Amalgam Clefts

“What we need is we need to help people find work in a legal way.”

(George W. Bush, 2004)

“He'll try to kill you is what he'll do.”

(Satan, South Park)

“It's sad, actually, is what it is.”

(Muriel's mother, A perfect day for bananafish, J.D. Salinger)

Amalgam pseudoclefts are a type of construction that can take two different forms. Examples of these two forms are shown in sentences (1a) and (1b):

1) a.*What he wants is he wants a good job.

b.*He wants a good job is what he wants.

The first form, as shown in (1a), consists of a wh-phrase (a phrase beginning with a question word such as what, why, or when), followed by a form of the verb be (usually is or was), followed by a complete sentence. In (1a), what he wants is the wh-phrase, is is the be verb, and he wants a good job is the complete sentence.

The other form in which amalgam pseudoclefts may appear contains the exact same three elements but in reverse order—namely, a complete sentence followed by a form of be followed by a wh-phrase, as in sentence (1b).

These sentences are described as "amalgams" (e.g. by Declerck 1988 and Lambrecht 2001) because they seem to contain amalgamated (that is, overlapping) sentence strings. In sentences (1a) and (1b), the overlapping string is he wants.

Amalgam pseudoclefts are called pseudoclefts because they appear very similar to more conventional, non-amalgam pseudoclefts like (2a) and (2b):

2) a.*What he wants is a good job.

b.*A good job is what he wants.

Sentences (1a), (1b), (2a), and (2b) all convey the same information as the following simple sentence:

3) He wants a good job.

Although the information conveyed is the same, the various types of pseudoclefts serve to highlight different information from the base sentence in (3).

Who says this?

Amalgam pseudoclefts are quite widespread in English and can be observed as early as the mid-17th century, especially with the verb do (Mair 2013). They are robustly attested in contemporary North American English and are also used in the UK. Related constructions have been observed in Australian English (McConvell 2004) and in a corpus of New Zealand English speech as well (Calude 2008). They are more frequent in speech than in written language, possibly because they do not conform to the standard grammar rules taught in school. In fact, amalgam pseudoclefts are even described as marginal or unacceptable in some of the linguistics literature (e.g., Higgins 1979 and Akmajian 1979), which suggests that this construction is not used by all English speakers. The factors conditioning the dialectal distribution of this construction are unknown, as the usage of amalgam pseudoclefts does not apparently correlate with any sociodemographic features.

Syntactic Properties

Detailed descriptions of the morphological and syntactic features of amalgam pseudoclefts can be found in Ross (2000) and O'Neill (2012); key observations from these papers are given below.


Amalgam pseudoclefts are characterized by three main ingredients: (i) a wh-clause or noun phrase with missing information, similar to an indirect question; (ii) a sentence supplying the missing information; and (iii) a form of be, usually is or was, linking the other two parts.

The wh-clauses

The wh-word representing the missing information is usually what, but also common are how, why, where, and when. Examples with these different wh-words are shown in (4):

4) a.*What I like is I like coffee.

b.*How he did it is he did it by calling an expert.

c.*Why I'm mad is I'm mad because you never told me you would be late.

d.*Where they met is they met at the park.

e.*When he arrived is he arrived yesterday.

Speakers vary with respect to which wh-words, other than what, they find acceptable. What is the most typical and the most widely-accepted wh-word in this context, but other wh-words may not be acceptable for some speakers. In particular, when the amalgam pseudocleft occurs in the wh-initial order, most speakers find who to be slightly unnatural, and which is not possible at all. For example, (5a) and (5b) would sound unnatural to most speakers:

5) a. ?Who I saw is I saw the President.

b. *Which book I like is I like War and Peace.

However, when the wh-clause follows is/was, these sentences improve significantly. Thus, (6a) and (6b) would sound much more natural to most speakers:

6) a.*I saw the President is who I saw.

b.*I like War and Peace is which book I like.

Amalgam pseudoclefts can also support phrases with multiple wh-words; in such cases, the clause on the other side of is/was provides a coordinated list of pairs, as in (7):

7) a.*Who ordered what is John ordered steak and Mary (ordered) salad.

b.*John ordered steak and Mary (ordered) salad is who ordered what.

The non-wh clause

The clause with the missing information usually has the form of a main/independent clause. When it comes after is/was, it may in some environments be introduced by that, but speakers generally prefer the that-less version. For example, the sentence in (8) is marginally acceptable when it contains that but is completely acceptable without that:

8)*What I like is (%?that) I like cheese.

