It was pointed out in section 2.7 that most of the rules required in the derivation of the passive are independently needed in the derivation of other forms. In this chapter the possibility is explored that T. Agent Postposing is also independently needed. Three constructions are examined where it appears that the subject noun phrase is postposed into a prepositional phrase as object of the preposition gan (by). These are impersonal passives, certain adjectival constructions, and certain nominal constructions. In each case the movement transformation is formulated and compared with T. Agent Postposing.

Impersonal passives

Impersonal passives Just as there are pairs of synonymous active and passive sentences, so there are also pairs of synonymous active and impersonal passive sentences. Examples are given in (1) to (6).

1. (1) Rhybuddiodd y dyn y plant.

2. Warned the man the children.

3. (2) Rhybuddiwyd y plant gan y dyn.

4. Warned the children by the man.

5. (3) Rhybuddiai'r dyn y plant.

6. Was warning the man the children.

7. (4) Rhybuddid y plant gan y dyn.

8. Was warning the children by the man.

9. (5) Rhybuddia'r dyn y plant.

10. Warns the man the children.

11. (6) Rhybuddir y plant gan y dyn.

12. Warns the children by the man.

In fact the impersonal passives are traditionally considered equivalent to the cael passives though stylistically superior to them.

This book is a revised version of my dissertation ‘The passive in Welsh’, which was submitted in September 1973 for the degree of Ph.D. of Cambridge University. The work was begun while I was a postgraduate student in Cambridge in 1968–9, but the bulk of the research was carried out while I was a lecturer in the Department of Phonetics at Leeds University. I am grateful to Leeds University for providing a grant towards the cost of publication.

I should like to thank my two supervisors for their help during this time; Dr P. A. M. Seuren originally suggested the topic for research and supervised the early stages, and Dr T. Moore supervised the later stages of the work. I have also benefited from the comments of my two examiners, Dr E. Fudge and Dr R. Hudson, from discussions with other linguists, in particular P. Meara, and from the intuitions of my Welsh-speaking friends and relations. I should also here like to thank Mr J. L. M. Trim, who first taught me linguistics and has always been a great source of encouragement and stimulation.

Till now work on Welsh within the framework of transformational grammar has been limited both in scale and in scope. It has been confined to a small number of articles and these have concentrated mainly on questions of phonology. This book is an attempt to rectify this situation, providing a full length study of Welsh within the transformational framework and concentrating on syntax. Owing to the dearth of previous work in this area this is bound to be an exploratory study, raising many questions for which no definite answers can be given. However, interesting explanations can be offered for much otherwise puzzling data which has caused difficulties within traditional and structuralist approaches, and it is felt that raising problems which have till now gone unnoticed is itself useful.

It is hoped that this book will be of interest to those who are working in the field of Welsh and the Celtic languages generally, as providing a new perspective on the language, offering new answers to some old questions and raising new questions instead. It is also hoped that the book will be of interest to those who are working in the field of transformational grammar, as providing an example of the approach being applied to a new family of languages. Any substantive work on linguistic universals must be based on such detailed studies as this for an increasing number of new languages.

It has been assumed throughout this study that the agent phrase of the passive should be derived by a postposing rule, which was formulated in Chapter 2 as T. Agent Postposing. Difficulties arose however in the last chapter in trying to give independent motivation to this rule, and there is in fact some evidence that the whole approach is mistaken. It is possible that the agent phrase is already present in sentence-final position in deep structure, not postposed into this position in the course of the derivation. This alternative analysis is now explored. In section 6.1 certain constructions containing the verb cael and the preposition gan will be examined and it will be argued that the prepositional phrase should be present in deep structure. In section 6.2 the cael passive will be compared with these forms and the possibility discussed that the prepositional phrase in this case too should be present in deep structure.

Constructions with cael and gan

Lexical objectCael may appear as the verb of a simple sentence with a lexical noun or a pronoun as direct object. Here it has the meaning ‘receive’. The subject is the recipient, the object the thing received and the object of the preposition gan is the source.

