Idioms and other constructions in American Sign Language

3.1 “Where are all the idioms?”

Section 2 proposes a cross-linguistic definition of idiom, developed in comparison to the related concepts prefab and snowclone. Treating idiom as a comparative concept lays a foundation for identifying idioms in a sign language like ASL, the topic of this section.

In sign language linguistics, there are two prevailing views of idioms. The first view is held by many ASL teachers and is found in their teaching materials. It is that idioms are signs that are difficult to translate from ASL to English or may require a full phrase to be expressed in English. This view conflates idiom with the concepts of slang and polysemy. Baker-Shenk and Cokely (1980: 119) noted this definitional tendency, which persists over 40 years later: “Most of the expressions which have been called “idioms” in descriptions of ASL are not really idioms. They do not fit the definition of an idiom. For example, the signs FINISH, NONE/ZERO, AT-LAST/SUCCESS are not idioms. They are simply frequently used signs which have several different uses.” Calling these single signs idioms, they argue, “is like saying that the English word run is an idiom since it has so many different meanings” (Baker-Shenk and Cokely 1980: 199).

The three ASL signs that Baker-Shenk and Cokely mention, pictured in Figure 1, are conventional meaning-form pairs in ASL. This quote illustrates one common glossing practice in sign language linguistics, which is to identify sign forms with uppercase, meaning-based glosses. Here I also use meaning-based glosses to refer to ASL signs, though I reserve uppercase type for more schematic elements in a constructional unit, as already used in Section 2, and for grammatical abbreviations (following the Leipzig glossing conventions: https://www.eva.mpg.de/lingua/resources/glossing-rules.php). Accordingly, the ASL signs in Figure 1 are glossed here as [finish], [none], and [success].

Figure 1:

The ASL signs [finish], [none], and [success] (left, center, and right, respectively; images from ASL Signbank 2025; https://aslsignbank.haskins.yale.edu/).

As Baker-Shenk and Cokely note, these sign forms have several different uses in ASL that are not captured by the English glosses. The sign [finish] is not only used as a verb (‘to finish’), but this form has also been grammaticalized as a completive and a perfective auxiliary, among other functions (Janzen 1995). The sign [none] is used to describe a quantity (as in ‘no money’) and for related types of negation (as in ‘haven’t seen anything at all’). The sign [succeed] is formed with the hands tracing an arc out and away from the signer’s face, but when it is also accompanied with the mouthing “pah!”, this sign emphatically celebrates that something has been ‘finally accomplished at last’.

Highly conventionalized meaning-form pairings such as these are often referred to as “lexical” or “fully lexical” signs in the sign language literature (e.g., Ferrara et al. 2023). For linguists who use this term, “lexical sign” is a category that contrasts with “non-lexical” or “partly lexical” signs, which must rely on contextual factors to “show” what they mean in some way. In practice, the primary difference between a lexical and a non-lexical sign is the degree of conventionality of the unit. A “lexical” sign is a form whose meaning is agreed upon primarily by convention. In contrast, the specific meaning of a “partly-lexical” sign is determined primarily by context, because it involves a mix of conventionalized and less conventionalized elements (see Enfield 2009). ASL pronouns, for example, physically point to their referents or to spatial locations that have been associated with their antecedents (Wilcox and Occhino 2016). An ASL pronoun is considered “partly lexical” because it incorporates the context of the physical space and preceding discourse to zero in on the referent.

This leads to the second view of idioms in sign languages, that “lexical” signs are idioms (Johnston and Ferrara 2012). This can be seen in Johnston and Schembri’s (2010: 31) argument that “the idiosyncratic meaning of a fully-lexical sign suggests that lexicalization in SLs is similarly ‘idiomatic’. Lexical signs are in a sense idioms”. Under this view, idiomaticity and conventionality are conflated: because they have fixed forms and involve conventional, learned meanings, lexical signs are considered a type of idiom.

Both of these views of single signs as idioms are rooted in the fact that, descriptively, few phrasal idioms have been identified and documented for any sign language to date. ASL famously has at least one idiom, which I refer to here as [train gone (sorry)] and which somewhat overlaps with the meaning of the English idiom missed the boat. In ASL, this phrase is used to brush off a latecomer to a conversation, as a way of saying “I’m not going to catch you up on what you’ve missed”. Like a train that has already left the station, the conversation is too far gone to join.

Figure 2 shows the three signs that make up this idiom, [train], [gone], and [sorry]. As indicated by the parentheses, in this idiom the final sign [sorry] is less fixed than the signs [train] and [gone], and is sometimes omitted. The sign [train] is formed with two fingers on the signer’s dominant hand rubbing back and forth over the fingers of the non-dominant hand, the sign [gone], in this context, is formed with the dominant hand moving away from the non-dominant hand with the fingers simultaneously closing, and the sign [sorry] is formed with the dominant hand tracing a small circle in the center of the signer’s chest.

Figure 2:

The signs that make up the ASL idiom [train gone (sorry)] ‘you’ve missed out on the conversation’ (images from ASL Signbank 2025; https://aslsignbank.haskins.yale.edu/).

Interestingly, the ASL idiom [train gone (sorry)] has also become a signifier of American deaf culture, in written English. In this respect, it is like the ILY handshape (, with thumb, index, and little finger extended), which means ‘I love you’ but is used as a friendly greeting by ASL signers. Both the ILY handshape and the phrase train gone sorry can be found in commercial and cultural products such as comics, GIFs, books, and shows that relate to ASL and shared deaf experiences, as a way to evoke the cultural frame of deafness and sign language use in America. Figure 3 shows a streaming series poster and a book cover for works that center on members of the American deaf community, and both are titled with variations of the English transliteration train gone sorry.

