Skip to navigation – Site map

HomeIssues22PapersCategory membership and category ...


Category membership and category potential: The case of vague because

Martin Konvička


I analyse the English causal connector because as used in the so-called because X constructions in terms of its word class membership. I show that three types of because can be distinguished based on syntactic criteria. Depending on the complement of the connector, because can be used as a conjunction, as a preposition, and as a member of a third category. I also show that prior to use, we can only talk about an abstract category potential of the connector, but not about its concrete category membership. It is only through the context of use that becomes category membership apparent. On a more general note, this means that linguistic categories should be understood as being emergent from language use, not essentialistically as existing before and being independent of language use.

Top of page

Full text

I would like to express my gratitude to the journal editors and to the two anonymous reviewers whose comments have greatly helped to improve the original manuscript. All remaining shortcomings and flaws of the present paper are, needless to say, entirely my own responsibility.


1Besides the traditional subordinate clauses, because can also be complemented by constituents lacking a finite verb such as noun or adjective phrases and even interjections (see Section 2). These constructions are known as because X. One of the central issues that emerged early on in the debates about these constructions in English (for a summary, see for instance Zimmer, Solomon & Carson [2014]) is the question of the parts of speech categorisation of because in this construction. While some argued that because in this construction has become (or alternatively that because has always been) a preposition (e.g. Whitman [2013]; Carey [2013]; Pullum [2014]), others argued that because in because X should be analysed as belonging to a completely new category (e.g. McCulloch [2014a]; Stefanowitsch [2014]).

2Against the backdrop of these discussions, I will argue in this paper that instead of a single umbrella category for all uses of because in because X, we need three categories for three types of because X construction. These three types are defined by the nature of the complement of because in because X, i.e. by the element X. We can then categorise because complemented by an elliptical subordinate clause as a conjunction, because complemented by a noun phrase as a preposition, and because complemented by interjections or emojis as a member of, what I call, a third category.

3The category of because is therefore best analysed as vague (Denison [2017: 293]) and emergent (Hopper [1987]; [1988]), stemming from its concrete usage context. Category membership is not an essentialist property inherent to every possible instantiation of because prior to its usage, but a concrete realisation of its category potential.

4The paper consists of two major parts: a description of the data and a discussion of its theoretical implications. In Section 1, I discuss the data used in my analysis, and in Section 2, I present the various types of complements found in the because X constructions in more detail. Subsequently, in Section 3, I discuss the three categories of because in because X. Finally, in Section 4, I use the example of because in because X to address the more general issues concerning linguistic categories and linguistic categorisation, prior to summarising and concluding the paper in the final section.

1. Data

5My analysis is based on a corpus of tweets (Stöcker [2020]) originally collected for a study of because X equivalents in English, German, and Dutch (Konvička & Stöcker [2022]). For the present study, however, I have only used the English part of the corpus.

6One reason for choosing tweets as the empirical basis were the frequent claims on the importance of online communication platforms for the emergence and spread of the because X construction. Either online platforms were the place of origin of the construction or these platforms at least played a crucial role in its spread (see, e.g. Bailey [2012]; Whitman [2013]; McCulloch [2014b]; van Oostendorp [2014]; but see also Bergs [2018]).

7Furthermore, as Bohmann [2016: 170] observed, Twitter as a medium is “tolerant towards deviations from the norms of Standard English”. Texts produced on this platform can therefore be seen as a source of conceptually spoken language (see Koch & Oesterreicher [1985], Landert & Jucker [2011] or Burger & Luginbühl [2014: 173–200]). In other words, although tweets represent a written form of language, at least some of their properties are closer to spoken language.

  • 1 I use the term connector or causal connector as a functional description of because without any com (...)

8The corpus consists of tweets collected between 18 and 23 November 2019. A set of 10,000 tweets containing because was collected using twarc [2019], a Python library for linguistic research on Twitter. In the first step, all tweets containing the causal connector because were collected.1 In the next step, the collected tweets were tokenised and tagged using SpaCy [2019], an open-source library for Natural Language Processing in Python.

9To ensure the quality of the results in light of the language used on Twitter, often containing typos and non-standard spelling, a conservative search pattern was designed which does not allow any finite verb between the causal connector because and the next punctuation mark indicating the end of a sentence such as <.>, <!> or <?>. The design of the formal query maximises precision and minimises recall. The exact query is given in (1).

(1) pattern = [{LOWER: because}, {TAG: {NOT_IN: [VBP, VBZ, VBD, MD]}, OP: +}, {LEMMA: {IN: [., !, ?]}}]

10The results were afterwards additionally manually checked to rule out any false positives. The final set contained 84 instances of because X in the collection of 10,000 tweets. Because X constructions therefore amounted to 0.84% of the overall collection.

11This is lower than the 6.3% of because X in Bohmann’s [2016: 160] analysis, but the discrepancy can be, at least partially, explained by the very conservative search pattern as well as by the inaccuracy of the annotation tool in light of English morphology. In order for a because construction to qualify as a case of because X, no finite verb can occur after the causal connector. However, since finite verbs in English are often formally undistinguishable, for example, from nouns, some genuine cases of because X were probably eliminated as false negatives.

2. Complement variability

12The label because X describes a structure consisting of the English causal connector because followed by a wide range of elements not traditionally associated with subordinating conjunctions such as noun phrases (2c), adjective phrases (2d), interjections (2e), pronouns (2f), compressed clauses (2g), but also emojis (2h). However, as was already mentioned above, no finite verb can occur in the X slot.

(2a) They liked him because he was very friendly.
(2b) They liked him because of his friendliness.
(2c) Can’t tell if I’m in a lot of abdominal pain because Crohns or because antibiotics. (Twitter, 2019)
(2d) Shipping always is a pain because expensive for another country (Twitter, 2019)
(2e) phone wallpapers because oof. (Twitter, 2019)
(2f) Who else does their makeup just to sit around in their room because ME. (Bailey, 2012)
(2g) I’m trying to muster up as much happiness as I can for today because LMAO. (Twitter, 2019)
(2h) Start my day with a yoghurt drink too because
😘 (Twitter, 2019)

13Although all of the possibilities for the complement of because exist as stated in (2), they differ in their frequency. Both Bohmann [2016: 160] and Schnoebelen [2014] conclude in their analyses that the subtype because NP is the most common one. Other variants, as indicated in Table 1, are also not uncommon.