1. (1) Cafodd Wyn anrheg gan ei frawd.

2. Got Wyn (a) present by his brother.

3. i.e. Wyn got a present from his brother.

In this chapter an analysis of the structure of active sentences is presented, which will provide a basis for the discussion of passive forms in later chapters. In section I.I simple active sentences are considered, and in sections 1.2 to 1.5 sentences containing periphrastic forms of the verb are discussed. It is suggested that they should be analysed as complex sentences containing a tenseless embedding.

Simple active sentences

Phrase structure The simple active sentence in Welsh consists of a sentence-initial verb, which is followed by the subject, object and any prepositional phrases in this order. Examples are given below.

1. (1) Diflannodd y ci.

2. Disappeared the dog.

3. (2) Gwelodd y dyn y ci.

4. Saw the man the dog.

5. (3) Soniodd y dyn am y ci.

6. Spoke the man about the dog.

7. (4) khoddodd y dyn y ffon i'r ci.

8. Gave the man the stick to the dog.

In traditional grammars of Welsh it is assumed that the sentence consists of two units, subject and predicate (Richards 1938 p. 5). This view is retained in more recent, structurally orientated, treatments such as Watkins (1961). The motivation for this division into subject and predicate seems to be a universalist semantic one rather than syntactic. For instance Watkins claims that ‘it is difficult to imagine any kind of language which would not include something to talk about (subject) and something to say about it (predicate)’ (Watkins 1961 p. 183).

Not all active sentences have a grammatical passive equivalent. In this chapter these restrictions on passives will be described and an attempt made to explain why they should appear. Section 4.1 concerns cases where the lack of a passive seems to be due to a restriction on the verb of the active sentence. Sections 4.2 to 4.5 concern cases where the lack of a passive seems to be due to a restriction on the subject or object of the active appearing as agent or subject of the passive.

Restrictions on the verb

Defective verbs Some verbs are defective. They can appear only with a very few inflections, arid lack the uninflected form of the verb completely. Such verbs are clearly not able to appear in the passive construction, or any other tenseless embedding, where the uninflected form of the verb is required for the verb of the embedded sentence. Examples of such verbs are dylai (ought he), which appears only in the imperfect or pluperfect forms, meddai (said he) which appears only in the present and imperfect forms, and moes (take) which appears only in the imperative.

Some of these verbs will be excluded on other grounds too. For instance, dylai takes a tenseless embedding as direct object, as in (1). Such forms are discussed in sections 4.2.1 and 4.2.2.

1. (1) Dylai Mair ddychwelyd.

2. Ought Mair returning.

And meddai takes a quotation from direct speech as direct object, as in (2). Such forms are discussed in section 4.2.6.

In this chapter an analysis of passive sentences is presented, which builds on the analysis of actives already established. It is suggested that passives, like periphrastic forms, contain an embedded sentence with a tenseless verb. In sections 2.1. to 2.4 the derivation of straightforward passives is considered, and in section 2.5 some variations on this type. In section 2.6 a problem first raised in section 2.4.3 is followed up more fully.

Active and passive pairs

Examples of active and passive pairs are given in (1) to (6). Each pair of sentences appears to be synonymous.

1. (1) Rhybuddiodd y dyn y bachgen.

2. Warned the man the boy.

3. (2) Cafodd y bachgen ei rybuddio gan y dyn.

4. Got the boy his warning by the man.

5. i.e. The boy was warned by the man.

6. (3) Rhybuddiodd y plant y merched.

7. Warned the children the girls.

8. (4) Cafodd y merched eu rhybuddio gan y plant.

9. Got the girls their warning by the children.

10. (5) Rhybuddiai'r ferch y dyn.

11. Was warning the girl the man.

12. (6) Câi'r dyn ei rybuddio gan y ferch.

13. Was getting the man his warning by the girl.

In each case, the passive sentence has a sentence-initial inflected form of cael (get) of the same tense and aspect as the verb of the active. This is followed by a noun phrase identical to the object of the active.