Figure 3:

An ASL idiom has become a signifier of American deaf culture, in English: poster for the 2020 streaming series TRAIN GONE SORRY (image from https://www.imdb.com/title/tt11905308/) and cover of the 1995 nonfiction novel Train Go Sorry: Inside a Deaf World (Cohen 1995; image from https://www.leahhagercohen.com/train-go-sorry).

In short, the expression [train gone (sorry)] is an ASL idiom, following the comparative concept definition from Section 2. It is affective, as it signals an unforgiving brush off, while also communicating that information has been missed. It is a phrasal unit, consisting of two to three distinct signs. It is also conventional, not only because it has a specialized meaning (‘you’ve missed out on the conversation’), but also because this phrase is so well-known that its written English transliteration has come to signify ASL itself.^[2]

A very similar idiom has been found in Auslan, the national sign language of Australia: Johnston and colleagues (Johnston and Schembri 2010: 30; Johnston and Ferrara 2012: 234) cite [you miss train] as a lone example of an Auslan idiom. This fact leads Johnston and Ferrara (2012: 236) to ask, “where have all the idioms gone?” From a cognitive and particularly a constructionist perspective, it would be unusual that idioms figure so predominantly in spoken language linguistics, but are virtually unattested in sign languages. Where are all the idioms, indeed.

Lepic (2019: 7–8) speculates that a descriptive lack of idioms, in ASL, at least, could be due to a number of factors. First, English is an old language with a long literary tradition, whereas ASL is comparatively younger and lacking a standardized or widely used writing system. It is possible that factors such as language age and absence of a standardized written modality could influence the conventionalization of idioms and other prefabs (e.g., Cutler 1982; Corson 1997). Second, sign languages like ASL have different patterns of transmission compared to nationally dominant spoken and written languages like English (e.g., Mitchell and Karchmer 2004; Morford and Hänel-Faulhaber 2011). English is found everywhere in the US, and even people who use a different home language are inevitably immersed in English in popular media, school, and work. In contrast, ASL is a minoritized language in the US, and it is uncommon for deaf infants to be born into families with caregivers who already know ASL, and who themselves learned ASL from signing caregivers (see Chen Pichler et al. 2019). Many ASL users learn ASL after childhood, and encounter ASL in more restricted social settings than English. These transmission patterns may also affect the development of idioms and other prefabs among users of a language. It also may be that English actually relies on idioms and prefabs more than other languages do, for whatever reason. However, Lepic (2019) also indicates that it is premature to conclude that sign languages like ASL lack idioms, since there have not yet been many systematic efforts to find and document them.

3.2 Evaluating ASL vocabulary in videos online

The remainder of this section analyzes a handful of ASL videos posted online for idioms and other vocabulary, following Hou et al. (2020)’s procedure for working with ASL internet data. At the time of writing (2023–2024), all videos discussed in this section are publicly available online, without any login credentials or permissions required, and all appear to be intended for a wider viewership, rather than as personal posts. I evaluate the ASL vocabulary provided in these videos for affective stance, conventionality, and especially phrasal complexity, in order to discuss their potential status as ASL idioms.

Four “ASL slang” videos provide the basis for this exploration. As mentioned in Section 3.1, in ASL teaching materials it is common to find lists of “ASL idioms” that include slang or other single-sign vocabulary. A search for “ASL idioms” using Google’s video search function yields dozens of videos posted online between approximately 2015 and 2022. Two instructional videos were selected for analysis, one in which an ASL instructor is providing ASL vocabulary from a prepared list and another with an ASL signer discussing the concept of idioms and providing some examples. In addition, two shorter videos using a similar format were found searching for “ASL urban dictionary” on Google. These latter videos also include an example context for each expression, as well. The URLs used to access the four videos, along with some additional descriptive information, are provided in Table 1.

Altogether, the four videos demonstrate 100 unique ASL expressions, five of which are also repeated in two different videos. Table 2 provides the full list, using the English labels provided by the signers in each video. The English labels for the five ASL expressions that are repeated across videos are underlined and the four English labels that are recycled with different ASL expressions are indicated with “-2”. In anticipation of the discussion in Section 3.3, ASL examples formed with multiple distinct signs are also emphasized in bold in Table 2. Finally, note that some of the English labels in Table 2 are not transparent to those who do not know ASL. For example, the label “athe” (for the sign meaning ‘ugh, I don’t want to’) matches the mouth movement and protruding tongue that accompany the manual sign, while the label “c fist” (for the sign meaning ‘screw them!’) is a partial description of how the sign is articulated, with the non-dominant hand forming a C handshape to grab the dominant arm while raising a fist (see Figure 4). These labels illustrate the metalinguistic knowledge that ASL signers have about how written English can be used to represent ASL sign forms. In the discussion below, I replace the English labels in Table 2 with a meaning-based gloss for each identifiable ASL sign in an expression, as needed.

Figure 4:

The label “athe” for the single sign meaning ‘ugh, I don’t want to’ evokes the tongue protrusion during the sign (left; image from https://youtu.be/I2OjqgWcrSc?t=16); the label “c fist” for the single sign meaning ‘screw them!’ evokes the configuration of the hands in the sign (right; https://youtu.be/I2OjqgWcrSc?t=64).

Though most of the English labels in Table 2 involve multi-word phrases, the majority of the 100 ASL vocabulary items in the four videos can be considered single signs (n = 52). These signs are formed with the hand or hands configured in a single handshape, performing either a single movement or repetitions of the same movement (Stokoe 2001). For example, the single sign [keep.quiet], identified as “that’s not what I’m talking about” in Table 2, is formed with the signer’s dominant index finger moving to contact the mouth in a single movement (Figure 5). This sign has a variety of uses in ASL, often used to mean ‘that’s not what I meant’ and ‘I didn’t say anything’, in addition to ‘shh, keep quiet’.

