In English, most grammars tacitly or explicitly recognise two types of if. One of these introduces subordinate interrogative clauses:

  • I don't know [if I passed the exam].

The other introduces conditional protases:

  • My dad will be over the moon [if I passed the exam].

Apart from the apparent difference in meaning, the clearest difference between these two occurrences of if is syntactic. The clauses introduced by so-called interrogative if appear as Complements and the clauses introduced by so-called conditional if appear Adjuncts [read 'Modifiers' or 'Adverbials'].

This homophony between the items used to mark conditional protases and interrogative clauses is not a peculiarity of English, but occurs in many languages: Tagalog, American Sign Language and Spanish, for example.

However there is, it has been argued, a third if in English. This if introduces clauses whose internal syntax makes them seem more like conditional protases and less like interrogative clauses (chiefly because such clauses allow the use of irrealis-were). These clauses, however, appear not as Adjuncts within conditional constructions but as Complements, often within extraposition constructions (bolding etc in the following added by me):

  • It would be a shame [if Bob left]. (Williams 1974, p. 158)
  • It would be wonderful [if unicorns existed]. (Pullum 1987, p. 260)

In the examples above we see a meaningless dummy pronoun it functioning as Subject. The if-clauses concerned appear as Complements (here, of predicate adjectives) within the VP, but are interpreted, semantically, as subjects. As described by Pullum (1987), on their most likely readings, the examples above mean something like For Bob to leave would be a shame and For unicorns to exist would be wonderful.

Some other examples:

  • Can you imagine [if people really believed that]? (Rocchi 2011, p. 1)
  • I'd prefer (it) [if you came on Thursday].

The question

  1. Are there other languages where clauses which seem to have the same form as conditional protases occur in other environments, for example as the complements of verbs, adjectives or prepositions?

  2. If so, are any of those languages ones where conditional protases may be introduced by the same items as interrogative clauses?


Pullum, Geoffrey K. 1987. Implications of English extraposed irrealis clauses. In: Ann Miller and Joyce Powers (eds.), ESCOL ’87: Proceedings of the Fourth Eastern Conference on Linguistics, October 2-4, 1987, 260–70. Columbus, Ohio: The Ohio State University.

Rocchi, Manuela. 2010. A third If. University of Edinburgh MSc dissertation.

Williams, Edwin. 1974. Rule ordering in syntax. Unpublished Doctoral dissertation, MIT.

  • 1
    Off the top of my head, the only languages I can think of where protasis-if and this ‘third ifdon’t always coincide in form are languages where protasis and apodosis are not freely switchable, like Chinese; and even there, I think at least the first two examples would still be formally like conditionals, while the latter two are just not distinctly possible at all (i.e., can’t be distinguished from ‘can you imagine people really believing that?’). In languages where interrogative and conditional if are distinct, ‘third if’ aligns with conditional if in all cases I can think of. Mar 16, 2022 at 7:49
  • 1
    For the group of languages where ‘third if’ would not coincide with the others? Definitely Chinese, and I think also Japanese – I don’t think if clauses as object-like complements are possible at all in either (except to the extent that Japanese can always just make anything a nominal by conversion). Interrogative protases are just embedded regular interrogatives there, though (à la “I don’t know [did I pass the exam?]”), so perhaps those aren’t really good examples after all… Mar 16, 2022 at 10:56
  • 1
    As for other languages that have the same alignment as English: definitely all the other Germanic languages (English is more or less alone in having interrogative = conditional, but all have conditional = ‘third’) as well as the Romance ones, which have int. = cond. = ‘third’. Goidelic and Finnic languages have cond. ≠ int. (int. are embedded as in Chinese/Japanese); in Goidelic, ‘third’ = cond., and I think the same is true in Finnic, but I’m not entirely sure. Mar 16, 2022 at 11:09
  • 1
    Like the Billy Crystal line %I hate when that happens.
    – jlawler
    Mar 23, 2022 at 16:45
  • 1
    BTW, the "first kind of if" is just a shorter version of whether, the wh-word for embedded yes/no questions. Since they're binary, if fills the bill as well as whether, and is one syllable shorter to boot. If you wanna call it Complementizer if or WH if, go ahead.
    – jlawler
    Mar 26, 2022 at 21:35


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.