However, when the clause precedes is/was, it is universally unacceptable to have it begin with that, as in (9):

9) (*That) I like cheese is what I like.

Although the clause is usually not marked by that, suggesting that it resembles an independent clause in structure, it also patterns with dependent clauses in some ways. For instance, it is sometimes possible to invert the subject and the verb in a main clause, such as when the sentence begins with a quote or a prepositional phrase like down the hill. Examples of this sort of inversion are given in (10):

10) a. Down the hill rolled the baby carriage.

b. "You look awful," said John.

However, dependent clauses do not support this kind of inversion, as illustrated by the fact that the sentences in (11) are unacceptable:

11) a. *She was sad [that down the hill rolled the baby carriage].

b. *I heard [that "You look awful," said John].

Like dependent clauses, the clause that specifies the missing information in an amalgam pseudocleft does not support subject-verb inversion, as illustrated by the fact that the sentences in (12) are ungrammatical:

12) a. *[Down the hill rolled the baby carriage] is why she was sad.

b. *What I heard is ["You look awful," said John.]

The phrase supplying the missing information is strongly focused. This focused phrase may be of any syntactic category, and may have almost any grammatical role. (13) illustrates several examples of different types of focused phrases, indicated inside brackets (capital letters indicate stress):

13) a.*He swam [FAST] is how he swam.

b.*I yelled [because I was MAD] is why I yelled.

c.*Where I went is I went [to the STORE].

d.*What I want is I want [you to get LOST].

For many speakers, amalgam pseudoclefts are less acceptable when the focused phrase is the subject of its sentence, especially when the pseudocleft is in the wh-initial order. Thus, (14) is less acceptable for many speakers:

14) ?What broke the window is [a ROCK] broke it.

The be verb

The be linking word of an amalgam pseudocleft can always have the form is and may optionally occur as was in the environment of another past tense verb form.

Instances without amalgamated components

Note that sentences that exhibit the same basic properties of amalgam pseudoclefts (namely, an information gap for which a root-like clause supplies the missing information) can be formed even without repeated strings. For example, (15a) and (15b) are very similar to (4b) and (4c), respectively:

15) a. How he did it is he called an expert.

b. Why I'm mad is you never told me you'd be late.

Sentences like (15a) and (15b) are also included in the amalgam pseudocleft construction despite the lack of amalgamated components.


The linear order of the two main phrases is generally reversible, as in sentences (1a) and (1b). However, as discussed above, the order with the wh-phrase after is is freer than the wh-initial order with respect to the wh-words it allows. Specifically, who and which are generally unacceptable in the wh-initial order but are acceptable when the wh-phrase occurs after the be verb.


The be verb form cannot be negated. Thus, (16a) and (16b) would be ungrammatical (Ross 2000, O'Neill 2012):

16) a. *I had a book isn't what I had.

b. *What I had isn't I had a book.

This contrasts with non-amalgam pseudoclefts, in which the be verb form may be negated, as in sentences (17a) and (17b):

17) a.*A book isn't what I had.

b.*What I had isn't a book.

Subject-auxiliary inversion

Amalgam pseudoclefts cannot undergo subject-auxiliary inversion. For example, (18b) and (19b) (which derive from the amalgam pseudoclefts in (18a) and (19a)) would both be unacceptable:

18) a.*What he had is he had a book.

b. *Is what he had he had a book?

19) a.*He had a book is what he had.

b. *Is he had a book what he had?

This is another feature that sets amalgam pseudoclefts apart from their non-amalgam counterparts because non-amalgam pseudoclefts may undergo question formation through subject-auxiliary inversion as in the following examples:

20) a.* What he had is a book.

b.* Is what he had a book?

21) a.* A book is what he had.

b.* Is a book what he had?


The be form of an amalgam specificational pseudocleft is morphologically restricted: it generally occurs only in a simple form as is or was. Tense matching between is/was and the main verb in other phrases is not obligatory—it may remain in present tense form in any environment, and it may optionally take the past tense form was if the main verb in the other phrases is in the past tense. For example, (22a) is acceptable with either is or was, but (22b) is only acceptable with is:

22) a.*What he needed {is/was} he needed a vacation.

b.*What he needs {is/*was} he needs a vacation.