Figure 5:

The single ASL sign [keep.quiet] is formed with one handshape (the index finger) executing one movement (moving to contact the lips) (image from ASL Signbank 2025; https://aslsignbank.haskins.yale.edu/dictionary/gloss/160/).

The remaining items (n = 48) vary in their symbolic complexity. Some consist of multiple identifiable signs (n = 12), and the rest fall somewhere between a “single sign” or a “combination of two signs” (n = 36), due to their analyzable internal structure. For example, the sign [think-appear] (meaning ‘thought of something’ or ‘came to mind’) could be analyzed as a combination of two signs, [think] and [appear], as is shown in Figure 6. However, as a unit, it has also undergone some phonological restructuring: the transitional movement of the former phrase has become the primary movement of the resulting unit (Liddell and Johnson 1986; Wilkinson 2016). Signed in a sequence, the signs [think] and [appear] require two primary movements, and [think-appear] has only one. Due to these changes in movement, and also due to the fact that both [think] and [appear] are signed with the same handshape, the sequential structure of this unit is partially obscured.

Figure 6:

The ASL sign meaning ‘come to mind’ can be analyzed as a sequential combination of [think] and [appear], though it has also undergone some phonological restructuring as a unit (images from https://youtu.be/MynzxdkfxfE?t=62).

The phonological structure in [think-appear] differs from that of [train gone (sorry)], which has three distinct signs in a sequence, each with their own hand configurations and primary movements. This difference in phonology also reflects the symbolic structure of these two examples. Many conventional ASL signs begin with the index finger at the forehead and relate to cognition or mental activity, suggesting that what is glossed here as [think] may actually be a component of a distinct construction. This construction is discussed more in Section 4.2.

Other “ASL slang” expressions are also phonologically complex, but the identifiable elements of the unit are less easily categorized as autonomous meaning-form pairings on their own. The sign [trivial], meaning ‘it’s no big deal for me’ (identified in Table 2 as “easy as pie” and “piece of cake”), is formed with the dominant hand raising a finger to the mouth and then closing to thump a fist against the chest twice (Figure 7). The form of this expression suggests that it may contain two words: it is articulated with a change in handshape from an extended index finger to a closed fist, the hand changes location from the mouth to the chest, and the second half of the sign is made with its own repeated movement. These two identifiable elements partially resemble the conventional ASL signs [keep.quiet] and [cough], though there is no transparent relationship between these elements and the meaning of the whole unit, if indeed they are etymologically related at all.

Figure 7:

The ASL sign meaning ‘trivial’ is phonologically complex, but is difficult to analyze as a sequential combination of autonomous signs (images from https://youtu.be/MynzxdkfxfE?t=137).

As a final example of a “borderline” case, the sign [word.on.street] also seems to contain two parts, the second of which is not a conventional ASL sign on its own (cf. “cranberry morphs”, Booij and Hüning 2014). The sign [word.on.street] begins with the index finger of the dominant hand contacting the ear, resembling the sign [sound], and then both hands shake slightly and move in an arc in front of the signer’s body (Figure 8). This second element can be analyzed as meaning something like ‘here and there’ or ‘going around’, in the context of the whole sign, but this two-handed shaking form is not used on its own outside of the sign [word.on.street], to my knowledge.

Figure 8:

The ASL sign meaning ‘word on the street’ is phonologically complex, but is difficult to analyze as a sequence of autonomous signs (https://youtu.be/UwobS6Gn0ss?t=272).

As a summary of the preceding discussion, and before turning to more sequentially complex examples in Section 3.3, Table 3 categorizes 12 selected examples from Table 2 according to their relative degree of analyzable symbolic structure. This table includes the five items that were repeated in two different “ASL slang” videos: [train gone (sorry)], [think-appear], [trivial], [dont.feel.like], and [outrageous]. Some glosses include angled brackets that describe non-segmental morphological complexity, such as changing the movement or location of another sign. The sign [arrogant], for example, is a single sign formed similarly to the sign [big], but the sign is relocated to the signer’s head. These signs are discussed in Section 4.2.

3.3 Looking for ASL idioms

Though the majority of the expressions in the “ASL slang” videos are single signs, there are phrasal units among these examples, as well. By virtue of appearing in videos designed to teach and share ASL vocabulary, the examples in these videos can all be considered conventional, at least to some degree, among ASL signers. In addition, some of these expressions convey affective meaning, in the broad sense indicated in Section 2, taking an emotional and attitudinal stance along with the propositional content of the utterance. These are candidates for ASL (“decoding”) idioms. The two most promising examples of idioms in this dataset are [keep.quiet hard] ‘just have to accept it’ (Figure 9) and [dry discuss] ‘end of discussion!’.

Figure 9:

The ASL idiom [keep.quiet hard] ‘just have to accept it’ (images from https://youtu.be/I2OjqgWcrSc?t=48).

The idiom [keep.quiet hard] (identified in Tables 2 and 3 as “opt hard”) contains two identifiable signs, each articulated with a different handshape and its own primary movement. This idiom conveys a frustrated resignation on the part of the signer; when you realize that you’ve made a mistake or will not get your way, you just have to shut up and accept it as a hard lesson. In Figure 9, some of the affective force of this idiom, particularly the signer’s depiction of resignation, also comes from the signer’s eye gaze and tightened lips (Siyavoshi and Wilcox 2021).

The idiom [dry discuss] (identified in Table 2 as “dry hotdog”) also contains two identifiable signs in a sequence. It is used to end a conversation with annoyance and impatience, declaring that a topic has already been discussed and decided, and nothing more can be wrung from it (Figure 10). Again, the signer’s facial expression in Figure 10 contributes to the symbolic structure of the construction, with the narrowed eyes, in particular, reflecting the signer’s assessment that the topic is unoriginal (Dachkovsky et al. 2013). Similar to prosody in spoken language, the non-manual elements of these constructions represent a simultaneous rather than sequential layering of meaning in signed discourse.