Table 1. Complement frequencies of because in English because X

Part of speech

Frequency (Schnoebelen [2014])

Frequency (Bohmann [2016])

Nouns/Noun phrases

32.02 %

38.8 %

Compressed clauses

21.78 %

14.15 %

Adjectives/Adjective phrases

16.04 %

9.8 %


14.71 %

20.3 %


12.97 %



2.45 %




16.6 %

14A note on the comparability of the two studies is needed. The categories nouns, adjectives, interjections, and compressed clauses, referring to acronymic complements such as yolo ‘you only live once’ or lol ‘laughing out loud’, are the same and therefore comparable in both studies. However, the other categories are comparable only to a certain extent. Schnoebelen’s agreement refers to expressions such as yes or no, which Bergs [2018: 49] proposes to combine with interjections in one category particles due to their similar syntactic behaviour. This would yield a category comprising 27.68% of the results, advancing it to the second position right behind the most frequent nominal complements.

15Furthermore, Bohmann’s [2016: 160] category other consists of “interrogative sentences, hashtags, emoticons, user mentions, adverbs, demonstrative and personal pronouns, hyperlinks, zero complements and a few cases that were unclear without further context.” This category overlaps with Schnoebelen’s pronouns, but also includes many complement types not at all captured in Schnoebelen’s statistics.

16Regardless of the exact figures, however, the quantitative analysis of the complement of because in because X shows two important things. First, that nominal complements are the most frequent type of complements and pronouns are the least frequent complements (if found at all). Second, the exact type of the complement is in fact not relevant for the categorisation of the construction as because X as long as the complement of because lacks a finite verb form.

17This allows us to make the first distinction in all the constructions containing because. If a construction contains the causal expression because followed by a finite verb form, it does not count as an instance of because X and falls out of the scope of the present study. If a construction, however, contains the connector because followed by anything except a finite verb form, we are dealing with an instantiation of because X. We can distinguish three types of because X constructions in this latter group, as I will show in the next section.

3. Categorisations of because

18Linguistic categorisation in terms of parts of speech (see, e.g., Schachter & Shopen [2007]) has been a topic of discussion at least since Antiquity (for a historical discussion see, e.g., Robins [1966]; Aarts [2004]). Generally speaking, word class membership can be based on various types of criteria (or on combinations of these types) (Givón [2001: 49]).

  1. Semantic: the kind of meanings that tend to be coded by words of a particular class;

  2. Morphological: the kind of bound morphemes that tend to be affixed to words of a particular class;

  3. Syntactic: the typical position(s) in the clause that words of a particular class tend to occupy.

19In my discussion of the categorisation proposals for because in because X, I will exclusively follow syntactic criteria and disregard semantic and morphological ones. For the analysis of equivalent constructions in other languages, in particular case languages such as German (Konvička & Stöcker [2022: 349]) or Czech (Konvička [2020]), morphological criteria would be relevant. The reason for this is the use of the nominative case in noun phrases occurring in because X equivalents in these languages. For English, however, morphological criteria play no role and will therefore be disregarded here.

20Based on syntactic criteria, in particular those based on the type of complement the causal connector because takes, I distinguish three categories of because as used in because X constructions:

  1. Because used as a conjunction;

  2. Because used as a preposition;

  3. Because used as a member of a third category.

3.1. Because used as a conjunction

21The first way to categorise because in because X constructions is to retain the categorisation of because as used in because clauses and analyse the connector as a subordinating conjunction. This applies particularly to instances of because X of the type AdjP1 because AdjP2. These constructions can be analysed as textual ellipses (Konvička & Stöcker [2022: 345–354]) due to the fact that the elided element can be recovered from the matrix clause preceding the because X construction. This elided element is usually, although not always, a copula verb.

22Example (3) contains a concrete instantiation of the pattern AdjP1 because AdjP2. At first glance, we see because followed by an adjective expensive – such a structure is incompatible with the analysis of because as a subordinating conjunction because this categorisation would require the presence of a finite verb in the complement of because.

(3) Shipping always is a pain because expensive. (Twitter, 2019)

We can, however, identify a parallel between the structure of the linguistic material preceding because in (3) and the structure of the linguistic material following it. In the matrix clause of the because X construction, the subject shipping is followed by the subject complement always is a pain which includes a copula verb is. In the linguistic material following because, we find the adjective expensive which, in analogy to the subject complement of the matrix clause, can function as the subject complement of the subject shipping in the first clause of the complex sentence.

23In other words, the structure in (3) can be described as a case of anaphoric textual ellipsis (see, e.g., Lemmens [1991: 15]; Konvička & Stöcker [2022: 351]). Based on the parallel structure of the complex sentence, we can precisely recover the material we assume to have been elided in the second clause containing the because AdjP construction, as illustrated in (4).

  • 2 I use square brackets in the examples to signal the (potentially) recovered parts of an elliptical (...)

(4) Shipping always is a pain because [shipping is / it is] expensive.2

24Against the backdrop of cases like (4), I argue that because in constructions of the type because AdjP should be categorised just as because in constructions of the type because VP. In other words, because in anaphoric textual ellipses of the type because AdjP can still be categorised as a conjunction that introduces a subordinate clause. The linguistic material preceding because and the linguistic material following because are parallel to such a degree that they allow for an unambiguous recovery.

25Crucial for this analysis of constructions of the type AdjP1 because AdjP2 is the interplay of both their parallel structure and the resulting possibility of unambiguous recovery. If both criteria are not met, the analysis of because as a subordinating conjunction, albeit in an elliptical construction, becomes untenable.

