RELATIVIZATION OR NOMINALIZATION?: A CASE OF STRUCTURE VERSUS MEANING

Ayo Bamgbose[1]

1. INTRODUCTION

In Bamgboṣe (1975), I tried to show "that not all clauses introduced by ti in Yoruba are to be regarded as relative clauses". This conclusion was arrived at after due consideration of the meaning of such clauses some of which have the same meaning as factive or manner nominalizations. Such was the 'novelty' of the conclusion that Awobuluyi felt obliged to write two papers to demonstrate how 'false' it is, and to re-assert the traditional position that ti always introduces a relative clause.

The purpose of this paper[2] is to examine, in the light of the points raised in these papers, the implications of the two positions. A convenient starting point, it seems to me, is a statement of the facts as they are, before going into the question of how these facts are to be accounted for in any analysis.

2. THE CONSTRUCTION TYPES IN QUESTION

Consider the following basic sentence:

(1) Ó ńlu ìlù

‘He is beating a drum’

This basic sentence must be included in the underlying structure of the following noun phrases:

(i) NP’s with Relative Clause meaning

(2) a. Ìlù tí ó ńlù

drum that he beating

‘The drum that he is beating’

b. ?lílù tí ó ńlu ìlù

beating that he beating drum

‘The beating that he is beating drum’

(ii) NP’s with Factive Meaning

(3)a ìlù tí ó ńlù

'The fact that he is beating a drum'

b. lílù tí ó ńlu ìlù

The fact that he is beating a drum'

c. pé ó ńlu ìlù

that he beating drum

‘The fact that he is beating a drum’

d. bí ó ṣe ńlu ìlù

‘The fact that he is beating a drum’

(iii) NP’s with Manner Meaning

(4)a lílù tí ó ń lu ìlù

‘the manner in which he is beating a drum’

b. bí ó ṣe ńlu ìlù

‘The manner in which he is beating a drum’

Note that (2a) and (3a) are identical in form and that the initial word in them is the noun object of the original sentence (1). Similarly ( 3b ) and (4a) are identical in form and the initial word in them is a nominalized form of the verb in the same sentence. If a basic sentence (5) containing an adverb is selected instead of (1), it can be shown that the adverb can be transposed to become the initial word of the corresponding noun phrase i.e.

(5) Ó ńlu ìlù lemọ́lemọ́

‘He is beating a drum insistently'

(6)a. lemọ́lemọ́ ti ó ńlu ìlù

"The fact that he is beating a drum insistently'

b. lemọ́lemọ́ ti ó ńlu ìlù

'The way in which he is beating a drum insistently'

Again, (6a) is identical in form with (6b), hence the NP can have a factive as well as a manner meaning.

In terms of the surface structures of the above NP’s and their meanings, the following may be given as a summary:

A. NP’s with tí- Clause

1. N tí… (a) relative e.g. (2a)

(b factive e.g. (3a)

2. V^NOM tí… (a) ? relative e.g. (2b)

(b) factive e.g. (3b)

3. Adv. tí… (a) factive e.g. (6a)

(b) manner e.g. (6b)

B. NP’s without tí Clause

1. pé… factive e.g. (3c)

2. bí… ṣe… facitve e.g. (3d)

manner e.g. (4b)

The above are the facts about the constructions in question and their meanings. The only disputed case is (2b) which Awobuluyi holds to have a relative meaning but which I maintain has only a factive or a manner meaning as in (3b) and (4a) respectively. We shall return to this point later.

The three types of meaning indicated above are self-explanatory. The relative meaning is associated with a restrictive relative clause construction and what it does is to provide further specification about the noun it qualifies. Thus (2a) specifies a particular kind of drum i.e. the one which the subject of the sentence is beating, and to the extent that any relative meaning may be ascribed to (2b), it should specify a particular kind of beating.[3] The factive meaning refers to the fact of doing something or being in a certain state. Hence all the NP's of (3) and (6a) are factive. The manner meaning refers to the manner of doing something. Hence NP's of (4) and (6b) are manner noun phrases.

