Inversion in embedded questions

“And he told them who was it.”

(Fought 2003:98)

Inversion in embedded questions is a phenomenon in which embedded questions have a word order more typically associated with non-embedded questions. An example is the phrase what color are we in the following example from Chicano English:

1) I don’t know what color are we, but it doesn’t matter. (Fought 2003:98)

An embedded question is a question that occurs as a subordinate clause rather than as the main clause of the sentence. For example, what you wrote is a question embedded in the sentence I asked what you wrote.

This phenomenon is referred to as inversion because it relies on a type of movement called subject-aux inversion. Subject-aux inversion is the movement of the auxiliary verb to the left of the subject. In standard English, subject-aux inversion is seen in non-embedded questions. For example, when the questions in (2b) and (2c) are formed from the statement in (2a), the auxiliary are moves to the left of the subject we:

  2) a. We are blue.

 b. What color are we?

 c. Are we blue?

In standard English, although subject-aux inversion does occur with main-clause questions, it cannot occur with embedded questions. For example, (3a) would be perfectly normal because it does not contain subject-aux inversion, but (3b) would be unacceptable because it does involve subject-aux inversion:

  3)  a. John asked what color we are.

  b. *John asked what color are we.

However, many dialects of English do allow subject-aux inversion with embedded questions, as illustrated by sentence (1) above. In these dialects, sentence (3b) would also be acceptable even though it is unacceptable in standard English.

Who says this?

In the syntactic literature, this construction is discussed in some detail by McCloskey (1992) for Hiberno English and Henry (1995) for Belfast English. It has been reported in a number of North American English varieties as well, including African American English (Green 2002:87-89), Appalachian English (Wolfram and Christian 1976:129), Chicano English (Fought 2003:98), Northeast U.S. English (Ross 1975) and Newfoundland English (Clarke 2004:315). This construction has become rather widespread, according to the following passage from Wolfram and Schilling (2016:388):

[...] the use of inverted word order in indirect questions, as in She asked could she go to the movies, is becoming just as much a part of informal spoken American English as indirect questions without inverted word order, as in She asked if she could go to the movies.

It may nevertheless be the case that dialects differ as to how widespread embedded subject-aux inversion is. For example, sentences like (4), with the verb tell rather than ask, seem much less common across dialects than sentences with the verb ask:

4) And he told them who was it. (Fought 2003:98)

The sentence in (4) corresponds to the standard English sentence in (5):

5) And he told them who it was.

Syntactic Properties

No complementizer allowed with inversion

When an embedded question exhibits subject-aux inversion, complementizer like if and whether are not possible (Henry 1995:110; Green 2002:88). Therefore, although (6a) would be acceptable, (6b) would be unacceptable because it contains both subject-aux inversion and the complementizer if:

6) a.*It’s gonna ask you [do you wanna make a transfer].

 b.*It’s gonna ask you [if do you wanna make a transfer].

Either inversion or a complementizer is required

Main clause questions sometimes have the word order of a declarative, and are marked as questions purely by intonation, as in (7):

7) You're leaving at 12:00?

According to Green (2002:89), this is not possible in embedded questions. If there is no complementizer like if or whether, inversion is obligatory. This is illustrated by the fact that (8), which contains neither inversion nor a complementizer, is unacceptable:

8) *It’s gonna ask you [ — you wanna make a transfer].

Thus, although embedded questions cannot have both inversion and a complementizer, they must have one of the two.

Inversion in intermediate clauses?

Normally, when a wh-word at the beginning of the sentence corresponds to the object in an embedded clause, inversion only applies to the main clause, where the wh-word sits, and not to intermediate clauses.

9) John had hoped he would see Mary.

10) Who had John hoped he would see?

However, Henry (1995:108) notes that, in Belfast English, the intermediate, embedded clauses can also exhibit inversion in these cases:

11)  Who did John hope would he see?

12) Who did John say did Mary claim had John feared would Bill attack?

It is not known whether examples like (11) and (12) are possible in North American English varieties. If not, it would be an interesting difference in embedded question inversion between North American English and Belfast English.

Page contributed by Jim Wood on September 19, 2013

Updates/revisions: August 16, 2015 (Tom McCoy); June 11, 2018 (Katie Martin)

Please cite this page as: Wood, Jim. 2013. Inversion in embedded questions. Yale Grammatical Diversity Project: English in North America. (Available online at Accessed on YYYY-MM-DD). Updated by Tom McCoy (2015) and Katie Martin (2018).


Phenomenon Category: 
Wh-words and relative clauses
Phenomenon Dialect: 
Widespread American English
African American (Vernacular) English
Appalachian English
Canadian English (including Newfoundland English)
Chicano English
Smoky Mountain English