Idioms: The Words We Twist and Just Can’t Let Alone
I. Introduction
As English language teachers well know, the current formal model of grammar cannot account for every common utterance in the English language. Currently, grammar is taught to language learners in a prescriptive manner. It begins with verb tenses, making mention of a few standard idiosyncrasies, like the varying semantic entailments of different verbs and the appropriate context in which to use them. There is a certain group of standard grammatical constructions that are taught to language learners, such as conditional sentences, comparative statements, and at the higher levels, even inversion. But it frequently happens in the classes of the higher levels, that a student will be able to produce a coherent utterance that is either (1) completely unexplainable by the grammatical construction being taught, or (2) seemingly adherent to a different set of rules which is not stated by the construction being taught. Oftentimes, these exceptions will come in the form of strangely acceptable sentence fragments. Consider examples (1.1) - (1.4) below.
(1.1) I am going to the store after school.
(1.2) *I am going to the home after school.
(1.3) *I am going to home after school.
(1.4) I am going home after school.
Second language learners often make this mistake. With almost any other noun phrase and the verb “go” in the present continuous tense, this general structure applies:
[Subject] + [be verb] + [go + ing] + [to] + [noun phrase, sometimes introduced by “the”].
Using this template, students understandably assume that (1.2) or (1.3) is the correct grammatical construction. But in reality, the phrase modern language users hear, speak, and write, is “I am going home after school.” The reason for this, is that “go home” is a type of idiomatic expression, which must be memorized on its own, the same way a new lexical item would be. These didactic observations are evidence of a deeply rooted inconsistency in the traditional conceptual model of language comprehension. Since the mid 1980’s, there has been a growing movement among linguists to revise the current understanding of grammar and create a new model to properly accommodate such exceptions.
This paper will describe both the seminal and more modern research on this topic, by reviewing the literature that has been written about one particularly troublesome and revealing idiom, let alone. After reviewing what was originally thought to be the only correct grammatical usage of let alone, and the updated take on why the “deviant” use of the idiom may actually be logical, this paper presents a small corpus study to investigate the historical usage of let alone, using the Corpus of Historical American English (COHA). Every (relevant) instance of let alone in COCA from the 1800’s will be analyzed, to investigate whether the contentions made about its modern use are also true of its historical usage.
2. Syntax Research on Idioms
2.1 Earlier Approaches to Let Alone
These seemingly aberrant constructions like go home, and let alone are part of what Fillmore, Kay, and O’Connor call the “Appendix to the Grammar,” a sort of “repository” of idiomatic structures that must be acquired independently of the standard lexico-syntactic processes, meaning that they are memorized in the same way a new word would be, and mentally stored as a discrete unit (Fillmore et al. 1988: 504). In a paper that became known as one of the major seminal works in a movement for Construction Grammar, these authors examined the idiom let alone to demonstrate that within this “appendix” to the grammar, lies a veritable wealth of hidden semantic and syntactic patterns, whose influence would resonate throughout established grammar. Most comprehensively proposed by Adele Goldberg, Construction Grammar seeks to provide the basis for a new, more inclusive model of linguistic understanding, capable of accounting for idiomatic “outliers” in large part by asserting that phrasal constructions carry their own meaning, and are stored in the mental lexicon as such (Goldberg 2013: 504).
This theory, defended by Fillmore, Kay, O’Connor, and other researchers like George Lakoff, comes in sharp contrast to what is considered a traditional or “atomistic” view of grammar, which places emphasis on the “computing” part of understanding and producing language and keeps that which must be stored to a minimum. This schema is realized in an ordinal manner, beginning with the proposition that speakers have internal and contextual knowledge of the words in their language, to which they apply grammatical rules to create structures. Once these structures are realized, they are interpreted by means of compositional semantic principles, and finally, these interpretations are matched to the particular situations in which they arise, by way of the speakers’ pragmatic knowledge (Fillmore et al. 1988: 502). Fillmore, Kay, and O’Connor were some of the first researchers to systematically debunk this view, though they shied away from comprehensively proposing a new model and focused rather on defining various categories of idioms and delineating enough of the particular constraints on let alone that its analysis can stand in isolation (Fillmore et al. 1988: 518).