3. THE TWO ANALYSES COMPARED

The problem posed for any analysis of these NP's is how to account for the three types of meaning which are associated with the ti clauses and how to relate them to similar meanings borne by non-tí clauses such as (3c) , (3d) and (4b).The two analyses under review recognize the need to group the NP's concerned and their meanings under construction types: nominalization and relativization The essential difference between the analyses lies in the range of meanings covered by each construction. If can refer to the two approaches as the A-analysis and the B-analysis respectively, the meanings covered c-shown as follows:

A – analysis

Nominalization Relativization

Factive meaning Relative meaning

Factive meaning

Manner meaning

B- analysis

Nominalization Relativization

Factive meaning Relative meaning

Manner meaning

3.1. Problems for the A-Analysis

The A-analysis predicts that the same NP containing the same relative clause can have as many as three different meanings e.g. a. relative meaning as in (2b), a factive meaning as in (3b), and a manner meaning as in (4a). It further states that one of these meanings happens to be identical with the factive meaning of a nominalization. e.g. (3b) compared with (3c) and (3d).

The two problems inevitably arise with the A-analysis. Firstly, how can we account for the three different meanings of a relative clause? and secondly, how can we relate the factive meaning of a relative clause to the factive meaning of a nominalization?

The answer provided to the first question is that this multiple meaning is a case of lexical ambiguity which is to be traced to the reduplicated nominalized form of the verb i.e. since the head of the relative construction is a nominalized form Ci-Verb[4], the ambiguity resides in this form. It seems to me that this kind of explanation for the ambiguity in meaning is unsatisfactory for three reasons:

(i) It does not really explain the basis for the ambiguity since there is absolutely nothing in the reduplicated nominalized form to provide evidence for the ambiguity. Why should the mere fact of a verb being nominalized by partial reduplication lead to three different meanings if the difference does not readily exist in the verb concerned? In any case, why should nominalisation by reduplication be unique in this respect?

(ii) Even if the ambiguity of the NP can be traced to the Ci-Verb head, there will be no explanation at all for the similar ambiguity of NP's such as (2a) and (3a) or ( 6a ) and ( 6b) whose 'heads' are a noun and an adverb respectively which do not involve any form of nominalization[5].

(iii) The lexical ambiguity hypothesis fails to account for the fact that both nominalization and relativization share a factive meaning.

The answer provided to the second question by the A-analysis is quite revealing. It states that a relative clause with a factive meaning is derived from a factive nominalization. Thus (7a) is derived from (7b)[6].

(7) a. wíwá ti ó wá

coming that he come '

The fact that he came'.

(7) b. pe ó wá

that he come

The fact that he came'

This in effect is a recognition that what is now called a relative clause in (7a) was at some point in its derivation a nominalization. Hence, the factive meaning which appears on the right-hand column of the A-analysis may be deleted since it derives from the same meaning under 'Nominalization' in the left-hand column. The implication of this is that the rigid claim that clauses introduced by the relative particle ti cannot be nominalization will have to be abandoned. At least the relative clause with factive meanings are nominalizations in their underlying structure. Secondly, the lexical ambiguity hypothesis will also have to be abandoned, since the ambiguity between factive and other meanings can be accounted for in terms of different underlying structure.

At this point. there are now only two differences left between the A-analysis and the B-analysis. The first concerns the treatment of the manner meaning and the second, the disputed relative meaning of NP's such as (2b) containing a nominalized form of the verb followed by a ti clause. I will now proceed to examine these two areas of difference.

3.2. Treatment of the Manner Meaning

We saw earlier that NP's containing nominalized verbal forms or adverbs followed by ti clauses are ambiguous in that they could have either a factive or a manner meaning. Hence,

b. lilu tí ó ńlu ìlù

'The fact that he is beating a drum'

(4)a. lilu ti ó ń lu ìlù

'the manner in which he is beating a drum'

(6)a. lemọ́lemọ́ ti ó ńlu ìlù

'The fact that he is beating a drum insistently'

b. lemọ́lemọ́ ti ọ́ ńlu ìlù

"The way in which he is beating a drum insistently'

The factive meaning, as we have seen, derives from nominalization: but what is the source of the manner meaning? The A-analysis maintains that the manner meaning is one of the meanings of relative clause, while the B-analysis states that there is a manner nominalization from which this manner meaning derive. One variant of this manner nominalization is represented by (4b) and (8) which are exact paraphrases of (4a) and (6b) respectively:

(4)b. bí ó ṣe nlu ìlù

‘The manner in which he is beating a drum’

(8) Lemọ́lemọ́ ti ó nlu ìlù

'The way in which he is beating a drum insistently'