26A case in point is given in (5). Syntactically, we see a structure comparable to (4) in that the linguistic material following because lacks a finite verb. In particular, we can see a structure of the type because NP. However, the linguistic material preceding because is not parallel to the linguistic material following it, thus a recovery, as in (4), is not possible.

(5) It is going to be blowout on this saturday, because djtripleks (Twitter, 2019)

If we only focus on the syntax of because and its complement, structures of the type VP because AdjP (4) and VP because NP (5) can be classified similarly. Both contain the expression because followed by a complement lacking a finite verb and both therefore qualify as because X constructions. However, while one type of because X, illustrated by (4), shows structural parallelism between the complement and the antecedent, the other type, illustrated by (5), does not. Although this difference does not affect the categorisation of both constructions as because X constructions, it does affect the categorisation of the expression because itself.

27While because X constructions of type (4) allow because to be classified as a conjunction in an elliptical clause, because in because X constructions of type (5) has to be classified differently, namely as a preposition.

3.2. Because used as a preposition

28Due to the fact that NPs constitute the most frequent group of complements of because in because X constructions (Schnoebelen [2014]; Bohmann [2016]; Konvička & Stöcker [2022]), the proposal has been made early on to categorise because in constructions such as (6a) as prepositions (Pullum [2009]; [2014]; [2014]). The causal connector because is in these constructions closer to other causal prepositional constructions such as because of (6b) or due to (6c).

(6a) Can’t tell if I’m in a lot of abdominal pain because Crohns or because antibiotics. (Twitter, 2019)
(6b) Can’t tell if I’m in a lot of abdominal pain because of Crohns or because of antibiotics.
(6c) Can’t tell if I’m in a lot of abdominal pain due to Crohns or due to antibiotics.

29The reason for categorising because in constructions such as (6a) as a preposition is twofold: The structure of the complement of because on the one hand and the structure of its antecedent – i.e. the linguistic material preceding the connector – on the other. As to the former, it consists of a noun phrase. As to the latter, it lacks the parallel structure that would allow for categorising because as a conjunction, as discussed in the previous section. Whereas we can assume that a structure of the type because AdjP (7a) is potentially recoverable into because VP (7b), an analogical recovery of because NP (7c) into something like because VP (7d) is not feasible.

(7a) Shipping always is a pain because expensive. (Twitter, 2019)
(7b) Shipping always is a pain because [shipping is / it is] expensive.
(7c) Can’t tell if I’m in a lot of abdominal pain because Crohns or because antibiotics. (Twitter, 2019)
(7d) *Can’t tell if I’m in a lot of abdominal pain because [I’m in a lot of] Crohns or because [I’m in a lot of] antibiotics.

30There is no evidence to justify any analysis of cases like (8a) as textual ellipses, but an analysis of such constructions as contextual ellipses (8b) can seem more relevant. This argumentation would allow to retain the traditional categorisation of because as a conjunction even in constructions like because NP.

(8a) You misunderstand. The point is that, while *anyone* can be a victim, *only men* can be perpetrators. Because patriarchy. (Twitter, 2019)
(8b) ?You misunderstand. The point is that, while *anyone* can be a victim, *only men* can be perpetrators. Because [we live in a] patriarchy.

31Although a categorisation of because NPs as contextual ellipses is possible in theory, in practice, however, there is no concrete evidence to support it. The reason why we could assume because in constructions of the type because AdjP to be a conjunction was the structural parallelism between its antecedent – i.e. the linguistic material preceding the connector – and complement. No such structural corroboration of the elliptical analysis of constructions of the type because NP is available.

32We could nevertheless assume that constructions of the type because NP are contextual ellipses. If we assume this, however, we should also be able to say what was elided and what is therefore supposed to be recovered. Unfortunately, since we lack any textual clues in constructions such as (8a), a recovery of because NP to because VP (8b) is equally plausible as a recovery to because PP (9).

(9) You misunderstand. The point is that, while *anyone* can be a victim, *only men* can be perpetrators. Because [of] patriarchy.

33We can sometimes make educated guesses about the potentially elided linguistic material. In the case of a because NP construction as in (10a), we can use our linguistic and world knowledge and assume that the NP in question is either the object (10b) or the subject (10c) in an elliptical clause.

(10a) Wake up in the middle of the night crying because bad dream. (Twitter, 2019)
(10b) Wake up in the middle of the night crying because [I had a] bad dream.
(10c) Wake up in the middle of the night crying because bad dream [disturbed my night].
(10d) Wake up in the middle of the night crying because [of a] bad dream.

Even in these cases, however, we will inevitably reach two insurmountable hurdles. Firstly, we will be always dealing with potential recoveries of potential ellipses. Secondly, even if we have good reasons to assume either (10b) or (10c) to be the case, it will always be the assumption of the linguist, not the reality of the language user. In other words, we have no way of knowing whether the language user meant a structure such as (10d) or a different one to be recovered by the addressee (for a critique see, e.g. Selting [1997]).

34These problems of the analysis of structures of the type because NP as ellipses ultimately boil down to a conflation of two distinctions. The first distinction has to do with the nature of ellipses, the second one with the nature of linguistic analysis.

35Regarding the nature of ellipses, we need to bear in mind that an elliptical structure requires the interplay of both the speaker and the addressee. The speaker produces a linguistic structure with the knowledge of its elliptic character expecting that the addressee will recognise the ellipticity as well as correctly recognise the elided material in order to recover it. Corpus attestations such as (10a), however, only enable us to see the surface structure, but say nothing about the intentions of the speaker, nor about the success with which the addressee potentially recognised such a structure as elliptical. Since we do not have access to these details, we are dealing with assumptions.

36Regarding the nature of linguistic analyses, we need to bear in mind that an ellipsis is a structure emerging between the speaker and the addressee. However, when analysing because NP constructions such as (10a), we leave the plane of language use and enter the plane of its analysis. In other words, we are not just describing what the language users produced, we are interpreting it. Because we do not have access to the speaker’s (and the addressee’s) mind, we as language analysts cannot assess what was meant and was not realised with any confidence. We can only make assumptions.