Figure 10:

The ASL idiom [dry discuss] ‘end of discussion!’ (images from https://youtu.be/9tkLIN6WEFA?t=56).

In contrast to these two idioms, other conventionalized phrasal units from the “ASL slang” videos seem to involve more “regular” ASL vocabulary. They have a somewhat diminished stance-taking function, behaving more like prefabs in ASL. Two examples are [money rake.in] ‘get funds’ and [finish touch] ‘have visited’ (Figures 11 and 12). Rather than pragmatically marked expressions, these phrasal units are the conventional ways to express the concepts they denote in ASL, and are noteworthy because they must be learned and used as fixed phrases. The construction [money rake.in] is conventionally used to describe a variety of events involving large amounts of money, such as being awarded a grant. The construction [finish touch LOCATION] contains a schematic slot and is commonly used to discuss whether a person has visited a given location before, in ASL conversation.

Figure 11:

The ASL prefab [money rake.in] ‘get funds’ (images from https://youtu.be/MynzxdkfxfE?t=50).

Figure 12:

The ASL prefab [finish touch LOCATION] ‘have visited LOCATION before’ (images from https://youtu.be/UwobS6Gn0ss?t=314; only the two fixed signs are demonstrated).

In addition to these four examples, Table 4 identifies the sequential structure for the remainder of the 12 phrasal “ASL slang” examples that contain multiple distinct signs. Not included in this list are examples like [think-appear], mentioned as “borderline” units in Section 3.2, which start with the index finger contacting the forehead and relate to cognition. These expressions are ambiguous between one and two signs. However, there are several such signs in this dataset, and, as already mentioned, they are discussed in Section 4.2.

Though the idiom [train gone (sorry)] and the prefab [finish touch] will likely be known to the majority of ASL signers, examples such as [dirty work] and [keep.quiet have plan] seem to be somewhat less commonly used, from my own observations. These may turn out to be regional or generational phrases, rather than “ASL-wide” conventions, or they may be particular to specific genres of ASL use. These possibilities cannot be addressed without further targeted research (e.g., Carrol 2023; Carrol and Segaert 2024), but the fact that these are featured in these “ASL slang” videos suggests that they are recognized by at least some ASL signers as conventional units.

For the most part, the few phrasal expressions in this small “ASL slang” dataset contain two signs. It is worth asking whether linguists can expect to find longer idioms and prefabs in ASL, as appropriate descriptive materials become available (e.g., Hou and Morford 2020, Hou 2022, Hochgesang et al. 2023). I suspect that, yes, there are longer ASL idioms yet to be discovered, still. Given that English idioms come from popular culture, including movies, songs, and literature, ASL idioms might eventually be identified in these domains as well.

In the past decade or so, ASL has undergone a shift in its communicative ecology. Traditionally, ASL is used in face-to-face conversation, and physical institutions such as regional schools for the deaf have served an essential role in ASL socialization (Padden and Humphries 2005). With the increasing availability of video recording and sharing in the internet age, ASL signers also increasingly experience ASL in videos and online (Hou et al. 2020; Lucas et al. 2013). At the same time, the visibility and popularity of ASL have risen in mainstream American culture (see Hou and Robinson 2020, Saunders 2023). In addition to the materials pictured above in Figure 3, ASL-signing characters have been featured in scripted media such as the television shows Switched at Birth (2011–2017) and Hawkeye (2021), and movies such as The Eternals (2021) and CODA (2021), which won an Academy Award in 2022. One result of this general societal trend is that language practices that might previously have been restricted to smaller communities of ASL signers occupying the same physical region can now be shared and experienced more broadly, though the internet.

Looking at internet discourse as one domain of contemporary ASL use, I have identified two cultural references that illustrate what the beginnings of the process of idiom emergence may look like. These two examples come from videos that are intended for the purposes of entertainment and information-sharing, rather than for metalinguistic discussion of ASL vocabulary. I have selected these examples because I have seen them referenced in social media posts online, but I do not have evidence that they are becoming highly conventionalized units known to the majority of ASL signers. Instead, I consider these internet references as memes that illustrate an early stage of the conventionalization process (Dancygier and Vandelanotte 2017; Hartmann and Ungerer 2023).

The first example is the phrase [hearing know best (right)] ‘hearing people know best, indeed!’ (Figure 13). This phrase is a reference to an ASL video posted online in 2017 with over 53,000 views on YouTube in 2024 (https://www.youtube.com/watch?v=MoxVdw6T0LA). The video is a sardonic, music-video style commentary on who is empowered to make decisions for deaf individuals. The phrase [hearing know best] is repeated throughout the video, satirizing the fact that deaf people are usually not allowed autonomy over their own lives and affairs. Better get a hearing person to make the important decisions, because “hearing people know best,” after all.

Figure 13:

An ASL reference: [hearing know best (right)] ‘hearing people know best, indeed!’ (images from https://youtu.be/MoxVdw6T0LA?t=35).

The second example is the phrase [let.you.know NAD sue] ‘the NAD is suing!’ (Figure 14). This reference comes from a press release posted online in 2020 (https://www.youtube.com/watch?v=0kRjZQfprTE). Unlike the video “hearing know best”, the tone of this video is serious. The press release is delivered by the then-CEO of the National Association for the Deaf (NAD), summarizing their lawsuit to compel the White House to provide ASL interpreters during their COVID-19 press conferences.

Figure 14:

An ASL reference: [let.you.know NAD sue] ‘the NAD is suing!’ (images from https://youtu.be/0kRjZQfprTE).