Some speakers accept certain modals (such as could or might) in combination with be, but the future modal will and forms of auxiliary have cannot occur. For example, (23a) is acceptable to some speakers with could be but is never acceptable with had been, and (23b) is acceptable to some speakers with might be but never with will be or has been:

23) a.*What he needed {%could be/*had been} he needed a vacation.

b.*What he needs {%might be/*will be/*has been} he needs a vacation.


Amalgam and non-amalgam pseudoclefts can be embedded as dependent clauses under a very limited set of verbs called bridge verbs, such as say and think. For example, (24a) and (24b) are both acceptable:

24) a.*He said that what he had {is/was} he had a book.

b.*He said that he had a book {is/was} what he had.

However, amalgam pseudoclefts cannot be embedded under other kinds of predicates. Thus, (25a) and (25b) are unacceptable when the embedded component is an amalgam pseudocelft even though they are grammatical when the embedded component is a non-amalgam pseudocleft:

25) a.*He regretted that what he had was (*he had) a book.

b.*He regretted that (*he had) a book was what he had.

It is impossible to embed amalgam pseudoclefts as small clauses, whether they have no be or the phrase to be. For example, (26a) and (26b) are both unacceptable:

26) a. *I believe what he read (to be) he read my book.

b. *I believe he read my book (to be) what he read.

The same restriction on embedding as small clauses has been reported for wh-initial non-amalgam pseudoclefts (Williams 1983), but these judgments are controversial; in any case, there is a strong contrast between the amalgam and non-amalgam pseudoclefts in embedded contexts. For example, (27a) and (27b) are acceptable, in contrast to (26a) and (26b):

27) a.*I believe what he read to be my book.

b.*I believe my book to be what he read.

Extraction restrictions

Amalgam pseudoclefts have been described as “frozen” (Ross 2000); extraction of and from the major constituents, to form a question for instance, is completely impossible. This is similar to wh-initial specificational pseudoclefts, which are also subject to some extraction restrictions (for an overview, see den Dikken 2005). (28) gives some examples of sentences that are ungrammatical due to extraction out of amalgam pseudoclefts:

28) a. *What do you think that she sent t to France is what she did?

b. *On the shelf, what she did is she put the book t.

c. *It's he wants a bagel, what he wants.

d. *What do you think that what she sent to France is t?

Syntactic Analysis

While there is a rich literature on the syntax of specificational copular sentences and pseudoclefts, amalgam pseudoclefts are typically dismissed or mentioned only in passing, and detailed syntactic analyses are scarce. Ross (2000) provides an in-depth descriptive profile of pseudoclefts of all sorts, but does not develop a model for their syntax in a contemporary theoretical framework. A puzzle for any analysis of amalgam pseudoclefts is how to account for the occurrence of a seemingly independent clause in the structural subject position of the sentence. Under standard assumptions about the relationship between finiteness and nominative case, a bare finite clause is not the sort of object that should be able to serve as the subject of a sentence, and indeed, outside of this construction, it cannot. Thus, the amalgam pseudocleft in (29a) is grammatical, but (29b) is ungrammatical:

29) a.* [I wanted sushi] is what I wanted.

b. *[I wanted sushi] surprised him.

The story cannot be simplified by assuming the pre-copular sentence is a dependent clause with a silent that (a that-clause can easily serve as the subject of a sentence, as in (30b)), because a that-clause cannot generally occur as the non-wh clause in an amalgam pseudocleft (under the desired interpretation), making (30a) ungrammatical even though a sentence like (30b) is grammatical:

30) a. *[That I wanted sushi] is what I wanted.

b.* [That I wanted sushi] surprised him.

A useful starting point for syntactic analysis of this construction is an analogy between (non-amalgam) specificational pseudoclefts and question-answer pairs (e.g., Faraci 1967). Specificational pseudoclefts, like question-answer pairs, consist of a phrase with missing information, and an expression that provides the missing information. The wh-phrase in a specificational pseudocleft is structurally similar to a question in many respects: for instance, it allows multiple wh-expressions as in sentences (7a) and (7b). In amalgam specificational pseudoclefts, the analogy is even more apt as the phrase providing the missing information has the same form as a sentential answer to a question.