37In order to successfully steer clear of these two problems, I exclusively look at the actual language structure. In the light of the lack of co-textual clues, as was the case in the analysis of because as a conjunction (Section 3.1.), I analyse constructions of the type because NP as containing because used as a preposition.

  • 3 The idea that English prepositions and conjunctions in fact form a single category distinguished on (...)

38Apart from these theoretical reasons, the analysis of because in because NP as a preposition is further corroborated by the syntactic properties of other English English expressions such as after, before or since. Depending on their complement, these expressions oscillate between being categorised as prepositions, if followed by a noun phrase (11a), or as conjunctions, if followed by a clause (11b). According to, among others, Pullum [2009: 269], this analysis holds true for the majority of all English prepositions.3

(11a) I arrived after his departure.
(11b) I arrived after he departed.

Moreover, it is also cross-linguistically not uncommon to find prepositional expressions (11a) being employed as subordinating conjunctions (11b). Disregarding evidence of because X equivalents outside of English, the expressions oscillating between a prepositional and conjunctional use can be found in languages as diverse as Japanese, Hausa and English (Schachter & Shopen [2007: 51]).

39In analogy to cases such as (11), I then analyse the connector because in constructions of the type because NP also as prepositions that are sometimes complemented by whole clauses (12a), and sometimes by noun phrases (12b).

(12a) I arrived because John called me.
(12b) I arrived because John.

40Despite these commonalities of because in constructions of the type because NP with traditional prepositions, some scholars (e.g. McCulloch [2014]) remark that because NP constructions nevertheless differ from other prepositional constructions. First, although the nominal complements of because in because NP can be relatively heavily modified (13), noun phrases containing determiners (14a) or pronouns (14b) are dispreferred (see Schnoebelen [2014]; Kanetani [2019: 159]; Bergs [2018: 49]).

(13) Maybe I’ll take dog and see if I can locate source...also peeing because aging small dog bladder. (Twitter, 2019)

(14a) guess we should not waste the peak of our teenage years in uncertain things when we could be enjoying these years, because these years? (Twitter, 2019)
(14b) Who else does their makeup just to sit around in their room because ME. (Bailey 2012)

41Although perhaps not a prototypical preposition, I have shown that there is enough evidence to categorise because in constructions of the type because NP as a causal preposition. The connector because in constructions of the type because X is, however, not only complemented by elliptical adjectival phrases (discussed in Section 3.1.) or noun phrases. In some instances, however, the X slot can also be filled by prepositional phrases, interjections, particles or emojis. These types of complements are neither compatible with categorising the connector as a conjunction, nor with categorising the connector as a preposition. Therefore, we need to make a step beyond the established word class system in order to account for these cases.

3.3. Because used as a member of a third category

42As I discussed in the previous section, some characteristics of because in because NP can be considered as unprototypical of prepositions, especially the fact that pronouns and determiners are uncommon in the construction. This observation as well as the fact that because can be combined with expressions of (dis)agreement like yes or no (15a), interjections (15b), emojis (15c) or prepositional phrases (15d) has led to suggestions to analyse because in these constructions as a member of a novel word class outside of the traditional parts-of-speech system (see, e.g., McCulloch [2014]; Stefanowitsch [2014]; Konvička [2020]). As a convenient way to refer to this category, I use the term third category.

(15a) My friends #1: don’t taking it seriously and saying it’s temporary. Friends #2: Asexual, Panromantic? Who is this? Crash: in scared I run away from him, because yes. (Twitter, 2019)
(15b) I just often wonder if those who actually call in (especially to some specific shows), aren’t actually masochistic because wow! (Twitter, 2019)
(15c) Start my day with a yoghurt drink too because
😘 (Twitter, 2019)
(15d) You’re a l[o]st cause and you don
t even know it. I read. Conservative and progressive, because in independent. (Twitter, 2019)

43One potential counterargument for treating because in constructions like (15) as members of a separate, third category is to consider them to be contextual ellipses. However, just as there is no direct evidence for the elliptical status of constructions of the type because NP (see Section 3.2.), there is also no evidence to treat constructions like (15) as ellipses (see also Konvička & Stöcker [2022: 345–354]).

3.4. Interim summary

44Looking back at the discussion so far, the first broad distinction we can make for all because constructions is to distinguish between because clauses and because X constructions, the criterion being the presence or absence of a finite verb form in complement of because in the construction. This distinction has led some to categorise because (+ clause) as a conjunction and because (X) as something else. As I have demonstrated in the previous three sections, however, a more fine-grained distinction of the because X constructions is required.

45Among the because X constructions, three different variants can be identified based on the complement type of because. First, constructions which can be identified as (anaphoric) textual ellipses and in which because can be categorised as a conjunction. Second, constructions of the type because NP in which because can be categorised as a preposition. Third, constructions with because complemented by elements incompatible with the two previous classifications in which because should be classified as a member of a third category.

46The concrete discussions of the various ways to categories different subtypes of because constructions in general and because X constructions in particular, will serve in the following sections as a foundation for a more general debate concerning the nature of word classes and the ways we, as linguists, categorise linguistic phenomena.

4. Approaches to categories

4.1. Linguistic description and categorisation

47The studies of because X have so far pursued two distinct, albeit related, goals: (i) a more detailed empirical description of the because X constructions, also from a cross-linguistic point of view (see, e.g. Konvička [2020]; Konvička & Stöcker [2022]; Wessman [2017]) and (ii) a theoretical discussion of category membership of because in these constructions (see, e.g. McCulloch [2014a]; Stefanowitsch [2014]; Konvička [2018]). While I have mostly focused on the former aspect of the discussions so far, I will turn my attention to the latter aspect in the remainder of this paper.

48What emerges from the theoretical discussions above is that if there is a construction in which because is used in a sufficiently different way from its traditional use as a conjunction, then we need a new category for that novel use. The question is, however, whether we do need one umbrella category to encompass all uses of because in the various instances of because X.