Though I do not have video evidence of these ASL phrases being used in ASL discourse, the written English phrases “hearing know best” and “NAD sue” can be found in written English posts shared between ASL users on various social media platforms online. GIFs created from these videos have also been uploaded to the popular GIF-sharing website Tenor (https://tenor.com/view/hearing-knows-best-gif-25369575; https://tenor.com/view/nad-sue-gif-24926212). These GIFs capture the ASL phrases in an animated image, and their availability on a GIF-sharing platform suggests that these ASL references have been quoted and shared digitally among at least some ASL users. The social media posts that I have observed using these memes are typically joking complaints about small, personal inconveniences suffered by deaf people. In these contexts, the English phrases “hearing know best” and “NAD sue” and the corresponding ASL GIFs serve as in-group jokes, extending references from the original videos to new contexts.

The source videos for these two memes differ markedly in their tone and intent, with “hearing know best” coming from an ironic music video and “NAD sue” coming from a sober press release. However, both these English phrases and ASL GIFs have been referenced in other contexts online. It seems likely that, as memes, these references are fated to fade out of use rather quickly. Nevertheless, these memes illustrate the possibility of phrases larger than two signs becoming conventional among ASL signers. As such, they show part of process through which a complex phrase may become conventionalized: language users who share in a particular linguistic experience can refer to that experience by quoting it in other contexts, and these repeated references allow the phrase to be further shared and used among other members of the linguistic community. With use, a repeated phrase may eventually drift from the source reference, becoming an increasingly autonomous construction in its own right.

4.1 An ASL “constructicon”

The primary goal of this paper thus far has been to identify idioms and prefabs as conventional phrasal units in ASL. Section 3 identifies a number of ASL expressions taken from ASL videos online, and from these examples it is clear that conventional ASL vocabulary items also include units larger than one or two signs. At the same time, the majority of conventional units that can be listed as fixed units seem to be quite short sequentially. This section describes a constructionist model of grammatical knowledge in ASL to account for these facts, and Section 4.2 revisits “morphological” examples mentioned in Section 3, to describe their constructional properties.

As discussed in Section 2, in usage-based theories of grammatical knowledge, a language user’s experience forms a structured inventory of constructions. Goldberg (2003: 223) famously captures this idea with the declaration that “it’s constructions all the way down”. In other words, linguistic knowledge is theorized to consist entirely of meaning-form pairs that vary in their size and fixedness (Langacker 2013; Dąbrowska 2014). Accordingly, a constructed, decontextualized English sentence like The window was broken in (2) can be analyzed as instantiating several component constructions, some of which are listed below:

Constructions are emergent patterns from a rich network of linguistic experiences often called the “constructicon” (Goldberg 2019). Following Croft and Cruse (2004: 255) and Goldberg (2013: 17), the variation in size and fixedness among constructions can be modeled as in Table 5. In this representation, the schematic category “word class” is an emergent generalization over the specific constructions in the “word” category. Similarly, the category “partially-fixed word” is an emergent generalization over the specific constructions in the “complex word” category. The exact number of construction types and the phrasing of their properties varies according to the researcher and the constructions under discussion, but the point is that any aspect of linguistic knowledge could be situated in the constructicon as a construction, according to the degree of symbolic complexity and schematicity it displays.

In order to extend this same approach to ASL, Figure 15 illustrates one example sentence from the four “ASL slang” videos analyzed in Section 3. This ASL sentence provides a constructed context in which a signer might use the idiom [keep.quiet hard] ‘just have to accept it’: “I got an F because I didn’t want to study, and I just had to accept it”. In this complex utterance, several shifts in facial expression can be observed, each coding displays of disengagement, surprise, and resignation (Janzen 2019; Siyavoshi and Wilcox 2021). Only two of these displays have been annotated in Figure 15, both because these are the most straightforward to gloss with a familiar grammatical abbreviation, and also to keep the glossing legible. These facial displays are the tilted head and wide eyes for marking a topic, [1SG study]<TOP>, and the disengaged facial expression signaling a short stretch of constructed discourse, [later dont.have.to study, think nothing]<QUOT>.

Figure 15:

An ASL utterance illustrating the use of the idiom [keep.quiet hard] ‘just have to accept it’: When studying, I was like, “ugh, I can do it later, I don’t have to study, no big deal”. But then I got an F! Well, I had to just shut up and accept it. (images from https://youtu.be/I2OjqgWcrSc?t=51; <TOP> indicates topic-marking and <QUOT> indicates constructed discourse.).

This utterance can be analyzed into several component constructions, including specific sign constructions and more schematic constructions. As is also the case in analyzing spoken language constructions, glossing presents a bit of a challenge for discussing schematic elements of form or meaning among signed constructions. Accordingly, Table 6 describes the form and meaning for a few constructions instantiated by the utterance in Figure 15. Each construction is given a name, a description of its fixed elements of form, and a short description of its meaning.

Table 6:

Some component constructions from Figure 15.

Construction	Description of form	Description of meaning
study (see Figure 15 f )	two-handed sign, the dominant hand repeatedly approaches the open palm of the non-dominant hand	‘study’
get.an.F (see Figure 15 j)	two-handed sign, the dominant hand forcefully strikes the open palm of the non-dominant hand	‘get an F’, ‘fail’
the WRITTEN Construction (Figure 15 f and j)	open.palm: two-handed signs in which the non-dominant hand is an open palm and the dominant hand articulates on the non-dominant hand; the dominant handshape and movement vary across signs	‘written text’: concepts that metonymically involve visible text on a flat surface, such as ‘reading’, ‘writing’, ‘learning’, and ‘evidence’
the Topic- Comment Construction (Topic: Figure 15 a and b Comment: Figure 15 c–h)	TOPIC<TOP>, COMMENT: Topic phrase signed with raised eyebrows, wide eyes, tilted head, followed by a change in face and body posture for the comment	establishing a topic and then making a focal comment about the established topic
the Subject- Predicate Construction (Subject: Figure 15 a Predicate: Figure 15 b)	(SUBJECT) PREDICATE	making a claim about a state or process, whose primary participant is either overtly stated or inferred from context

Five example constructions are listed in Table 6, two of which are conventional signs as fixed meaning-form pairings: [study] and [get.an.F]. Also listed is a schematic morphological construction partially describing the symbolic structure of these signs: the WRITTEN Construction, [‘written text’/open palm]. This schematic construction is a generalization over signs like [study] and [get.an.F], as well as other conventional signs like [write] and [lesson] (Figure 16). These signs all conventionally refer to concepts related to information and evidence. All are also formed with the non-dominant hand configured as an open palm that the dominant hand articulates against.

