The syntax of the Welsh language has much in common with the syntax of other Insular Celtic languages. It is, for example, heavily right-branching (including a verbâÂÂsubjectâÂÂobject word order), and the verb for be (in Welsh, ) is crucial to constructing many different types of clauses. Any verb may be inflected for three tenses (preterite, future, and unreality), and a range of additional tenses are constructed with auxiliary verbs and particles. Welsh lacks true subordinating conjunctions, and instead relies on special verb forms and preverbal particles to create subordinate clauses.
There are at least four registers or varieties of Welsh that the term Modern Welsh is used to describe. There is Biblical Welsh, which is archaic and not part of colloquial usage, although some educated Welsh speakers are familiar with it. Two more registers are Literary Welsh and Colloquial Welsh; this article primarily describes Colloquial Welsh, except where noted. Finally, there are also a number of other dialects which diverge from these three varieties of Welsh. These various dialects are understudied, with the exception of some research by .
Welsh is a language with verb-initial word order, the usual word order being verbâÂÂsubjectâÂÂobject (VSO).
In addition to a verb and a subject, which are obligatory in a canonical clause, Welsh typically organizes additional information as follows:
Preverbal particle â Verb â Subject â Direct object â Indirect object â Adverbials (prepositional phrase, adverb, etc.)
The syntactic analysis of the VSO word order of Welsh is currently under debate. Richard Sproat and Ian Roberts have argued for an underlying subjectâÂÂverbâÂÂobject (SVO) word order with the surface VSO word order derived by syntactic movement of the verb to a higher position in the clause. On the other hand, Robert Borsley has argued against an underlying SVO analysis.
The arguments that Roberts makes about Welsh syntax are largely based on data from the Literary Welsh dialect.
The first step in the argument that Roberts makes for an underlying SVO analysis of Welsh word order is to argue that the subject moves out of the verb phrase to a position higher in the clause. This argument is made on the basis of data from passives, unaccusatives, and raising predicates in Welsh. The derived subjects in all three of these constructions behave like subjects of other predicates in Welsh in that they cannot be separated from the verb. That is, the subject must immediately follow the verb, as can be seen in (1)âÂÂ(3).
This suggests that the subjects in these three constructions are true subjects. On the assumption that all subjects in the language occupy the same position in the clause, this entails that the subject in Welsh must raise to a higher position in the clause than where it was base generated.
Another argument for movement of the subject in Welsh comes from reconstruction effects. Andrew Barss noticed that there is an interpretive difference between (4a) and (4b).
In (4a), himself can be interpreted as either coreferential with John or Bill. However, in (4b), it can only be interpreted as coreferential with Bill.
Cheng-Teh Huang analyzes these English facts by adopting the VP-Internal Subject Hypothesis and assuming that the subject raises from the verb phrase to a position higher in the clause. That is to say, the structure of (4b) is what is given in (5).
The closest c-commanding element that binds the reflexive pronoun, himself, is the trace of Bill (see Principle/Condition A of the binding (linguistics) theory).
The facts in Welsh parallel the facts in English. Specifically, the reflexive in (6a) can be interpreted as coreferential with either John or Bill, whereas the reflexive in (6b) can only be interpreted as coreferential with Bill.
The most straightforward analysis of these facts is to adopt the same analysis that Huang gives for English. That is to say, if one adopts the VP-internal subject hypothesis and assumes that the subject raises to a higher position in the clause, then an account of these facts is straightforward. Moreover, this suggests that the underlying word order is indeed SVO.
The second step in the argument that Roberts gives to motivate an analysis of Welsh word order in which the underlying structure of the clause is SVO and the verb has moved to a higher position in the clause is to observe that the verb appears in a higher position than the subject. If the subject has raised from a VP-internal position, then it follows that the verb must have also raised in order to be in a higher clausal position and to show up to the immediate left of the subject.
On the other hand, Borsley has argued against an underlying SVO analysis with the surface word order derived by verb movement. One of the arguments that he gives against such an analysis is based on negation. In Welsh, the negative adverb cannot be immediately followed by an object noun phrase, as the following examples show.
Borsley claims that this means the Welsh grammar must have a constraint against appearing next to an object noun phrase. He further argues that it would not be possible to state such a constraint since is not underlyingly next to the object noun phrase if one adopts an underlyingly SVO analysis of Welsh.
Note that Borsley takes the of a sentence like "We didn't see a dog" not to be this negative adverb, but a homophonous negative quantifier.