49As I discussed in the previous sections, because X constructions have in common that they are not because clauses and yet all use the causal connector because. This is, however, not enough to warrant a uniform categorisation of all uses of because in because X.

50The theoretical troubles arising from the search for a single categorisation for because in because X become obvious if we take a brief look outside of linguistics. By way of comparison, the chemical compound of two hydrogen atoms and one oxygen atom, represented as H2O and colloquially referred to as water, is in our everyday lives categorised as a fluid. We also know that if we heat water above its boiling point, it will cease to be fluid and turns into vapour. By lowering the temperature below the melting point of water, it turns into ice and becomes solid. In linguistic terms, we could paraphrase this by saying that water is context-dependent or construction-dependent. Just as it is not helpful to categorise water as always fluid (or always solid or always a vapour) because it is always water, it is not helpful to categorise because always as a conjunction (or a preposition or as something else) only because we are always dealing with because.

51The category membership of because, as I will show, can be best understood as an emergent property, emerging from its concrete usage context, and not as an essentialist property inherent to all instantiations of because prior to use.

4.2. Category membership as prototype-based

52In general, two approaches to word class categorisation can be distinguished: the classical or essentialist approach and the prototype-based approach. The former approach assumes that each expression belongs to exactly one category and category membership is based on one or more necessary and sufficient features shared by all of its members. This approach to word class membership is present in the generativist tradition (see, e.g., Baker [2003]) where lexical, but sometimes also functional categories, are understood as universal and cross-linguistically valid (for a critique see, e.g., Haspelmath [2007]). The latter approach follows the principles of the Prototype Theory (Rosch [1973], but see also e.g. Taylor [1995] for an overview) and by contrast does not assume that all members of a category necessarily share the same features. Category members are rather classified as more or less prototypical members, depending on the number of features a member shares with other members of its category.

53These different approaches to linguistic categories are mirrored in the different approaches to categorisations of because. In its conjunctional use, because in because X can still be considered a conjunction even though it is not followed by a finite verb due to the elliptical character of the complement (see Section 3.1.). For the purposes of this discussion, we can say that because used as a conjunction is prototypically followed by a subordinate clause and thus by a finite verb (16a). Uses of because in cases like (16b) would therefore not count as conjunctions following the essentialist approach to word classes since the necessary feature of a finite verb in the complement of because is absent. Addressing this issue from the prototype-based perspective, (16b) can still be categorised as a conjunction in spite of the fact that the necessary criterion, the presence of a finite verb, is not met.

(16a) becauseCONJUNCTION + VP + AdjP [+ finite verb]
becauseCONJUNCTION [+ VP] + AdjP [‒ finite verb]

54Prototype-based categories do not imply that every outlier is automatically a non-prototypical member of a category. Both essentialist and prototype-based approaches rely on categories and boundaries between categories. The difference lies, however, in the nature of the category boundaries. Just as a fish is not a less prototypical member of the category of birds, but a member of a different category, so the prepositional because is not a less prototypical conjunctional because, but rather a member of a separate category.

55This can be seen in the third group of because X constructions discussed in Section 3.3. The connector because in these constructions cannot be categorised as a conjunction or as preposition so that it is categorised as a member of a third category. While the first two types of because are defined in positive terms either by the presence of a textual ellipsis or by the presence of a noun phrase in their complement, the third category of because is defined ex negativo. The connector because in this type of because X construction is complemented neither by a textual ellipsis nor a noun phrase.

Table 2. Feature-based classification of because construction


Construction type

Finite VP



because clause

because X












4.3. Category membership as emergent

56Both the essentialist and prototype-based approaches to word class membership discussed in the previous subsection presuppose that an expression belongs to a certain category and can be used accordingly. They differ, however, in their view of the categorial boundaries and the gradience of category membership. By contrast, emergentist approaches to word class membership do not presuppose that any expression a priori belongs to a category. The category membership of any expression, according to the emergentist principles, is something that can be only determined a posteriori based on the context of use.

  • 4 Both references contain early accounts of Hopper’s concept of Emergent Grammar. Based on a chronolo (...)
  • 5 Although the concept of Emergent Grammar can be traced back to Hopper, its underlying principles ca (...)

57By emergentist approaches, I mean all theories in line with the tenets of Emergent Grammar (see Hopper [1987]; [1988]; [2011]).4,5 The basic idea is that grammar – or linguistic structure in general – is, as Hopper [1987: 141] argues, “a real-time, social phenomenon, and therefore […] temporal, its structure is always deferred, always in a process but never arriving, and therefore emergent”. In other words, linguistic structure is not assumed to exist prior to language use, but to emerge from it.

58If we recall the discussion of the word class membership of because in because X, I wish to argue that because cannot be said, in essentialist terms, to be a conjunction, to be a preposition or to be a member of a third category. Rather, because can only be said, based on its complement, to be used as a conjunction, to be used as a preposition or to be used as a member of a third category.

59Without knowing anything about its complement or about its broader co-text or context, because can be described as having the potential to be used as a conjunction, as a preposition, or as a member of a third category. The expression can be said to be pre-categorial. It has no fixed a priori category membership, but it has a certain category potential.

  • 6 This terminology goes back to Hengeveld [1992] who proposed a distinction between flexible and rigi (...)

60This analysis is compatible with two earlier approaches to category membership: flexible words (Hengeveld [1992]) and Radical Construction Grammar (Croft [2001]; [2013]). Flexible words6 (see, e.g., Rijkhoff & Lier [2013: 5–6]) are words such hammer, kiss or dance in English which can be categorised as nouns or verbs depending on their context (Farrell [2001]). Flexible words can undergo a functional shift by means of conversion. It is not possible to say a priori whether an expression such as hammer is a noun or a verb unless we know in what context the expression is used. The same applies, as I have shown earlier, to the category membership of because which can be also classified as a flexible word due to its categorial indeterminateness.

  • 7 Precursors of this radical constructionist position can be found in some of the early works on gram (...)