According to the A-analysis, (4b) and (8) are relative clauses even though they do not contain the relative particle ti which is supposed to be an unfailing indication that a clause is a relative clause. Apart from this, how do we relate the meaning of (4a) and (6b) to that of (4a) and (8) respectively? As far as I can see, the A-analysis does not have any answer to this question since each of the NP's undergoes an independent process of relativization. Whatever similarity of meaning results is simply a changed one. Finally, alloting the manner meaning to a relative clause means that such a clause will still have two different types of meaning (a relative and a manner meaning), the basis of which cannot be traced to any structural difference. What is even worse, although all NP's having simple nouns as their head nouns will have a relative meaning (for example, as in (2a)), an NP having the noun bí as head (as in (4b) and (8)) will have only a manner meaning[7].In view of the unsatisfactory aspects of this analysis as outlined above, the B-analysis is believed to be preferable, since, under it, all manner meanings will be associated with a manner nominalization.

3.3. Is there a Relative Meaning for the Nominalized Verbal form?

According to the B-analysis, an NP such as (9) has only two meanings, a factive meaning and a manner meaning:

(9) lilọ tí ó lọ

Going that he go

a. ‘The fact that he went’

b. ‘The manner in which he went’

The A-analysis, on the other hand, contends that (9) is three-way ambiguous i.e. in addition to the factive and manner meanings, it also has a relative meaning which may be glossed as (c) 'The trip which he made'. It is significant that this claim is only made in the most recent account of this approach and that hitherto, only the factive and manner meanings have been accepted by the author of the A-analysis. For example, in Awobuluyi (1973), it is unambiguously stated that

Yoruba nouns formed from verbs through partial reduplication typically have two meanings each of which one conveys the fact, and the other the manner of performing the action named by the verb concerned[8].

Similarly, in Awobuluyi (1975a), it is specifically stated that sentence (10) has only (11a) and (11b) as synonyms:

(10) lílọ tí ó lọ dára

going that he go good

(11) pé ó lọ dára

‘The fact that he went is good’

b. bí ó ti lọ dára

‘The fact that he went is good’

‘The way in which he went is good

Again, nowhere is any gloss like (c) given. One can conclude, therefore, that the author of the A-analysis has just "discovered" that an NP such as (9) can have a relative meaning. Now, the question is: how valid is this discovery? If (9) has the meaning 'The trip that he made', one would expect (12a) to be synonymous with (12b), and (13a) with (13b).

(12)a. lílọ tí ó lọ yà mi lẹ́nu

The fact that he went surprised me

b. ìrìn-àjò tí ó lọ yà mi lẹ́nu

trip that he go open me in mouth

‘The trip that he made surprised me’

(13)a. ó gbádùn lílọ tí ó lọ

He enjoyed the fact that he went

He enjoyed the manner in which he went

b. ó gbádùn ìrìn-àjò tí ó lọ

‘He enjoyed the manner in which he went

But this expectation is not fulfilled, since the (b) sentences refer to a trip, while the (a) sentences refer only to the fact or manner of going. The logical conclusion, therefore, is that (9) does not have a relative reading, and any attempt to impose such meaning on it is best seen as a way of providing spurious support for the analysis.

One further evidence that what is involved in sentences such as (9) is not a relative meaning is provided by sentences such as:

(14)a. ó bá mi lọ

he with me go '

He went with me'

b. bíbá tí ó bá mi lọ

? that he with me go

'The fact that he went with me'

'The manner that he went with me'

c. lílọ tí ó bá mi lọ

'The fact that he went with me'

'The manner in which he went with me'.

(15)a. ó ba dùrù jẹ́

'He spoiled an organ'

b. bíbà ti ó ba dùrù jẹ́

'The fact that he spoiled an organ'

'The manner in which he spoiled an organ'

In (14a) there is a preverbal item bá 'with' which can only occur with a main verb such as lọ'go'. In (14b) and (14c), both items are supposedly relativized, yet there is no relative meaning. What is more, (14b) and (14c) are synonymous, and this would certainly not have been possible if there were two different noun heads lílọ trip' and bíbá (meaningless in isolation!) being qualified by the relative clause. Finally, no conceivable relative meaning can be provided for (14b), since its so-called head is meaningless if interpreted as a noun. (15a) and (15b) further demonstrate the absurdity of imposing a relative meaning on these NP's. In (15a), the verb bàjẹ́ ‘spoil' consists of two parts (ba and jẹ́) either of which is meaningless by itself. Yet we see one of these meaningless parts bà being 'relativized as bíbá in (15b). Since no meaning can be given to this so-called relativized form, it follows that a relative meaning is simply out of the question for (15b). On the other hand, both the factive and manner meanings are possible for the NP in question which is really a nominalized sentence.