Figure 16:

Many ASL signs referring to text and information, including [write] (left) and [lesson] (right), are formed with dominant hand articulating on the open palm of the non-dominant hand (images from ASL Signbank 2025; https://aslsignbank.haskins.yale.edu/).

The WRITTEN Construction captures the fixed aspects of meaning and form that can be extracted from fixed signs like [study] and [write]. This construction is not hypothesized to “derive” these conventional words each time they are used, but rather is meant to capture an ASL user’s knowledge that these conventional words have aspects of meaning and form in common (Booij 2010). This morphological knowledge about patterns among ASL signs can be taken advantage of in connected discourse or in the creation of new signs (cf. Aronoff et al. 2005: 75; Lepic and Occhino 2018; Lepic 2021).

On this point, following a suggestion from a reviewer, it is necessary to make a brief side comment about an additional type of construction that is often found in ASL discourse, though it is not represented in the decontextualized data that have been analyzed in this article. ASL signers, like users of other (sign) languages, often shift between different “modes” of signing in discourse (Dudis 2011; Capirci, Bonsignori and Di Renzo 2022; Ferrara and Hodge 2018; Hodge and Ferrara 2022). In a more descriptive mode, signers communicate using primarily conventional signs, as in the example in Figure 15. In a more depictive mode, signers communicate in ways that invite the viewer to see the iconic and symbolic potential of the signs they produce. Under the present analytic approach, this second type of signing would be analyzed as making productive use of schematic constructions such as WRITTEN Construction identified in Table 6, in which, for example, the flat hand is conventionally used to represent a surface of written text. However, as theorists in the Semiological Approach have demonstrated, these different modes of signing are not signaled by the structure of the manual signs themselves (Cuxac 1999; Garcia and Sallandre 2020). Instead, they argue that in discourse, signers move between simply saying and saying by showing through their use of eye gaze to indicate their communicative intent (Garcia and Sallandre 2020; see also Janzen 2019). By looking within the space around them, including their hands, signers call attention to the iconic potential to be found in a sign. This approach to depictive signing seems entirely correct to me. However, since this type of depictive signing is not found in the data I analyze here, I restrict my analysis to only identifying morphological schematic constructions among signs, for the time being.

In addition to “word-sized” schematic constructions, sequential constructions are also identifiable in the sentence in Figure 15. The two syntactic constructions listed in Table 6 are the Topic-Comment Construction and the Subject-Predicate Construction. The Topic-Comment Construction is characterized by a constellation of formal features, such as sequential order, timing, and non-manual facial expressions that accompany manual signing (Janzen 1999, 2017). Its function is to establish a discourse referent as known to the signer and the addressee (“About studying…”), and the signer then comments further on the established topic (“I thought I could put studying off”). Similarly, the ASL Subject-Predicate Construction contains open slots that specify the order of a Subject and Predicate. In Figure 15, the identified Topic phrase contains both a Subject and a Predicate ([1SG study]). For other Predicates, such as [got.an.F], the Subject is inferred from the preceding discourse context, rather than coded with separate manual sign. Omission of Subjects is quite common in ASL (Wulf et al. 2002); here I use this example only to illustrate how this type of construction can be analyzed for its complexity and schematicity: it is a phrasal template containing a sequence of two open slots. With these constructions identified, Table 7 provides an ASL counterpart to Table 5 above, listing a few ASL constructions according to their degree of symbolic complexity and schematicity.

A constructionist approach seeks to describe the varying structure of meaning-form units in ASL as a single type of knowledge: constructions as conventional symbols that vary continuously in complexity and fixedness. The example sentence analyzed in Figure 15, accordingly, can be analyzed as containing constructions that range in size from simpler, fixed forms, such as the single sign [keep.quiet], to more complex and schematic forms, such as the Topic-Comment Construction.

A constructionist approach also assumes that the majority of constructions actually exist somewhere between these two ends of a continuum, exhibiting degrees of symbolic complexity and phonological fixedness. Under this conceptualization of ASL structure, the variety of ASL structures that have been observed as straddling the “borderline” between words and phrases can be understood as belonging to the category of “partially-fixed words”. In addition, non-sequential morphology, which is commonly found in ASL, can also be explained through reference to a partially-fixed word as a type of construction.

4.2 Morphological complexity and “partially-fixed words”

The discussion in Section 3 emphasizes the phrasal complexity of conventional units in ASL. However, to know ASL is to know a wide variety of constructions that fall somewhere between single signs and complex phrasal templates. Individual ASL signs exhibit a great deal of analyzable structure, and it is well known in ASL linguistics that a single sign form may function as a complex, stand-alone predicate. This is likely because the visual modality offers more opportunities for simultaneous coding of information than the auditory modality (e.g., Emmorey and Corina 1990; Dudis 2004; Davidson and Gagne 2022), and perhaps because manual signs are slower to articulate than speech sounds (e.g., Bellugi and Fischer 1972; Mathur and Rathmann 2013; Börstell et al. 2016). In line with these explanations, this section explores how ASL signs code complex information, with examples from the “ASL slang” dataset.

