It would be rather strange to call it a relative adverb, given that relative adverbs can also function as interrogative adverbs, but "that" cannot:
That's true only if you make it one of the defining characteristics of relative adverbs that they must be able to function as interrogative adverbs.
OK, but perhaps it is not a mere coincidence in English that ALL the words that function
indisputably as relative adverbs (
when, where, how, why) also function indisputably as interrogative adverbs, especially since both types of words are generally understood to have a syntactic role within the clauses they introduce. I shall now explore this at some length. Please forgive me for the mini-essay that follows.
In the formal sentences
Whom did John punch? and
Tom is the man whom John punched, the word
whom, regardless of whether it is the interrogative pronoun or the relative pronoun, is understood to function as the direct object of
punch(ed); indeed, it is because of this function of
whom within each type of clause that speakers who wish to speak very formally are grammatically entitled to use
whom rather than
who in those sentences and be formally correct in doing so.
Similarly, in the sentences
When do they go to church? and
Sunday is the day when they go to church, the word
when, regardless of whether it is the interrogative adverb or the relative adverb, functions as an adverbial modifier/adjunct within the clause
they go to church, in the same way that the prepositional phrase
on Sundays functions as an adverbial adjunct in
They go to church on Sundays and the word
then functions as an adverbial adjunct in
They go to church then.
It might give us pause, then, in our rush to deem
that a relative pronoun or relative adverb in sentences like
Sunday is the day that they go to church (a) that the word
that bears no resemblance to a word like
then and (b) that it is ungrammatical to try to use
that as an interrogative adverb (
*[strike]
That do they go to church?[/strike]) even though (all?) words that function as relative adverbs also function as interrogative adverbs. We might even question whether
that plays any semantic role in those clauses at all.
No lexical category is not 'recognised'. However, I thought that that as complementiser was the that in, for example, He said that he was happy, in which the that clause functions in a similar way to an object or, in other sentences, a subject.
In mainstream generative grammar,
that is indeed parsed as a complementizer in sentences like
He said that he was happy. It is also parsed as a complementizer in sentences like
That he was happy was well known, His statement that he was happy was not believed by everyone,
He was happy that she was there, and
The man that was happy was the only happy one in the group. It is
not parsed as a complementizer in sentences like
That was nice, That man was nice, or
He wasn't really all that nice.
Interestingly, in infinitival clauses, it is the word
for that functions as a complementizer, but it shows up (in modern English) only when the infinitival clause has an overt subject. Thus, if we wished to add a subject to the infinitival relative clauses in the Bible verse "
To every thing there is a season . . . A time to weep, and a time to laugh; a time to mourn, and a time to dance" (Ecclesiastes, chapter 3, verses 1 and 4, King James Version), we would also need to add
for:
a time for people to weep
a time for them to laugh
a time for them to mourn
a time for them to dance
Notice that we would not say that
for is either a relative pronoun or a relative adverb in such phrases (or ever!). It could be that at one time it was possible to have relative adverbs in infinitival relatives without
for (
? a time when to dance), but that clearly doesn't work in modern English. It would be far worse, however, to try to include for:
*! [strike]
a time when for people to dance[/strike]. Now, what about these noun phrases?
a time when people dance
a time that people dance
a time at which people dance
Each of those noun phrases contains a finite relative clause, and of course finite relative clauses always have subjects. Thus, we can't say that that serves the function of allowing us to inject the subject of the relative clause (
people). But is
that really any more meaningful in there than
for is in the infinitival relatives? The fact that we can omit it altogether and still have the same meaning (
a time people dance) suggests that it likewise contributes absolutely nothing to meaning.