There is, however, a valid example given in Awobuluyi (1975b) which may look like a counterexample to the claim that no relative meaning is involved in sentences like (9). This example is (16a) which has the inelegant variant (16b)[9].

(16)a. lílọ kù tí o máa bá mi lọ

going remain that you will help me go

'There are some more errands for you to

run for me'.

b. lílọ ti o maa ba mi lo kù

'there are some more errands for you to

run for me'

Since the NP in (16b) looks very much like the one in (9), and since neither a factive nor a manner meaning can be given for it, the conclusion is drawn that (16) justifies the assignment of a relative meaning to NP's having a nominalized form of the verb followed by a tí clause.

A closer examination, however, reveals that (16a) must have been derived from

(17) líkọ kù; o máa bá mi lọ lílo

‘Errand remains: you are going to run

errand for me

In (17), there are two sentences. The first one has its subject the noun lílọ 'errand'. This same noun is object in the second sentence. The construction in (16) is a non-restrictive relative clause whose source is typically two sentences between which the relative particle tí is inserted. Compare this with the derivation of a factive nominalization where the nominalized form of the verb does not already exist in the underlying structure but is brought into being by promotion or copying of the verb in the sentence. Thus (9) is derived from (18a), and not from (18b)

(9) lílọ tí ó lọ

Going that he go

a. ‘the fact that he went’

b. ‘the manner in which he went’

(18)a. ó lọ

‘He went’

b. Ó lọ lílọ

‘He went going’

The sentence in (17) may be compared with those in (19a) from which the non-restrictive relative clause (19b) can also be derived.

(19)a. ìwé ‘kù; o máa bá mi ra ìwé

‘Book remain: you are going to help me

buy books’

b. ìwé kù tí o máa bá mi rà

‘There are more books that you are going to help

me to buy’.

When (17) is compared with (19a), we can see that lílọ ‘errand’ behaves just like a simple noun ìwé ‘book’ and that in both cases, the noun which is supposed to be relativized is already present in the underlying structure. The obvious conclusion that can be drawn from this is that in sentences such as (16), the head of the relative clause is a noun, and it should not be confused with the nominalized form of the verb in a factive or manner nominalization such as (9).

Another instance where a form such as lílọ is to be analysed as a noun is where it is qualified by a qualifier in the underlying structure. A good example of this is an NP such as (20a):

(20)a. lílọ òsì kan ti ó lọ

going poverty one that he go

'A certain poor way of going that he went'

b. ó lo ni lílọ òsì kan

'He went in a certain poor way'

which is clearly derived from (20b ) by the relativization of the NP lílọ òsì kan ‘a certain poor way of going'. Here again, one can see that the NP in question already contains a noun lílọ and that (20a) is simply not (9) with qualifiers inserted as has been erroneously suggested.

To summarize, we can say that there are some nouns which obviously are originally formed from verbs but which have acquired specialized meanings. An example of such a noun is lílọ when used to mean 'errand' as in (16) and (17) or síṣe in the sense of a ‘ceremony’ (cf. the verb ṣe 'do') which can be used in the place of lílọ in (16a) to mean 'There are some more ceremonies that you are going to .join me in performing'. In addition to such specialized senses, the same forms of the noun occur in their ordinary sense (i.e. in a sense close to the meaning of the verb) when attention is specifically drawn to a particular kind or species of the noun concerned e.g. in (20) where attention is drawn to a particular kind of going. It is in such cases that relativization can take place just as with any other noun.

4. CRITICISMS OF THE B-ANALYSIS EXAMINED

At this point, I would like to turn to the point made by Awobuluyi (1975b) in his criticism of the B-analysis and his support of the A-analysis. These criticisms fall into three broad categories

(a) Criticism based on the mistaken identity of the nominalized verbal form.

(b) Criticisms arising from a disregard of meaning

4.1. Mistaken Identity of the Nominalized Verbal For:

Criticisms under this category are:

(i) lílọ has a meaning 'going' and it can be qualified to give the meaning 'a type of going' as in (20a)

(ii) Contrary to the position taken by Bamgbose (1975), the relative clause in (9) can be questioned .wo 'which?'

(iii) Under certain circumstances, (16b) can be extraposed as (16a) just like any other NP with relative clause.

(iv) Since (16a) does not have a factive or manner meaning, it follows that the only other possible meaning it can have is a relative one.

