,,--------=-------;,:--------, t t I Enlargement Subject Predicate Object ,, of object t , . when the Qualificative is in • company of the Substantive it qualifies, and therefore 'a word that qualifies;: I I ,•• kgomo : e e gangwang ntate ; 0 rekile : , t is in i~olation and.now a when the Qualificative Pronoun, a substant1ve Relative Qualificat}ve and Still a word: : ,, t :e e gangw-:. ----- I ' ntate ; 0 rekile , I ang : , I t , ' 99 We find that if the Relative is viewed as a vord, it performs two functions here. qualificative and substantival. It has other functions - with the aid of an adverbial formative it can function adverbially. With the copula a predicative function is performed. Its functions are a study in syntax. If the Relative is viewed as a clause the analysis would be as follows: b} GENERAL ANALYSIS Clause Kind and Relation a. Ntate 0 rekile kgomo Main Clause E e gangwang Relative Qualificative Clause, qualifies object keomo in main clause a. We must teach that the same Relative Qualifica- tive, a word, functions as enlargement of object. as object. a word again. and as a qualificative clause. If this is wrong we mUst teach that the same Relative Qualiticative clause functions as qualificative clause, as a substantive (8. word) and as an enlargement of a substantive (a word). How the qualificative word can be a clause simultaneously-or the qualificative clause a word simultaneously defies anybody's conception of a word a~d a clause. We are not yet at the stage of stating which we think it is. Other factors should deter- mine how many words constitute the form e e gangwan~/ ••gangwan~. When we say that this form is a Relative Qualtficative Pronoun we are in fact saying VOlubly that it functions SUbstantivally when in iSOlation, and are still in the field of syntax, the classification of functions and not the classification of words. We will return to the utter. • 100 Is concord a morphological or syntactical requirement? Concord means agreement, the opposite of discord or disagreement. This is never a feature of one pole. Concord/discord presupposes two poles. In our case there must be two distinct poles, the Substantive and the Relative Qualificative, in functional discord in the absence of their concord, incapable of function until the concord is introduced, yet identified severally as words, thus:- kgomo ~angwang cow being milked Admittedly gangwang commands a lesser degree of iso1ation~ ability or word-likeness than kgomo. This gangwang can be used in different ways, thus:- .&0 gangwang? what is being milked? go gangwanei? what being milked? df gan!1;wang? what is being milked/tapped out of them (those huge trees)? di" gangwang? : what being milked out of them? Depending on the class of the noun qualified, in this case kgomo, n-/din- class, singular, the concord we require in the gap above is ~ (which is), which we require not for the identification or even classification of ~angwang but for its usage for purposes of qualifying kgomo. As seen above, for other usages we also select other forms, ~. di~ gan~wang already standing identified. These are all concordial forms replacing each other according to the meaning intended. We come to the con- . t We do elUsion that concord is a syntactical requ1remen • not know which concord to select when we have ~an~wang only. so that these concords do not inhere to the sO- determinant sub- called relative stem unless we know the t ting a clause. We stantive qualified and we are cons rue f d ur ther down. Suffice discuss the nature of the conco~. 101 it to add that gangwang is one of the forms of the lexeme gama LXb.: m£J( to milk). And the forms of this lexeme are legion and are all verb-stems with inflections that are semantic twists always seen in the light of the mother lexeme. This form of this lexeme and those of other verbal lexemes can be used to qualify a substantive in addition to other usages. When e e gan~wang is used in isolation, the concord betrays the noun-class of the substantive referred to. One would not know which qualificative or other concord to employ unless one had a particular noun or state in mind. Speech is by suppression, expression, impression. Strictly speaking, it is only to the eye and to the ear that any of the Qualificatives above, with its concord, is used in isolation. The construction e e -gangwang has the semantic consequence of n-/din- class-noun, singular + e e gangwang to the hearer, and this noun is never absent in the speaker's situation, or the speaker would not be able to select this concord. This noun is concord-determinant. A form identifiable by a concord is therefore a syntacti- cal unit in our view. In semantic terms therefore there is no isolation of a ~ualificative, with concord, from the substantive qualified. This substantive is consci- ously and deliberately suppressed as a matter of usage, but consciously held in the speaker's situation. In mathematical am trying to