According to these researchers, there are several relevant distinctions to be made among different types of idioms. First, they follow the lead of Makkai (1972), in categorizing idioms as either “decoding” or “encoding” (Fillmore et al. 1988: 507). The difference mainly lies in how easily the meaning of the idiom could be guessed by someone hearing it for the first time. Decoding idioms are those with a meaning that would be difficult for language users to guess with full confidence unless they had learned them before. Encoding idioms by contrast, are those that may or may not be understandable upon first hearing but are nonetheless easily recognized as atypical relative to standard grammar. Based on these definitions, every decoding idiom is also an encoding idiom, but some encoding idioms are not decoding. For example, kick the bucket and pull a fast one are both encoding and decoding, since their meanings would be more difficult to guess on first hearing, but it is theoretically possible. Answer the door and wide awake are easier to guess, but conceptually and structurally not quite aligned with the standard grammar, so they are encoding only (Fillmore et al. 1988: 507).
Next, the researchers define “grammatical” and “extra-grammatical” idioms. Despite being exceptional phrases which are stored separately in the mental lexicon, some idiomatic expressions do adhere to familiar grammatical rules; examples of this type include kick the bucket, spill the beans, and blow one’s nose (Fillmore et al. 1988:. 505). Extra-grammatical idioms by contrast, are those which the standard grammar cannot account for, like all of a sudden, so far so good, and our first example, go home.
Finally, the researchers divide the idioms into “substantive” and “formal” idioms. The former title refers to idioms that are essentially set expressions, whose “lexical make-up is (more or less) fully specified.” The example they give is, The bigger they come, the harder they fall. “Formal” idioms by contrast, are considered lexically “open,” and derived primarily from syntactic patterns. The given example for this category is The X-er, the Y-er--seen in lexically mutable grammatical constructions such as: The more you study, the better your grades will be, or The longer you wait, the harder it’s going to be. The researchers then use these distinctions to investigate some of the more novel instances of idiomatic expressions, categorized in terms of whether their pieces (words) are familiar (common to the modern lexicon), and whether their order or grammatical structure is familiar in the standard grammar.
Having distinguished these categories, the researchers are able to examine various cases of the idiom let alone. The researchers saw this idiom as an attractive entry-point to the conversation around Construction Grammar, having the impression that let alone sentences represent a collection of interesting syntactic and semantic properties distinct from standard grammar, but shared by a family of other phrasal constructions (Fillmore et al. 1988: 511). It is both an encoding and decoding idiom: an aspect of its meaning may be guessed upon first hearing from the lexical items that comprise it, but due to the extra-grammatical syntactic patterns it entails, its complexity cannot be gleaned with full confidence until it is learned.
To begin with, the researchers draw a few important conclusions about the syntax of let alone: firstly, that it functions like a coordinating conjunction joining two clauses as does the word and (see example (2.1) below). More specifically, the researchers consider let alone a negative polarity item, as it occurs mainly in negative contexts, with an element in the first clause acting as a “polarity trigger” (Fillmore et al. 1988: 512). In example (2.1) below, the word barely acts as the affective trigger, setting up the sentence in a negative frame, such that only lexical denoting a negative experience will be able to fill it out.
(2.1) I barely got up in time to EAT LUNCH, let alone COOK BREAKFAST.
(2.2) You’ve got enough material there for a whole SEMESTER, let alone a WEEK.
Within these two conjuncts are found one or more sets of paired, contrasting foci (p. 512). In (2.1) above, the focal points are capitalized-- “eat” pairs with ”cook,” while “lunch” pairs with “breakfast.” It is noted that the second conjunct, occurring after let alone, is a sentence fragment which can only be understood by adding or “restoring” constituents from the first part (p. 514). Essentially, “cook breakfast” is not a complete sentence, but its meaning can be interpreted by redistributing the first constituents “I barely got up in time to...”
The researchers do provide examples that challenge that theory (see example (2.2) above), as they came up in their research and lead them to believe that if let alone is a negative polarity item, it is “not simply and straightforwardly one” (p. 519). The contention has since been challenged by other researchers, most recently by Cappelle, Dugas, and Tobin (2015), whose objections will be discussed in a later section of this paper.
The semantic analysis of these conjuncts and its implications is the source by which the present corpus study is inspired. While at first glance, it may seem that the relationship between the two parts of a let alone construction is one of entailment, this is not the case (p. 523). In example (2.3) from the Corpus of Contemporary American English (COCA) below, a surface-level reading might suggest that the meaning is, “he didn’t look old enough to be out of high school, therefore, he didn’t look old enough to be out of law school.” But a close reading reveals that this construction actually implies a “presupposed” semantic scale, on which the paired foci are fixed points (p. 523). The semantic scale implied here, to use similar terms to the ones found in Fillmore, Kay, and O’Connor 1988, would be one in which the speaker assumes a “dimension” of youthfulness, in that the prototypical attendees of high school are (or at least look) younger than the prototypical attendees of law school (p. 525).