These criticisms can be shown to be based on a failure to distinguish between the nominalized verbal form that arises from a sentential nominalization as in (9) and a noun having the same form and already present in underlying structure as in (16) or (20). Hence, where 'going' can be qualified to give the meaning ‘a type of going' as in (20) or where it occurs in a restrictive relative clause and has a non-factive non-manner meaning as in (16), what we are dealing with is not the nominalized verbal form arising from nominalized sentence, but rather a noun of that form. The difference between the two has been discussed exhaustively above. As far as questioning by .wo ‘which?’ is concerned, this is certainly not possible with (9) but it is possible where what turns up as the head of the relative clause is already a noun in the underlying structure. Consider, for example, the following:

(21) a. wọ́n na ọmọ náà ní nínà kan ní àná

they beat child the in beating one in yesterday

‘They gave the child a certain kind of beating yesterday

b. nínà (kan) tí wọ́n na ọmọ náà ....
beating one that they beat child the ...
'The beating that they gave the child

c. nínà wo ni ẹ ńsọ̀rọ̀ rẹ̀

'Which beating are you talking about?'

In (21b), there is a relative clause qualifying a noun nínà 'beating'. This noun is already present in the underlying structure (21a) from which (21b) is derived. Since the reference of the noun is to a particular kind of beating (e.g. a brutal one), it is possible to have (21c) as a questioning of the type of beating, and this can evoke (21b) as a response.

4.2. Disregard of Meaning.

Criticisms under this category are:

(i) Deletion of a relative clause is not possible in all NP's: consistently it cannot be used as a criterion for distinguishing between a true noun and a nominalized verbal form after which such deletion is not acceptable. In any case, given an appropriate context, and ignoring difference in meaning, deletion of a relative clause is even possible after a nominalized verbal form.

(ii) Contrary to the assertion made in Bamgbose (1975), a factive nominalization can occur with a non-impersonal verb.

Both these criticism can be shown to be due to a failure to look closely into the meanings of the constructions concerned. As far as the first criticism is concerned, it is true that some nouns always take qualifiers but that is not the reason why a relative clause cannot be deleted after a nominalized verbal form. To say this would be to identify the form with nouns that must always take qualifiers. And this cannot be correct in view of the example provided to demonstrate that, given an appropriate context, a relative clause may after all be deleted after a nominlized verbal form. Some attention to meaning will show that when a relative clause qualifying a true noun is deleted, the resulting sentence still retains its reference to the original topic. Thus when (22a) is reduced to (22b), there is no change in the topic, but this is not the case with the reduction of (23a) to (23b):

(22)a. ìwé tí ó rà dára

book that he buy good

‘The book which he bought is good’

b. ìwé dára

'The book is good'.

(23)a. lílọ ti ó lọ dara ju kíkun

going that he go good pass grumbling

ti o ńkùn kiri

that you grumbling about

'The fact that he went is better than your grumbling about'.

b. lílọ dára ju kikun kiri

'Going is better than grumbling about'

In (22a), one is talking about a book, the particular book bought by the subject. In (22b), the sentence is still about a book. In this sense, the topic of both sentences remains unchanged in spite of the deletion the relative clause. Compare this with (23a) where the topic is not a particular kind of going but rather the fact of going. When (23a) is reduced to (23b), the topic changes to 'going' in general as may be demonstrated by substituting for lílọ such variants as àti lọ ‘going’, kí ènìyàn lọ 'that someone should go'. The point therefore needs emphasizing that deletion of a relative clause after a true noun does not lead to a change of topic and only rarely to an ungrammatical sentence, while such deletion after a nominalized verbal form usually leads to an ungrammatical sentence and always to a change of topic.

As far as the second criticism under this category is concerned, it can be easily demonstrated from the so-called counterexamples that what occurs with impersonal verb is not a factive but a manner nominalization. Take for example, (24a) which is supposed to be a counterexample to the statement that factive nominalizations do not occur with non-impersonal verbs.

(24)a. lílọ tí ènìyàn máa ń lọ pé mẹ́ta

going that people HABIT, go reach three

'There are three ways by which people go’

b. bí ènìyàn ṣe máa ńlọ pé mẹ́ta

'There are three ways by which people go’

c. *pé ènìyàn máa ńlọ pé mẹ́ta

*’The fact that people go amounts to three’.

A little attention to the meaning of (24a) will show that it is not a factive nominalization. For if it were, it would have a paraphrase (24c) which does not occur. On the other hand, it is identical in meaning with (24b) which is a manner nominalization. Therefore, all that the so-called counterexample proves is that a manner nominalization can occur with a non-impersonal verb, but then no one has said that such a nominalization could not so occur.