Some linguists have proposed that amalgam pseudoclefts are self-answering questions, as though the speaker is asking a question and answering it in the same sentence (den Dikken et al. 2000; Schlenker 2003; Lambrecht and Ross-Hagebaum 2006; Caponigro and Davidson 2011). Some of these authors observe that the “question” phrase serves as a “topic,” something either mentioned or hinted at in a discourse as being of current relevance to the conversation. These proposals do not always analyze the sentence structure in detail; neither do they analyze amalgam pseudoclefts in which the “answer” phrase comes first instead of the topical “question” phrase.

O'Neill (2012) proposes that amalgam pseudoclefts of both orders have a structure comparable to coordination, where the word is serves a grammatical function similar to words like and or or.

Related Phenomena

Several other syntactic phenomena resemble amalgam pseudoclefts. In particular, there are several phenomena that appear to use copulas in non-standard ways and/or to have finite clauses as the subjects of sentences. First, there is the that's X is Y construction illustrated in (31):

31) That’s what I was about to say is that everyone needs to be tested.

(Ross-Hagebaum 2004:403 (4))

Secondly, there are question-answer sentences such as (32):

32) You know what I need? is (I need) a new job.

Thirdly, there are sentences with free bes, such as (33):

33) Anne has the same problems with her anxieties is that she wakes up in the night.

(Massam 2013: (30c))

Finally, there is the double-IS construction shown in (34):

34) But what is also important for us to understand is, is that for America to be successful in this region there are some things that we're going to have to do here at home as well.

(President Barack Obama, cited in Zimmer's 2012 Language Log post)

See our Double-IS page for more information about Double-IS.

Page contributed by Teresa O'Neill on April 10, 2014

Page updated by Tom McCoy on August 17, 2015


Akmajian, Adrian. 1979. Aspects of the grammar of focus in English. New York: Garland.

Calude, Andreea. 2008. Demonstrative clefts and double cleft constructions in spontaneous spoken English. Studia Linguistica 62:78-118.

Caponigro, Ivano and Kathryn Davidson. 2011. Ask, and tell as well: clausal question-answer Pairs in ASL. Natural Language Semantics 19:323-371.

Declerck, Renaat. 1988. Studies on copular sentences, clefts and pseudo-clefts. Walter de Gruyter.

den Dikken, Marcel. 2005. Specificational copular sentences and pseudoclefts. The Blackwell companion to syntax, eds. Martin Everaert, Henk Van Riemsdijk, Rob Goedemans, and Bart Hollebrandse, 292-409. Wiley-Blackwell.

den Dikken, Marcel, André Meinunger, and Chris Wilder. 2000. Pseudoclefts and ellipsis. Studia Linguistica 54:41-89.

Faraci, Robert. 1971. On the deep question of pseudo-clefts. English Linguistics 6:48-85.

Higgins, Francis. 1979. The pseudo-cleft construction in English. New York: Garland.

Lambrecht, Knud. 2001. A framework for the analysis of cleft constructions. Linguistics 39:463-516.

Lambrecht, Knud and Sebastian Ross-Hagebaum. 2006. Apo koinou and intrusion: Toward a syntactic-pragmatic typology of amalgam constructions in spoken English. Proceedings of the 4th workshop on discourse structure. University of Texas, Austin.

Mair, Christian. 2013. Writing the corpus-based history of spoken English: The elusive past of a cleft construction. Language and Computers 77:11-29.

Massam, Diane. 2013. Intrusive be constructions in (spoken) English: apposition and beyond. Proceedings of the 2012 annual conference of the Canadian Linguistic Association.

McConvell, Patrick. 2004. Catastrophic change in current English: emergent double-be's and Free-be's. Talk given at the CRLC Seminar at the Australian National University.

O'Neill, Teresa. 2012. Coordination in English copular amalgams. Qualifying Paper, CUNY Graduate Center. Available here.

Ross, John Robert. 2000. The frozenness of pseudoclefts--towards an inequality-based syntax. Manuscript, University of North Texas.

Ross-Hagebaum, Sebastian. 2004. The that's X is Y construction as an information-structure amalgam. Paper presented at the 30th meeting of the Berkley Linguistics Society, 403-414.

Schlenker, Philippe. 2003. Clausal equations (a note on the connectivity problem). Natural Language and Linguistic Theory 21:157-214.

Williams, Edwin. 1983. "Semantic vs. syntactic categories." Linguistics and philosophy 6.3: 423-446.

Zimmer, Ben. 2012. Obama's "is is". Language Log post, October 23, 2012. Available here.