Following the approach laid out in Section 4.1, sign-internal complexity is viewed here in terms of partially-fixed constructions abstracted over conventional signs. As in the preceding discussion, these aspects of sign complexity are written in the following way: sequentially analyzable elements in a single sign are separated by hyphens, and non-sequential patterns such as changes in location are indicated with angle brackets. Schematic sign constructions are represented by square brackets: only the aspects of a construction’s meaning and form that are fixed are listed, with the meaning and the form separated by a forward slash (/), when both are described.

The first example of a non-sequential pattern is the ASL sign [arrogant] (identified in Table 2 as “conceited”). The sign [arrogant] resembles the sign [large], which is signed with both hands moving out and away from each other in front of the signer’s body (Figure 17). The sign [arrogant] is signed with the hands configured similarly, and moving in a similar trajectory, however the hands start closer to the signer’s head, and the movement of the hands is more tense. Moving the hands to articulate near the head codes a change in meaning (“big-headed”), and this change is accomplished paradigmatically rather than sequentially (Frishberg and Gough 2000; Meir et al. 2013). Like a prefab that can be analyzed into its component words, the sign [arrogant] can also be seen as exhibiting complex morphological structure: [big<signed at forehead>].

Figure 17:

The ASL sign [arrogant] (left; image from https://youtu.be/UwobS6Gn0ss?t=66) is formed similarly to the sign [large] (right; image from https://aslsignbank.haskins.yale.edu/dictionary/gloss/772/), however it starts much closer to the signer’s head.

Similarly, the sign [feebleminded] exhibits this same non-sequential complexity. The sign [feebleminded] resembles the sign [weak], which is signed with the fingers of the signer’s dominant hand buckling against the flat non-dominant hand (Figure 18). The sign [feebleminded] is signed with the dominant hand articulating the same buckling movement, however the hand contacts the signer’s forehead instead. Moving the dominant hand to articulate at the head again codes a change in meaning through paradigmatic contrast in location (“weak-minded”).

Figure 18:

The ASL sign [feebleminded] (left; image from https://youtu.be/UwobS6Gn0ss?t=104) is formed similarly to the sign [weak] (right; image from https://aslsignbank.haskins.yale.edu/dictionary/gloss/2485/), however it is signed at the signer’s temple.

As Wilcox (2000, 2005) explains, in these and many other conventional ASL signs, such as [brilliant], [open.minded], and [hearing.minded], the relationship between meaning and form instantiates the metaphor of “the mind as a container”. The physical location of the forehead in these signs metaphorically represents mental states, thinking, and cognition. For present purposes, the elements of meaning and form that are shared among these signs can be analyzed as an emergent, partially-fixed construction: in many conventional signs referring to ‘cognition or mental activities’, the sign is articulated at the forehead. Moreover, ASL signers know that changing the location of an existing sign to be articulated at the forehead has the potential to create a new sign that metaphorically refers to mental states and processes. The schematic MENTAL Construction, [‘mental’/forehead.location], describes these shared elements among conventional ASL signs. This construction describes a sign-sized schema that has its location specified, and all other aspects of its form are unspecified.

In addition to changing the location of a sign, non-sequential complexity can also be found in the movement patterns among groups of related signs. Movement patterns have been found to distinguish aspectual values in ASL (Klima and Bellugi 1979), as well as coding referential uses of action concept signs (Supalla and Newport 1978; Padden and Perlmutter 1987). Many conventional signs are distinguished from one another by differences in the overall movement, such as movements in opposing directions or with different intensities (Frishberg and Gough 2000; Lepic and Padden 2017; Shaw and Delaporte 2010). Two signs that illustrate this type of movement contrast in the “ASL slang” dataset are [outrageous] and [worthless].

The sign [outrageous] is formed with the signer’s dominant hand moving in front of their face in several long, tensely articulated arcs. The manual component of this sign is similar to two other conventional ASL signs (Figure 19): the sign [silly] is formed with the forearm held stationary in front of the body while oscillating the hand in front of the signer’s face, and the sign [ridiculous] is formed with the dominant hand executing a single movement in front of the face, while simultaneously twisting the forearm. The concepts that these signs denote all share a sense of ‘absurdity’, and are all signed with the same Y handshape (, with thumb and little finger extended) moving in front of the signer’s face. From these signs, a partially-fixed sign can be abstracted: [‘absurd’/Y.handshape, near face]. I am unaware of any recent neologisms that instantiate this schematic ABSURD Construction, though presumably the first time the slang sign [outrageous] was coined, its form was motivated by the forms of the existing ASL signs [silly] and [ridiculous], as an instantiation of this same schematic construction.

Figure 19:

In the ASL sign [outrageous] the hand executes tense repetitions of a path movement (left; image from https://youtu.be/MynzxdkfxfE?t=110); in [silly] the hand executes a tense oscillation in place (center); and in [ridiculous] the hand executes a single path movement (right; images from ASL Signbank 2025; https://aslsignbank.haskins.yale.edu/).

A similar movement-based contrast can also be seen in the sign [worthless]. This sign is formed with both hands configured in the ring handshape (, with thumb and index finger contacting to form a ring) starting in contact and then moving to separate from one another. In ASL, other concepts relating to ‘value’ are also signed with two ring handshapes but with different movements (Figure 20): the sign [important] is formed with the two hands moving to contact each other in front of the signer, and the sign [judge] is signed with the two hands alternatingly moving up and down without contacting each other.

Figure 20:

In the ASL sign [worthless], the hands move apart (left; image from https://youtu.be/UwobS6Gn0ss?t=154); in [important], the hands move together (center); and in [judge] the hands alternatingly move up and down (right; images from https://aslsignbank.haskins.yale.edu/).