4.3. Preoccupation with Surface Structure

Criticisms under this category are:

(i) Relative clauses are different in appearance from nominalizations.

(ii) Several relative clauses can be combined into nesting structures to qualify the same noun head, but "no nominalizations can be combined into a nesting structure in the language".

(iii) There is a semantic constraint in relative clauses such that a noun head must be matched by some ether element in the relative clause. If NP's like (9) are treated as nominalizations, it will be found that "no other types of nominalization are subject to [this] semantic constraint".

(iv) Other qualifiers may be substituted for a tí clause, and its order of occurrence conforms to the general sequence of qualifiers in an NP.

All these criticism are in varying degrees, based on an excessive preoccupation with surface structure considerations. To take first the argument about appearance which is to the effect that sentences and verb phrases are nominalized by the addition, in the initial position of prefixes or introducers, and that if (9) were to be seen as a nominalization, tí which could be regarded as the nominalizing element would be an exception in that it occurs in the middle of the NP. My comment on all this is that appearance is not a reliable guide for making a deep structure analysis which is what the distinction between nominalization and relativization is. In the first place, it is not correct to say that the particle tí is responsible for nominalizing a sentence to yield an NP such as (9). The whole of nominalization involves moving or copying a constituent to the sentence-initial position and adding tí. Secondly, nominalized sentences take different surface forms including addition of introducers, of prefixes, or even not adding any formative Considering this divergence, why should a particular type of nominalization be considered unusual? In any case, the A-analysis accepts that in their factive meaning, NP's like (9) are derived from nominalizations. This being the case, the argument about appearance becomes meaningless. The futility of the appearance argument is best demonstrated by:

(14) bí ó se ńlu ìlù

'The manner in which he is beating a drum’.

According to the A-analysis, this NP is a clause: yet it does not have the relative particle tí which is supposed to indicate a relative clause. Judging by appearance alone, it means that there is no basis at all for saying that (4b) is a relative clause.

The criticism about nesting structure is misplaced. Just as relative clauses such as (25a) can occur in a nesting structure, so can nominalizations such as (25b), (25c) and (25d).

(25)a. owó ti mo gba tí mo fún ẹ pamọ́

money that I receive that I give you

'The money that I received and gave to keep'

b. pé wọn ní kí ó lọ tí ó ní

that they say let he go and he say

òun ò níí lọ

he not will go

'The fact that they asked him to go said he would not go'

c. bí mo se lọ ti mo tete de

‘The manner in which I went and came back early'.

d. bí Òjó ba lọ ti ó bá dé

'If Ojo goes and comes back'

As the examples in (25) show, these so-called structures involves linking tí. (25a) is a case of relativization, while (25b) and (25c). are cases of nominalization. According to Awobuluyi (1975b), (25d) is also a case of nominalization. All this goes to show that it is simply not true that nominalizations cannot be combined into nesting structures. However, the point needs to be made that even this type of consideration is a mere surface structure one[10].

The criticism about semantic constraint is no more than a statement that the underlying structure of a relative clause as well as a factive nominalization must contain the element that is shifted or copied to the position before tí. Hence both (2a) and (3b) derive from (1).

(1) ó ńlu ìlù

‘He is beating a drum’.

(2)a ìlù tí ó ńlù

‘The drum that he is beating’.

(3)b. lílù tí ó ńlu ìlù

‘The fact that he is beating a drum’.

The argument is that (3b) must be regarded as a relative clause since the same constraint typical of (2a) is also typical of it. Again this is a mere surface structure phenomenon. If it shows anything, it is simply that the process of relativization is similar to those of this type of nominalization in that the surface structures are similar. Proof that this type of matching is not confined to relative clauses is that in Awobuluyi (1974), it is stated that

The nominalization in

[(26) Adé kò lọ lílọ kan

Ade didn't go at all'

must be a derivative of the verb in that sentence and of that verb only .

So much then for the same author's assertion that if NP's such as (26) are regarded as nominalizations, they would be the only normalizations that are subject to this semantic constraint!