convey; A 102 Utterance heard/read A ~~ C C A The hearer lives himself into the speaker's situation and is led thereto by the intersection of the sets. The hearer is sure that the speaker had an n-class noun in mind,and reckons with this situation. The usage appears to us to be a matter of idiom and not word- bUilding. One would therefore not speak of a Pronoun but a Noun Clause, when referring to C above. This is where NATURAL speech is primary to the written language. Coming now to the nature of the concord we may learn from A. J. Wookey who identified Proper Adjectives and Verbs, Nouns and Adverbs ~ adjectivally. He would have seen gangwang in e e gangwan$ as a verb used adjectivally. His mistaken reference to the first ~ of e e ~angwang as Relative Pronoun (translated 'whiCh') is significant as pointing out in effect that there exists a word-boundary after this form. Earlier on we showed that this form takes the usual inflections of a Demonstra- tive in the normal way:- e e gangwanL : (which is milked) eo e gangwang : (which there is milked) ele e gangwang : (which yonder is milked) It is also perfectly separable as in~- e tota e ~anRwan~ : (which indeed is milked) 103 We believe that it is this consideration that accounts for classifying this form both as a Relative (a word) and as a Relative Qualificative Clause (a string of words). Wookey would have classified it only as a string. We support this view, especially where gan~wan~ commands negative conjugation as a verb normally does:- e e sa gangwan~ : (which does not get milked) commands tense formation:- e e tla gangwang : (which will get milked _ future) and oommands modal formation:- e e ka gangwang (which can be milked - potential). From the above then e e gangwang would be a olause and not a word and there would be eventually no word-olass oalled Relative Qualificative in Tswana. The detailed analysis of the clause woutd be: Subject Predicate E (Demonstrative fun- otioning as con- e gangwang junctive) (Yona) SUbjeot proper understood The second element of the concordial segment e e, namely !, remains subject concord. In faot, it is the only conoord, the first element being a Demonstrative of the olass of the substantive qualified. The formula for the Relative Qualificative Clause above would therefore be: Demonstrative of the class and number of substantive qua- lified + SUbject (often suppressed/understood) + SUbject ooncord of class and number of noun qualified (suppressed/understood) 104 + Conjugational and/or tense or modal formative elements (often omitted) + Verb-stem plus passive morpheme ~ ~lus participal formative -~ In the orthography the Demonstrative would stand separa- ted. There is the argument of separability by conjuga- tional and tense formative elements from the verb-stem regarding the subject concord, as well as our earlier idea that the concord, being a syntactical requirement, does not necessarily inhere to the forms of the lexeme ~ama, even more so when the Demonstrative is separated. This is why we would have no objection to this form being treated as one of the not-so-word-like words in orthography. and at the same time would sympathise if orthographers decided to join it to succeeding forms. Orthography is practical. We are aware that the arguments above will not suit a Nguni language just as well, but this need not be so. The trouble is where it was made to look as if what suited Nguni always suited Sotho. Having aligned ourselves with the stand that there is only a Relative Clause in Tswana, which may function qualificatively or substantivallY or predicati- vely or adv~rbially, and therefore no Relative Qualifica- tive or Relative Qualificative Pronoun as word_classes, we briefly consider the rest of the Qualificative types. In our view the remaining four .Qualificatives t form There are fall into two categories according 0 • . . bielemental and the those three whose concord1al form 1S e element, thus:- one whose corresponding form has on A Adjective ngwana LYo montli? - a ~ child Enumerative letsatsi /5.e lengw.i7 - another day Quantitative: basetsana jja botlh~ - all the girls B Possessive selepe 5a mil - !SL axe Starting from the bottom the Possessive is to us a form which means possession. It has primitive forms which are not necessarily qualificative and deserve to be treated simply as Possessives, their qualificative, descriptive, predicative, substantive, interjective func- tions being a separate study in syntax. The primitive forms are:- Singular Plural First Person melka (my) ~ (our) Second Person ~ (your) eno (your) Third Person gagwe(his/hers) abo (their) The other noun-classes use other word-classes, mainly the Absolute Pronoun. The plural forms have a rather subtle interpre- tation, really meaning:_ ~ of my (singUlar) family or clan or kraal ~ of your (singular) family or clan or kraal abo of his (sin~ular) family or clan or kraal If intended to convey the idea of 'our' family ••• the Absolute Pronoun takes the place of the Posse- ssive, thus:- selepe sa ~ our (plural) axe selepe sa lona your (plural) axe selepe sa bona their axe or selepe sa gabo bona : axe of their household, etc. An important point of principle arises here : whether the Absolute Pronoun thus functions in a posse- 106 ssive construction it is to be classified as a Possessive. But for the fact that it is prece1ed by a possessive concord it retains all its properties as a Substantive, as an ABSOLUTE and UNQUESTIONABLE pronoun. The posse- ssive concord is determined by and is proper to the preceding substantive, the possessee. It operates between two poles and is simply a matter of orthography (conjunctive or disjunctive) and a matter of whether a word must always be capable of being spoken alone, whether or not the concord must be separated from or joined to the possessive stern or other stem used in a possessive construction. This is important since our primitive forms above would be viewed as stems. We accept this analysis. We find it better than defining a word in terms of its concord, which makes it syntactic. It is the problem of the near boundary of a word, which must not be so contrived as to make it a word-group or clause. In any event if when the Absolute Pronoun is used in a possessive construction it becomes a possessive, where do we end in re-classifying it? As a result of such tendency to re-classify a form ad infinitum there is a Sotho school grammar which states that the possessive and the Absolute Pronoun are the genuine Southern Sotho Pronouns. Only because the Absolute Pronoun functions in possessive constructions it would be viewed as a Possessive, and since it retains the properties of a Pronoun intact in this construction, it is a genuine it pronoun too. 8 Why is it not the cenuine adverb when functions in a nescriptive construction? In still higher e v,. Mofokeng's places it happens where C. M. Doke an d ~. ,., 9 ,.. t .tative' scheme of Qualificatives excludes the ~uan 1 , 10 . eludes the D. T. Cole's corresponding scheme 1n 11. 1 des the Quantitative Quantitative; and S. M. Guma's 1nC u . as well as the Demonstrative, which latter category 1S e two genuine pronouns. generally regarded as one 0f th 107 For as long as function is the touchstone for word-classification, all the above scholars are right. The differences of classification are the logical result of the approach to word-classification, the criterion of function. We would like to see the Quantitative, the Adjective, the Enumerative simply as Quantitative, Adjec- tive, Enumerative, forms operating in most instances with noun-class-prefix plus stem which means quality, a count, quantity, respectively, and which, in another study, ALL function adjectivally, sUbstantivally, adverbially, predicatively, and otherically. The term Qualificative is understood as classifying function and not words. as classifying a function which can be performed by a word, a word-group, a clause, and a word of practically all word-classes for that matter. As to the nature of the concords of the Adjec- tive, the Enumerative, the Quantitative the first elemen~ when expressed, is demonstrative and the second class- prefixal, with minor exceptions. This study needs no further elucidation. Hence we have written examples of them under A above as two words each. It is also not found absolutely necessary to set out the word-classes under review fully according to the noun-class system as the aim of this paper is to highlirht problems. We would finally like to close with a paragraph on yet another problem which we think has a solution in NATURAL speech rather than in grammar. The Adjective compares with the Noun not only by the inherent class- prefixal inflexion, but by the diminutive transformation too. As a form of endearment, the 'fine one' is yo montlenyanejyo montlane. The 'black one' may be trans- lated ~o montshonyana (male) mOswana (female). The diminutive formation is intriguing. witness:_ 108 montlana (. montle moswana tiC.. motsho < montsho to s speaks for itself. The dissimilation of ~ In both cases the terminative vowel has dis- appeared. It is not grammar which so decrees, but NATURAL speech. In the case where the terminative was a front vowel (-e) the replacement is -~. In the case where the terminative was a back vowel the replacement is -wana. In the latter case the terminative -0 and the initial vowel of suffix _~ have disappeared during NATURAL speech. A sound-change has taken place. It does not take place in NATURAL speech, after the root-consonant -~. In fact it takes place before this consonant, after the prefixal syllable ~_. Our Adjective could in fact have been written mo _ wsana just as well. There would still be no _~ in NATURAL speech in spite of what the eye saw. This is because there is lip-rounding which spells the sound-change, before the root-consonant -~. To say the d diminutive suffix is _~ amounts to