61My analysis of because in the three types of because X is also compatible with the tenets of Radical Construction Grammar (Croft [2001]; [2013]) whose central axiom is that word classes are language-specific and construction-specific7. This is something that has in principle been long accepted in the discussion of because X. What has not been sufficiently accepted is the fact that the relevant construction is not because X as such, but its three variants, which I have described in the subsections of Section 3.

62In the light of the observation that the word class membership of because is potential and emergent, the question arises as to why this is the case. In the following section, I will argue that the reason for this is the categorial vagueness of the connector because. Its category membership, which is dependent on its context of use, is ultimately not relevant to language users, but only to linguists.

4.4. Category membership as vague

63One aspect of linguistic categorisation and linguistic categories is the distinction between their ontological status for language users and for language analysts. However, neither as language users nor as language analysts do we ever have complete knowledge about the utterances we encounter (Denison [2010]; [2017]; [2018]). For illustration, let us consider the use of rubbish in (17) (Denison [2018: 136]). For some speakers of English, rubbish can function in such constructions both as an adjective and as a noun. In the former case, rubbish (17a) can be used as a synonym of prototypical adjectives such as bad or terrible (17b). In the latter case, rubbish (17a) can be used analogically to prototypical nouns such as nonsense or drivel (17c).

(17a) This is rubbish.
(17b) This is terrible.
(17c) This is drivel.

Theoretically, rubbish in (17a) could be categorised both as an adjective and as a noun, which highlights two points about the nature of linguistic categories and linguistic categorisation. First, as language analysts, we are dealing with a situation where we are unable to categorise rubbish as either a noun or an adjective unless we base this decision on an arbitrary criterion. Just as the proverbial Schrödinger’s Cat is simultaneously both dead and alive, the expression rubbish in (17a) is simultaneously both an adjective and a noun. Second, from the point of view of language users, this categorial indeterminateness is arguably inconsequential, according to Denison [2018: 136]. For the addressee, the question whether rubbish is an adjective or whether it is a noun does not play a role because the word class membership of the expression does not affect the meaning of the utterance. Moreover, if the category distinction does not make a difference on the part of the addressee, it cannot be utilised by the speaker to convey any meaning.

64This is an example of categorial vagueness (Denison [2017: 293]). A vague expression is an expression that is in some relevant aspect underdetermined, but this fact is of no consequence for its interpretation. The meaning of (17a) for the addressee remains the same regardless of whether rubbish was meant/interpreted as being more like an adjective (17b) or more like a noun (17c).

65Related to vagueness is the concept of ambiguity (Denison [2017: 293]). An ambiguous expression is in some relevant aspect also underdetermined. Unlike vague expressions, however, the fact that an ambiguous expression is underdetermined is of consequence to the language user. Classic examples are the so-called garden-path sentences such as the one in (18).

(18) The old man the boat.

66Both old and man are ambiguous. The word old can be categorised as an adjective or a deadjectival noun and the word man in turn as a noun or verb. If the addressee of (18) categorises old as an adjective and not as a noun and man as a noun and not as a verb, they will inevitably be led down the garden path and will experience the consequences of categorial ambiguity.

67The causal connector because does not show these effects in any of its uses discussed so far. The category membership of the connector is emergent and context-dependent. It is not ambiguous, but vague. A case in point is the construction because NP in (19).

(19) Wake up in the middle of the night crying because bad dream. (Twitter, 2019)

Depending on the criteria taken into consideration, because can be categorised as a preposition or a conjunction. Only considering the linguistic material in (19), the fact that because is complemented by a noun phrase and the matrix clause gives no reason to assume an elliptical structure, because can be categorised as a preposition (see Section 3.2.). Alternatively, given a suitable extralinguistic situation that would allow an analysis of (19) as a contextual ellipsis, because could also be categorised as a conjunction (see Section 3.1.).

68From the perspective of the language user, however, both categorisations are inconsequential since because used both as a conjunction and as a preposition still retains its core causal meaning. This causal meaning is expressed by all uses of because, unlike the different uses (and therefore categorisations) of old and man in (18).

69I have demonstrated that because in isolation, as an entry in the mental lexicon or constructicon, has the potential to be used as a conjunction, as a preposition, and as a member of a third category, those cases of because where the connector is followed by complements that rule out its categorisation as a conjunction or as a preposition. In many cases, the context of use disambiguates this category potential by providing clues determining the category membership of because. In (20a), because is unambiguously used as a subordinating conjunction and in (20b) as a preposition. In cases such as (20c), however, the context of use is insufficient to resolve the categorial under-specification – or the word class potential – of the causal connector because. The expression in these cases remains categorially vague.

(20a) She skipped breakfast because she was running late.
(20b) It is going to be blowout on this saturday, because djtripleks. (Twitter, 2019)
(20c) Wake up in the middle of the night crying because bad dream. (Twitter, 2019)

70In their treatment of recontextualisation and the role of context for language use, von Mengden & Kuhle [2020: 259] claim that “under-specification and context-dependence, are themselves, at any point in time, default characteristics of a linguistic expression of whatever kind.” This certainly holds true for the connector because. Prior to its use, it exhibits potential to be used as a conjunction, as a preposition or as a member of a third category. It is only when the connector is used in a specific context that its potential is realised. Its category membership is, in other words, context-dependent.

71In the light of my discussion of because in because X, I would slightly adjust the emergentist tenet above and say the following: all linguistic expressions are context-dependent, but not all of them are (equally) under-specified. Vague (and ambiguous) expressions are more under-specified than other expressions because vague expressions are not categorially specified by their context. It is precisely this categorial vagueness that enables language users to recontextualise because and gives rise to the various types of because X.

Summary and conclusions

72When we analyse a linguistic expression, we want to know what the expression is (like) and what it is not (like). This has been expressed in the dictum that to cognise is to categorise (Harnad [2005]). In terms of the specific analysis of because X I have presented in this paper, this means knowing in what ways the causal connector because is used like regular causal conjunctions and in what ways it is used like other word classes. I have shown that three types of because can be identified in because X constructions: because used as a conjunction, as a preposition, and as a member of a third category. At the same time, in some cases the context does not provide sufficient data to cognise unambiguously, which means that categorial vagueness is something that needs to be considered when we categorise.