Finally, we shall now look at the two criticisms about substitutability and order of occurrence. The former depends crucially on having a nominalized verbal form follow by a tí clause. But even here, it is simply not true that a qualifier substitutes for the tí clause. Consider the following sentences:

(9) lílọ tí ó lọ

‘the fact that he went’

‘The manner in which he went’

(27) lílọ rẹ̀

‘His going’

(3b) lílù tí ó ńlu ìlù

‘The fact that he is beating a drum’

(28)a. *lílù rẹ̀

‘His beating’

b. lílu ìlù rẹ̀

‘His beating a drum’

(29)a. kíákíá tí ó lọ

quickly that he go

‘The fact that he went quickly’

b. *kíákíá rẹ̀

* Its quickness’

In (27), the qualifier rẹ̀ 'his' substitutes for tí clause in (.9), but the same qualifier in (28a) cannot substitute for the tí clause in (3b); rather what it substitutes for is the clause minus the object which must be shifted to the position after the nominalized verbal form. This shows that what is involved here is not just a substitution since in spite of their difference (27) and (28) share the same factive or manner interpretation. Where the tí clause follows an adverb as in (29a), substitution of a qualifier is not even possible, as the non-occurrence of (29b) clearly shows.

As far as the question of order of occurrence is concerned, again this is a mere surface phenomenon; and one that does not say much about the true nature of the constructions concerned. Consider again the following sentences:

(30)a. ìlù tí ó ńlù yi

This drum that he is beating'

b. ìlù yi

'This drum'

(31)a. kíákíá ti ó lọ yi

'This fact that he went quickly'

b.*kíákíá yi
*'This quickly'

c. pe ó lọ kíákíá yi

'This fact that he went quickly'.

Superficially, the order of occurrence of qualifiers in (30a) and (31a) is the same i.e. 'relativized' element, + tí clause + qualifier. But where the qualifier in (30b) qualifies the relativized element, this is not the case with (31b) which does not occur. On the other hand, the factive nominalization in (31c) which has no nominal head can take the qualifier vi 'this', and it is a paraphrase of (31a). What this suggests is that the qualifier yi in (31a) does not qualify any noun head but rather the whole factive nominalization kíákíá ti ó lọ 'the fact that he went quickly'. In other words, in spite of the superficial similarity between (30a) and (31a), the deep structural affinity of the latter is really with (31c).

TWO THEORETICAL ISSUES

After this long digression on the criticisms of one of the two approaches, we may now return to the theoretical issues involved in the analyses under review. Basically these are two:

(a) Given identity of structure and difference of meaning, which should take precedence?

(b) Could separate constructions be subject to similar rules of derivation?

As we saw earlier in the tables of structures and their corresponding meanings, when an NP contains a tí clause, it can be interpreted as having a relative, a factive or manner meaning depending on the element preceding the tí clause. The position of the A-analysis is that all such NP's should be identified primarily as the same construction i.e. relativization, and that differences in meaning should be explained as different semantic interpretations of the same structural string[11].The B-analysis on the other hand puts together tí clauses with nominalizations having the same meaning, and then treats the difference in structure as surface realizations of identical underlying semantic structures. The A-anaiysis is reminiscent of the structural approach in linguistics, but it is inadequate even by the standards of the Aspects model of transformational grammar. Since it fails to provide different underlying structures for ambiguous noun phrases or sentences. I believe that the B-analysis which recognizes the primacy of the semantic structure is closer to the facts of language and the task of a grammar, which is to account for the underlying meaningful relations in language.

The second theoretical issue raised is whether different constructions can be subject to similar derivational processes, and even end up in a similar surface structure. Specifically, can relativization and nominalization be derived in similar ways? The A-analysis thinks that they should not:

Bamgbose is in effect saying that nominalizations and relative clause constructions are derived in very similar ways. The question that this immediately raises is: why this marked similarity between those two supposedly distinct constructions?

(Awobuluyi 1975b)

As far as I can see, as long as the underlying semantic representation is able to distinguish between construction types, it does not matter at all if derivational rules are similar; specifically, the claim made about Yoruba that certain nominalizations as well relative clause involve shifting or copying of an element in the underlying sentence. Notice that this sort of phenomenon is also true of topicalisation in the language, yet the author of the A-analysis sees no objection to recognizing relativisation and topicalisation as different construction types which nevertheless are subject to what he calls "two completely parallel syntactic processes". The question is: if relativisation and topicalisation can share similar derivational processes in spite of their difference meaning, why can't relativisation and nominalisation which yield ambiguous surface structures do the same?

Far from being completely unrelated, there is some evidence that there is a basic similarity between relativisation and nominalisation[12]. This basic similarity is recognised by Schachter (1973:43)

We may say that relativization is a process by means of which an unlimited number of sentences can be turned into nouns. In this respect, relativization appears to be quite similar to nominalization.