Welsh has a highly developed system of fronting constituents in focus in which parts of a sentence can be moved to the front for emphasis, rather than stressing them phonetically as English does. Most elements of a sentence can be moved to sentence-initial position.
The subject of a verb causes a soft mutation.
Sentence elements following , such as verbnouns, lose the when moved initially:
Determiners precede the noun they modify, while adjectives generally follow it. A modifier that precedes its head noun often causes a mutation, and adjectives following a feminine noun are lenited. Thus:
Genitive relationships are expressed by apposition. The genitive in Welsh is formed by putting two noun phrases next to each other, the possessor coming second. So English The cat's mother, or mother of the cat, corresponds to Welsh literally, 'mother the cat'; 'the project manager's phone number' is literally, 'number phone manager the project'. Only the last noun in a genitive sequence can take the definite article.
"be" is used for a number of constructions, including equating two noun phrases, using adjectives predicatively, and forming a wide range of grammatical tenses.
One way to equate noun phrases is to use what Gareth King calls "identification" forms of , with the word order NP<sub>1</sub>NP<sub>2</sub>.
Alternatively, a verb-initial word order may be used, with the "affirmative forms" of and a particle which triggers the soft mutation: NP<sub>1</sub>+SMNP<sub>2</sub>. This construction has both interrogative and negative variations which utilize different verb-forms and require, in the case of the negative, the addition of "not".
The predicative adjective construction uses this same verb-initial construction: NP+SMadjective.
In addition to the inflected preterite, future, and conditional tense forms, âÂÂsubjectâÂÂâÂÂverbnoun (with no mutation) is used to express a range of other times:
While the present and imperfect have special interrogative and negative forms, the future and conditional forms:
All of these constructions may be given perfect meaning by replacing with (lit. "after"), while substituting (lit. "newly") for (together with lenition of the verbnoun) expresses what may be termed the immediate perfect ("has just", etc.). Thus:
Welsh has inflected preterite, future, and conditional tenses. These do not take any particle such as , but instead soft mutation occurs after the subject: 'We saw a dog' (where is the lenited form of 'dog'). In negative sentences the soft mutation is instead placed on "not": 'We didn't see a dog'.
Questions are formed the same way as with the future and conditional forms above, as are negative statements except when there is a specific noun phrase functioning as the direct object. A specific noun phrase is a pronoun (, , etc.), a definite noun (, , etc.), or a noun preceded by a definite adjective (, , etc.). In these cases, is replaced by (a contraction of ). Thus:
The preterite, future, and conditional can also be formed with the appropriate inflected tense of 'to do' with a verbal noun (again with soft mutation after the subject). The preterite may also be formed with (which is the third person singular preterite of 'to happen'), which does not alter its form.
For affirmative statements with inflected verbs, it is particularly common to attach or , preverbal particles which trigger the soft mutation:
The passive voice can be expressed with the verb 'get' followed by the verb noun modified by a possessive adjective. For example:
The agent is introduced with the preposition 'with, by'. A "static passive", expressing the result of an action, can be expressed with the verb 'to be' followed by the preposition 'after' and, again, the verbal noun modified by possessive adjective. For example:
The prepositional phrase can also be used attributively:
The construction can be negated by replacing with 'without'.
There are two relative pronouns in Welsh, and . (which causes soft mutation) is used in "direct" relative clauses, i.e. those where the relativised element is the subject of its clause or the direct object of an inflected verb (rather than a periphrastic construction with ):
cannot coexist with . Instead, a special form, or , is used:
In all other cases, known as "indirect" relative clausesthose where the relativised element is genitival or the object of a preposition, , the complementizer, is used.
Note that because the object of a verbal noun is genitival, all periphrastic constructions take .
In the colloquial language, both and are typically omitted, and soft mutation occurs in both types of relative clause:
which in more formal Welsh would be
Welsh has a number of complementizers used under different circumstances. is used in non-focused affirmative clauses other than the present periphrastic with :
Affirmative clauses with the main verb in the preterite are an exception. The construction with is ungrammatical in many spoken dialects (as well as in the literary language), and a construction based on the preposition and the verbnoun is used instead:
The present periphrastic with tends to use a construction with the verbnoun in a genitival construction with the subject of the subordinate clause:
Negative clauses can be made negative normally or by replacing with :
Focused clauses are complementized with (in the North) or (in the South):
Focused clauses can be made negative with , or made negative normally (with or ):