73Not only is context essential for describing the three types of the because X construction mentioned above, but it also plays a crucial role in our understanding of categories. I have shown that the causal connector because does not have a set category membership. Without context, the connector has only a category potential. It can be used as a conjunction, it can be used as a preposition, and it can be used as a member of a third category. It is only through context, however, that this potential category membership becomes concrete.

Top of page


Aarts Bas, 2004, Conceptions of gradience in the history of linguistics, Language Sciences 26.4, 343–389.

Bailey Laura, 2012, “Because reasons”, Linguistlaura, (16 August 2022).

Baker Mark C., 2003, Lexical categories: Verbs, nouns, and adjectives, Cambridge: Cambridge University Press.

Becker Alton J., 1988, “Language in particular: A lecture”, in Tannen Deborah (Ed.), Linguistics in Context: Connecting Observation and Understanding. Lectures from the 1985 LSA/TESOL and NEH Institutes, Norwood, NJ: Ablex, 17-35.

Bergs Alexander, 2018, Because science! Notes on a variable conjunction, in Seoane Elena, Acuña-Fariña Carlos & Palacios-Martínez Ignacio (Eds.), Subordination in English, Berlin: De Gruyter, 43–60.

Bohmann Axel, 2016, “Language change because Twitter? Factors motivating innovative uses of because across the English-speaking Twittersphere”, in Squires Lauren (Ed.), English in computer-mediated communication. Variation, representation, and change, Berlin: De Gruyter, 149–178.

Burger Harald & Luginbühl Martin, 2014, Mediensprache. Eine Einführung in Sprache und Kommunikationsformen der Massenmedien, 4th edition, Berlin: De Gruyter, 173–200.

Carey Stan, 2013, “‘Because’ has become a preposition, because grammar”, Sentence First. An Irishman’s blog about the English language, (9 December 2017).

Coseriu Eugenio, 1974, Synchronie, Diachronie und Geschichte. Das Problem des Sprachwandels, München: Wilhelm Fink.

Croft William, 2001, Radical Construction Grammar, Oxford: Oxford University Press.

Croft William, 2013, “Radical Construction Grammar”, in Trousdale Graeme & Hoffmann Thomas (Eds.), The Oxford Handbook of Construction Grammar, Oxford: Oxford University Press, 211–232.

Denison David, 2010, “Category change in English with and without structural change”, in Traugott Elisabeth Closs & Trousdale Graeme (Eds.), Gradualness, gradience, and grammaticalization, Amsterdam: John Benjamins, 105–128.

Denison David, 2017, “Ambiguity and Vagueness in Historical Change”, in Hundt Marianne, Mollin Sandra & Pfenninger Simone E. (Eds.), The changing English language: Psycholinguistic perspectives, Cambridge: Cambridge University Press, 292–318.

Denison David, 2018, “Why would anyone take long? Word classes and Construction Grammar in the history of long”, in van Goethem Kristel, Norde Muriel & Coussé Evie (Eds.), Amsterdam: John Benjamins, 119–148.

Emonds Joseph E., 1976, A transformational approach to English syntax, New York, NY: Academic Press.

Farrell Patrick, 2001, “Functional shift as category underspecification”, English Language and Linguistics 5(1), 109–130.

Givón Talmy, 2001, Syntax. Volume 1, Amsterdam: John Benjamins.

Harnad Stevan, 2005, “To cognize is to categorize: Cognition is categorization”, in Cohen Henri & Lefebvre Claire (Eds.), Handbook of Categorization in Cognitive Science, Amsterdam: Elsevier, 19–43.

Harris Roy, 1991, “On redefining linguistics”, in Taylor Talbot J. & Davis Hayley G. (Eds.), Redefining linguistics, London: Routledge, 18–52.

Haspelmath Martin, 2007, “Pre-established categories don’t exist: Consequences for language description and typology”, Linguistic Typology 11.1, 119–132.

Hengeveld Kees, 1992, Non-verbal predication. Theory, typology, diachrony, Berlin: Mouton de Gruyter.

Himmelmann Nikolaus P., 1992, Grammaticalization and Grammar, Köln: Institut für Sprachwissenschaft, Universität zu Köln.

Hopper Paul J., 1987, “Emergent grammar”, in Aske Joe, Beery Natasha, Michaelis Laura & Filip Hana (Eds.), Berkeley Linguistics Society. Proceedings of the Thirteenth Annual Meeting, February 14-16, 1987: General Session and Parasession on Grammar and Cognition, Berkeley: Berkeley Linguistics Society, 139–157.

Hopper Paul J., 1988, “Emergent Grammar and the A Priori Grammar Postulate”, in Tannen Deborah (Ed.), Linguistics in Context: Connecting Observation and Understanding, Norwood, NJ: Ablex, 117–134.

Hopper Paul J., 2011, “Emergent Grammar”, in Gee James Paul & Handford Michael (Eds.), The Routledge Handbook of Discourse Analysis, London: Routledge, 301–314.

Kanetani Masaru, 2019, Causation and Reasoning Constructions, Amsterdam: John Benjamins.

Koch Peter & Oesterreicher Wulf, 1985, Sprache der Nähe - Sprache der Distanz. Mündlichkeit und Schriftlichkeit im Spannungsfeld von Sprachtheorie und Sprachgeschichte, in Deutschmann Olaf, Flasche Hans, König Bernhard, Kruse Margot, Pabst Walter & Stempel Wolf-Dieter (Eds.), Romanistisches Jahrbuch, Berlin: Walter de Gruyter, 15–43.

Konvička Martin, 2018, Want/omdat X en de vaagheid van de linguïstische categorieën, Neerlandica Wratislaviensia 28, 17–31.

Konvička Martin, 2020, Protože změna: K české kauzální konstrukci protože X, Naše řeč 103.3, 243–263.