Even in Awobuluyi i'1975a), the similarity is apparent in the statement that

relativization involves (i) nominalizing some items within a sentence (to be known as the relativized item) and (ii) adjectivizing or turning whatever is left of such a sentence into a qualifier for the nominalization.

In other words, the process of relativization involves a nominalization as well.

Conclusion

If the position taken in this paper about the primacy of underlying semantic structure over the syntactic one is correct, then all the arguments designed to show that a given surface structure cannot be a token of two different construction types must be seen as misdirected. That being the case, there is no reason why a tí clause in Yoruba cannot occur in a nominalization as well as in a relative clause.

References

Awobuluyi, O. (1973), ‘Predicate and/or Verb Phrase Relativization’, An Unpublished Paper, Department of Linguistics and Nigerian Languages, University of Ibadan.

Awobuluyi, O. (1974), ‘Binary and Non-binary Aspects of Transitivity’, Paper Read at the Eleventh West African Languages Congress, Younde, Cameroon.

Awobuluyi, O. (1975a), ‘Adjectivisation, Nominalisation and Relativisation in Yoruba’, Paper Presented at the Department of Linguistics and Nigerian Languages Seminar, University of Ibadan.

Awobuluyi, O. (1975b), ‘Nominalisation or Relativisation?’, Paper Presented at the Department of Linguistics and Nigerian Languages, University of Ibadan.

Bamgbose, Ayo (1975), ‘Relative Clause and Nominal Sentences in Yoruba’, in Proceedings of the Annual Conference on African Linguistics: Ohio State University Working Papers in Linguistics, pp. 202-209, Columbus, Ohio.

Chafe, W. (1970), Meaning and the Structure of Language. Chicago: University of Chicago Press.

Schachter, P. (1973), ‘Focus and Relativization’, Language 49, 1: 19-46.

[1] This paper was published as Bamgbose, Ayo (1992), ‘Relativization or Nominalization?: A Case of Structure Versus Meaning’, Research in Yoruba Language and Literature (edited by L.O. Adewole) 3: 87-109.

[2] It should be noted that this paper was written in 1975.

[3] Cf. Chafe (1970: 288-289). What is called ‘relative meaning’ here relates specifically to this restrictive relative clause meaning and not to any other kind of meaning such as non-restrictive relative clause meaning.

[4] Ci-Verb represents the form of the nominalized verbal form which concicts of a reduplication of the initial consonant of the verb plus the vowel I prefixed to the verb e.g. lọ ‘go’, lílọ ‘going’. The lexical ambiguity hypothesis is stated in Awobuluyi (1973), foot note 3.

[5] According to Awobuluyi (1975a), even such heads which are already nouns are further nominalized vacuously. No matter how ridiculous this may sound, the fact is that any such vacuous nominalization is still different from nominalization by reduplication.

[6] See Awobuluyi (1975a).

[7] The same problem arises in an analysis which regards adverbs as simple nouns, for such an analysis will have to explain why they have only the manner meaning as in (6b) while other nouns have a relative meaning as in (2a).

[8] One example given in this paper is wíwà tí mow a ọkọ̀ náà which is glossed as ‘The fact that I drove the vehicle’ and ‘The way in which I drove the vehicle’.

[9] There is a good reason why (16a) is preferred to (16b) for non-restrictive relative clauses, but I will not go into that here.

[10] Note that apart from relative clauses and nominalisations, nesting structures are also possible with topicalized sentences. For example, compare (25a) with the topicalized sentence owó ni mo gbà tí mo fún ẹ pamọ́ ‘It is money that received and gave to you to keep.

[11] The only exception to this is in the treatment of a relative clause having a factive meaning which is derived from a factive nominalization.

[12] It is even possible that at an abstract level, the three processes of relativization, nominalization and topicalization are sub-types of the same phenomenon.

Search This Blog

LINGUISTICS AND YORÙBÁ STUDIES E-LIBRARY

RELATIVIZATION OR NOMINALIZATION?: A CASE OF STRUCTURE VERSUS MEANING

RELATIVIZATION OR NOMINALIZATION?: A CASE OF STRUCTURE VERSUS MEANING

Comments

Post a Comment

Popular posts from this blog

YORÙBÁ LITERATURE E-LIBRARY

SYNTAX AND GRAMMATICAL THEORIES E-LIBRARY SECTION

YORÙBÁ GRAMMAR E-LIBRARY SECTION