The shared meaning and form among these and other conventional signs again potentiates an emergent construction: [‘value’/ring.handshapes]. This schematic VALUE Construction captures an ASL signer’s knowledge that this hand configuration is fixed across signs, while other aspects of form vary. In the case of [worthless], specifically, the movement pattern also matches another family of signs. Just as the pair of signs [important] and [worthless] are antonyms that are signed with the hands configured similarly, but with opposing movements, so too are pairs of signs like [want] and [dont.want] and [like] and [dont.like], as in Figure 21.

Figure 21:

In the ASL sign [like], the hand moves away from the signer’s chest in a straight line (left); the sign [dont.like] starts similarly, but the hand twists away from the signer while opening (right; images from ASL Signbank 2025; https://aslsignbank.haskins.yale.edu/).

In these pairs of signs, the configuration of the hands is shared, but the ‘negative’ signs are formed with a twisting movement that ends with the palms facing away from the signer. Woodward and Desantis (1977: 385) explain that this construction has been grammaticalized from a separate negative manual form. However, in present-day ASL, this pattern is only found in a few signs, and it has been greatly reduced to a single twisting movement away from the signer, rather than a sequence of distinct signs. Accordingly, this NEGATION Construction, to the extent that it is a part of an ASL signer’s knowledge, can be represented as [‘negation’/away.movement].

The final construction to be analyzed here is a partially-fixed word that overlaps in meaning with the MENTAL Construction and, like the NEGATION Construction, involves phonological reduction due to grammaticalization. Recall from Figure 6 that the ASL construction [think-appear] ‘came to mind’ is formed with the index finger starting at the signer’s forehead, and then moving to articulate on the non-dominant hand. In ASL, numerous other signs also relate to cognitive concepts and are articulated with a brief initial contact at the forehead. In these signs, including [up.to.you], [agree], [purpose], and [goal] (Figure 22), the brief initial contact with the forehead signals the mental nature of the referent concept. Following a suggestion from a reviewer, I represent this construction as [THINK-X], with the uppercase gloss suggesting that this element is somewhat schematized relative to the independent verb [think]. The distribution of this initial contacting movement across these many signs suggests that it functions more like a grammaticalized “affix” that signals a cognitive referent.

Figure 22:

Many ASL signs referring to cognition and intention, including [THINK-mean] ‘purpose’ (left) and [THINK-point] ‘goal’ (right), are formed with the index finger initially contacting the signer’s forehead (images from ASL Signbank 2025; https://aslsignbank.haskins.yale.edu/).

The construction [THINK-X] is a partially-fixed word that is instantiated by fixed constructions like [THINK-appear] and [THINK-mean]. This template specifies the index finger’s initial contact at the signer’s forehead, but the rest of the sign’s form is unspecified. Several constructions from the four “ASL slang” videos are instantiations of this schematic construction, such as [THINK-disappear], which means ‘slipped my mind’ (Figure 23). Table 8 also lists other “ASL slang” examples that instantiate the [THINK-X] construction.

Figure 23:

The ASL sign meaning ‘slipped my mind’ begins with the signer’s index finger contacting their forehead at the temple, before articulating a second movement under the non-dominant hand (images from https://youtu.be/UwobS6Gn0ss?t=183).

As indicated by the parentheses in Table 8, some signs within this family also exhibit phonological variation, and need not be articulated with the initial contact at the forehead, at all. For example, the citation form provided in the “ASL slang” dataset for [not.interested] (identified as “garage door” in Tables 2 and 8) is articulated only with the dominant hand contacting the non-dominant hand. However, in the example sentence provided with this expression, the signer’s index finger very briefly contacts their temple at the onset of the sign (Figure 24).

Figure 24:

The ASL sign meaning ‘not interested in that topic’ can be signed with or without the index finger initially contacting the temple (images from https://youtu.be/9tkLIN6WEFA?t=24).

Similarly, the linguistic resource ASL Signbank (Hochgesang et al. 2024; https://aslsignbank.haskins.yale.edu/) lists two phonological variants for each of the two signs identified as “lose it” in this dataset. ASL Signbank glosses these forms as TEMPER-BLOW-UPix, TEMPER-BLOW-UPnoix, LOSE-TEMPERix, and LOSE-TEMPERnoix. In the ASL Signbank glosses, the final “ix” identifies forms with the index finger briefly contacting the forehead, and “noix” identifies forms without this initial contact (Figure 25). This glossing convention captures the phonological variation that can be observed among these two signs.

Figure 25:

The ASL sign [(THINK-)“blow.top”] ‘lose your temper’ can be signed with or without the index finger initially contacting the temple (images from https://aslsignbank.haskins.yale.edu/dictionary/gloss/2801/).

I do not have additional data to address this variation in the [(THINK-)X] family of signs directly, so I end this discussion simply by speculating that the relative frequency of occurrence between signs for ‘mental’ signs within this family likely drives the phonological variation that can be observed among them. I expect that constructions that are used most often to refer to metaphorical mental activities (‘not interested’), rather than to literal objects in motion (‘close garage door’), will also exhibit more advanced and consistent reduction of the initial contact at the forehead. In these signs, the second element in the sign effectively takes on the ‘mental’ meaning of the whole.

To summarize the preceding discussion, here I have sought to describe morphological patterns in ASL as constructions, using the idea of a “partially-fixed word”. This label names a type of construction that is considered to be a single sign, that exhibits complex symbolic structure, and that only has certain aspects of its meaning and its form specified in the construction, while others are unspecified. In ASL, morphological contrasts are often coded through changes to the location or movement of an existing sign, as a type of paradigmatic contrast (Lepic and Occhino 2018). In addition, sequential morphological patterns can be observed. However, the sequential morphological patterns analyzed here also seem predisposed to undergo grammaticalization to reduce “back” to single signs, as observed in some members of the [(THINK-)X] family of signs. This is consistent with observations in the literature, and in Section 3 of this study, that few highly conventional units in ASL seem to exhibit fixed sequential symbolic structure.