Konvička Martin & Stöcker Kristin, 2022, “(Non-)Ellipses in Dutch, English, and German: The case of because X”, Nederlandse Taalkunde 27.3, 333–367.

Landert Daniela & Jucker Andreas H., 2011, “Private and public in mass media communication: From letters to the editor to online commentaries”, Journal of Pragmatics 43.5, 1422–1434.

Lemmens Marcel, 1991, Want-zinnen. Mag het ietsje minder zijn?, Onze Taal 59, 14–15.

McCulloch Gretchen, 2014a, “Why the new “because” isn’t a preposition (but is actually cooler)”, All things linguistic, (16 July 2019).

McCulloch Gretchen, 2014b, “Where “because noun” probably came from”, All things linguistic, (16 August 2022).

von Mengden Ferdinand & Kuhle Anneliese, 2020, “Recontextualization and language change”, Folia Linguistica 54.41, 253–281.

Pullum Geoffrey K., 2009, “Lexical categorization in English dictionaries and traditional grammars”, Zeitschrift für Anglistik und Amerikanistik 57.3, 255–273.

Pullum Geoffrey K., 2014, “Because syntax”, Language Log, (16 July 2019).

Pullum Geoffrey K., 2014, “The promiscuity of prepositions”, Language Log, (16 August 2022).

Rijkhoff Jan & van Lier Eva Helena (Eds.), 2013, Flexible word classes: Typological studies of underspecified parts of speech, Oxford: Oxford University Press.

Robins Robert H., 1966, “The Development of the Word Class System of the European Grammatical Tradition”, Foundations of Language 2.1, 3–19.

Rosch Eleanor H., 1973, “Natural categories”, Cognitive Psychology 4.3, 328–350.

Schachter Paul & Shopen Timothy, 2007, “Parts-of-speech systems”, in Shopen Timothy (Ed.), Language typology and syntactic description. Volume I: Clause structure, 2nd edition, Cambridge: Cambridge University Press, 1–60.

Schnoebelen Tyler, 2014, “Innovating because innovation”, Corpus Linguistics, (22 February 2022).

Selting Margret, 1997, “Sogenannte ,Ellipsen’ als interaktiv relevante Konstruktionen? Ein neuer Versuch über die Reichweite und Grenzen des Ellipsenbegriffs für die Analyse gesprochener Sprache in der konversationellen Interaktion, in Schlobinski Peter (Ed.), Syntax des gesprochenen Deutsch, Opladen: Westdeutscher Verlag, 117–155.

Stefanowitsch Anatol, 2014, “Weil ist faszinierend, weil Sprachwandel”, Sprachlog, (8 February 2022).

Stöcker Kristin, 2020, “Because X in Dutch, English, and German”, GitHub, (18 February 2022).

Taylor John R., 1995, Linguistic categorization: Prototypes in linguistic theory, 2nd edition, Oxford: Clarendon Press.

Van Oostendorp Marc, 2014, “Want boos/verdrietig/bekaf”, (8 February 2022).

Weber Tilo, 1997, “The Emergence of Linguistic Structure: Paul Hopper’s Emergent Grammar Hypothesis Revisited”, Language Sciences 19.2, 177–196.

Wessman Kukka-Maaria, 2017, “Rating the acceptability of non-standard language. How Finnish language users rate variants of the verbless koska X ‘because X’ internet meme construction?”, Poster presented at the 14th International Cognitive Linguistics Conference, Tartu, Estonia.

Whitman Neal, 2013, “Because as a Preposition”, Quick and Dirty Tips, (16 August 2022).

Zimmer Benjamin, Solomon Jane & Carson Charles E., 2014, “Among the new words”, American Speech 89.1, 89–110.

Top of page


1 I use the term connector or causal connector as a functional description of because without any commitment to its word class membership.

2 I use square brackets in the examples to signal the (potentially) recovered parts of an elliptical structure.

3 The idea that English prepositions and conjunctions in fact form a single category distinguished only by their complements – nominal or clausal – goes back to Emonds [1976: 172f.]. The main problem with this analysis, however, is its object. It is an analysis of expressions as such, while the object of the present analysis are expressions in specific constructions. For a discussion see Croft’s [2001, especially chapter 2.3] treatment of the differences between the so-called lumping and splitting approaches to parts of speech categorisations.

4 Both references contain early accounts of Hopper’s concept of Emergent Grammar. Based on a chronological perspective, Emergent Grammar [1987] is usually taken as the primary reference, because the second text, Emergent Grammar and the A Priori Grammar Postulate [1988], was published one year later. Upon closer scrutiny, however, we find out that the latter article appeared in the proceedings of a conference that had already taken place in 1985.

5 Although the concept of Emergent Grammar can be traced back to Hopper, its underlying principles can be found in earlier works (see, e.g., Coseriu [1974]; Becker [1988]; Harris [1991]) and even outside of linguistic (see, e.g. Weber [1997] for a discussion of the links between Emergent Grammar and Jacques Derrida’s deconstructivism).

6 This terminology goes back to Hengeveld [1992] who proposed a distinction between flexible and rigid words. Whereas the latter category describes those parts of speech that are specialised in one particular function, the former category refers to expressions that can have multiple functions.

7 Precursors of this radical constructionist position can be found in some of the early works on grammaticalisation. Himmelmann [1992: 21–22], for instance, acknowledges the primacy of constructions and the fact that word classes such as nouns are merely “convenient cover label[s]”. Moreover, he also stresses the fact that word classes primarily exist on the ontological plane of the language analysts, i.e. of the linguists. The ontological reality of word classes for the language user remains an empirical question.

Top of page


Electronic reference

Martin Konvička, Category membership and category potential: The case of vague becauseLexis [Online], 22 | 2023, Online since 29 December 2023, connection on 25 February 2024. URL:; DOI:

Top of page

About the author

Martin Konvička

Freie Universität Berlin, Germany

Top of page



The text only may be used under licence CC BY-SA 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search