(2.3) He hardly looked old enough to be out of HIGH SCHOOL, let alone LAW SCHOOL. (COCA, 2017, FIC, Bk:ArsenicWithAusten)
Other examples, such as (2.4), are less straightforward, as they involve multiple paired sets of foci. In this example, which has been invented for our purposes here, there are at least two dimensions at play, both seeming to allude to the speaker’s lack of trust in John. The two scales in operation are those of “distance” and “value.” The preposterousness of the notion of the speaker trusting John to take the more valuable item (the car), the farther distance (Las Vegas) is highlighted by the parallel example of his not trusting John to take a much less expensive item (the bike) a much shorter distance (down the street).
(2.4) I wouldn’t let JOHN take my BIKE down the STREET, let alone my CAR to LAS VEGAS.
A point Fillmore, Kay, and O’Connor underscore is that the listener is not able to interpret the exact nature of the scale being dealt with from the let alone sentence itself, rather, they must use the preceding pragmatic context as a guide (p. 525). In fact, as later researchers have pointed out, the semantic scale itself must be predicated upon what Grice originally termed “common knowledge”: information assumed to be known by both the speaker and the listener (Cappelle et. al, 2015, p. 72 ). In example (2.3) the speaker assumes that the listener is aware of a scalar relationship in terms of youthfulness and youthful appearance, and that high school naturally precedes law school. As Cappelle et. al mention, the scalar implicature of two focal points can follow from the lexical items, and that is the case here with “high school” and “law school” (p. 72). However, there are cases in which the implied scale is specific to pragmatic context, in which the order is not necessarily logical entailment, but contextual entailment (Cappelle et. al p. 72). (2.4) for example, could instead deal with concern for John’s wellbeing, such as natural disasters that would prevent him from safely traveling any distance, long or short.
The true predictability in the semantic interpretation of let alone constructions lies in the notion of “Relative Strength” (p. 527). The two clauses joined together by let alone are interpreted as two “propositions” from the same scalar model (p. 528). These two propositions are assumed to be of the same polarity (in most cases, negative), and of the two, the syntactically first proposition should be the stronger one (p. 528). The researchers have termed it as “stronger” because within the assumed context, it is the more surprising proposition, and it is the one being emphasized to underscore the unlikeliness of the second proposition. In example (2.3) above, we may assume the context is that the person being described is either in law school, or another speaker in the dialogue made some mention of his being or appearing to be in law school. Given this frame of reference, it is a stronger proposition to assert that he does not look old enough to be in high school, as it illustrates the scalar distance between the reality of his appearance and the appearance of a prototypical law student. The reason for this order can be understood in pragmatic terms as well: let alone constructions are said to uphold both the Gricean maxim of informativeness, and that of relevance, by presenting the two propositions in a reliable order: the first being more informative (ie “he does not even look old enough to be in high school”), and the second being more relevant (“he does not look old enough to be in law school”) (p. 513).
2.2 Contemporary Approaches to Let Alone
As previously mentioned, there have recently been dissenting researchers who object to Fillmore, Kay, and O’Connell’s conclusion that let alone is primarily a negative polarity item that exists within the semantic and pragmatic constraints listed above, and that divergent examples are mostly caused by the speaker’s misunderstanding or misuse of the construction. For instance, Cappelle et. al (2015) argue instead that of the seemingly deviant manifestations, there is at least one that they consider “fully legitimate, and fairly common,” constituting 4.5% of 200 corpus examples drawn from COCA and the British National Corpus (BNC) (p. 81). In this use, let alone introduces a sort of “afterthought”, in which the speaker introduces something she wants to lightly allude to, reinforcing the statement she just made without delving too much into it, as the strength of the first statement renders it unnecessary (p. 81). This type of utterance, they contend, amounts to “apophasis…[a] rhetorical trick of speaking of a subject by denying that it is your intention to bring it up” (p. 81). An example of this type of expression would be something like, Never mind WHAT THE CRITICS SAID, let’s talk about WHAT YOU THOUGHT of your performance. In this example, the speaker covertly mentions the critics opinions of the performance, by dismissing it as unnecessary to discuss.
Defining Fillmore, Kay, and O’Connell’s standard let alone construction as “canonical” and their deviant but viable alternative as “afterthought,” Cappelle et. al note that the main structural difference between the two constructions is that the order of the conjuncts is reversed, which begets both semantic and pragmatic ramifications. The researchers divide the afterthought cases into three categories: “illogical,” “underspecified,” and “illogical with more explicit pragmatics” (pg. 79-81). The “illogical” use is the semantic reverse of the canonical one: the weaker proposition is placed first, and the stronger proposition second, after the let alone clause (pg. 79). The connection between these will sometimes involve the a fortiori entailment of the canonical use, but often will not, the connection being more simply an additive one, like “and.” The clauses will also be the inverse of the canonical use in pragmatic terms, the first one being the more relevant, and the second more informative (pg. 79).
To illustrate this contrast, example (2.5) from the Corpus of Contemporary American English (COCA) below is of the canonical ilk, while example (2.6) from Cappelle et. al, is the afterthought use. Having antipathy toward immigration of any kind is a much more intense standpoint than having antipathy toward refugees in particular, thus example (2.5) follows the canonical pattern of stronger proposition first, weaker proposition second. Because refugees are a subset of the general category of “immigration,” the relationship between the two propositions can be seen as one of entailment. Similarly, because “immigration” sets up a categorical frame for the comparison, the first proposition is more informative--and because refugees are the specific topic at hand, the second proposition is more relevant.
Example (2.6) by contrast, is the afterthought use. The speaker is emphasizing his relative lack of daily practice time in comparison with the guitarist and communicating that a significantly lower number of hours is for him, a feat. Within the scalar dimension of “lack of practice,” practicing for three hours would be a stronger proposition than practicing for four, so the first proposition in this case is weaker, while the second is stronger. This inversion of the relative strengths causes the phrase to function differently. Rather than contextualizing the second proposition in scalar dimension relative to the first, the speaker presents the more relevant proposition first, and the more informative one in the fragment after let alone.
(2.5) And then on top of it, at this particular political moment, there is so much antipathy toward IMMIGRATION of ANY KIND, let alone REFUGEES in particular. (COCA, 2017, NEWS, Los Angeles Times)
(2.6) One of the guitarists in our band, to his full credit, averages five hours a day practice on the guitar. For me on the drums, I’m pleased if I make it to FOUR, let alone THREE.
The “underspecified use” is unique in that neither proposition is necessarily stronger on the semantic level, and the two conjuncts are semantically joined by an additive, (“and” equivalent) (p. 80). But the defining characteristics of this use are in its pragmatic constraints. To clarify this distinction, Cappelle et. al suggest that the canonical use of let alone may in fact be more complex than simply, “the first proposition is more informative, and the second is more relevant” (p. 80). The greater depth they suggest here is that the first proposition seems so sufficient on its own in the context, that it is almost unnecessary—or even irrelevant—to state the second (p. 80). For an example of this use, see example (1.7) from data Cappelle et. al took from Verghaven (1994:279) below (p. 80)
(1.7) I seldom spend 100 hours on ANYTHING these days, let alone A VIDEO GAME.
According to Cappelle et. al, the “underspecified” use opens up new pragmatic possibilities. Since it is seen in the underspecified construction that the speaker does not consider the second proposition as noteworthy as the first (perhaps due to lack of “sufficient contextual relevance”), the Gricean constraints of the canonical use may be reversed, such that the first proposition should be more relevant, and the second more informative (p. 81). The researchers termed this use “illogical with more explicit pragmatics,” and it holds a few important implications: firstly, that this order of propositions more closely follows the typical information flow, from old to new (p. 81). Additionally, the second conjecture, since it is seen as not worthy of much attention, is being stated to preemptively “cancel a conversational implicature triggered by [the first proposition]” (p. 81).
Cappelle et. al hold that while “synchronically, it is most plausible to consider the afterthought use as an extension of the canonical use, given that the latter is much more frequent,” it is reasonable “to assume that it is actually the afterthought use that originally provided the basis for an extension to what is now the canonical type,” diachronically speaking (p. 79). The researchers take as evidence the oldest examples of let alone from the OED, finding they are of the afterthought type (79).
Having reviewed the prior research, this paper will now present data from COHA to investigate whether historical uses of let alone trend more toward the canonical use or the afterthought type, and within the afterthought type, if they fall into the three categories the researchers defined.
3. The Present Study
To gather data for this corpus study, the phrase let alone was searched in an online database with historical records of American English (COHA). Each decade of the 1800’s was reviewed, and each individual case of let alone was analyzed to determine whether it exemplified the correct sense to be used in this study. (Let alone can also equate to “leave alone,” or “despite.” In those senses it is like a different lexical item; it does not occur between two propositions or exhibit the syntactic patterns in question). There were 385 instances of let alone between 1820 and 1900, and of those, 121 cases were the correct sense (Davies 2010).
Of those 121 cases, 117 were the canonical use, with only four clear-cut cases of the afterthought use. Statistically, this number equates to 3.31% of utterances. While it might appear small, this result is not far off from the results of Cappelle et. al’s corpus study of 200 random cases from the British National Corpus (BNC) and COCA, in which 4.5% of utterances of let alone were the afterthought use. Of the examples found in the present study, one was the “illogical” use, two were the “underspecified use,” and one was the “illogical, with more specified pragmatics.”
First let us examine example (3.1), from Moby Dick, published in 1850. Since the orlop (the lowest deck on a ship) is where the character in question (Fedallah) lives, it seems less shocking for him to appear there than in the admiral’s cabin (which we can lexically infer is of higher prestige), making that a stronger proposition. In the canonical use of let alone, the phrase about being in the “orlop...where he lives” would be emphasized in the first clause, and the “admiral’s cabin” clause would be in the fragment after let alone.
But there is more to the story here, as the fragment after let alone includes not only “the orlop...where he lives” but also “the upper decks where he sneaks so much.” This is a semantically complicated pairing of foci, since they expect him to be in the orlop, as he belongs there, and they also expect him to be on the upper decks, though it is implied that he doesn’t belong there. The “upper decks” location seems to be more similar to the one in the first proposition, “the admiral’s cabin,” and is perhaps mentioned to cancel out any conversational implicature that he might be expected or belong in that first location. The pragmatics of this reading makes this example most closely aligned with the third type of “afterthought” use defined by Cappelle et. al, “illogical, with more specific pragmatics.”
(3.1) "Suppose he should take it into his head to duck you, though -- yes, and drown you -- what then? " " I should like to see him try it; I'd give him such a pair of black eyes that he wouldn't dare to show his face in the ADMIRAL’S CABIN again for a long while, let alone down in the ORLOP there, where he lives, and hereabouts on the UPPER DECKS where he sneaks so much. (COHA, 1851, FIC, MobyDickWhale)
The context of example (3.2) is that one character has just informed the other of some local news—namely, a murder that took place during a wedding the previous night. The relative strengths of the propositions here may be read in two ways. The deictic that here references the murder, which is inherently a more surprising or unexpected event to occur at a wedding than a bride falling into a fit. In this reading, it would seem that the bride falling into a fit is the weaker proposition, and since it is placed in the fragment after let alone, the phrase is rendered a classically “illogical” afterthought use. However, the “bride falling into a fit” is more relevant to the wedding itself, so in pragmatic terms, the pattern could be seen as the canonical usage, with the more relevant information in the fragment clause.
(3.2) No wedding could go on after THAT, you know, ma'am, anyhow, let alone THE BRIDE FALLING INTO A FIT the minute she saw the bloody corpse of her murdered father, and being of a raving manyyack ever since. (COHA, 1894, FIC, LostLadyLone)
In example (3.3), the first proposition is the more expected, unsurprising one--the defining characteristic of an orphan. As such, it is the weaker proposition, and also the more relevant, which in the canonical use would be relegated to the fragment clause. Because the first proposition is by far the most important feature of being an orphan, the second proposition is hardly worthy of mentioning. This pragmatic configuration makes this construction a clear case of the “underspecified” afterthought use mentioned by Cappelle et. al.
(3.3) That child, sir, is an orphan; got NO FATHER NOR MOTHER, let alone GRANDF’THER or GRANDM’THER, in the land of the living. (COHA, 1895, FIC, Nautilus)
Example (3.4) is also a case of the “underspecified” afterthought use, in that it gives the weaker, more relevant proposition first, before the fragment. The author is talking about his ability to write, as a sort of letter to the reader before he begins recounting a personal anecdote. This essay was submitted to a journal called Lippincott’s Monthly Magazine in 1896, which is a bit ironic given the context of the present paper and may lend credence to the theory that the afterthought use of let alone is sometimes born of misuse or misunderstanding. As it is written, the proposition of the second clause “other studies” is hardly relevant or worthy of information compared to the first proposition of “Rhetoric and English Composition,” making this a clear case of the “underspecified” afterthought use.
(3.4) Any one who hopes to find in what is here written a work of literature had better lay it aside unread. At Yale I should have got the sack in RHETORIC and ENGLISH COMPOSITION, let alone OTHER STUDIES, had it not been for the fact that I played half-back on the team, and so the professors marked me away up above where I ought to have ranked. (COHA, 1896, FIC, GreatKAmpARobbery)
In both studies, a small but not insignificant percentage of cases of let alone were of the “afterthought” type, and there was a fairly even distribution of its sub-variations within them. Though a larger corpus study would be more revealing, the results of this study seem to indicate that the let alone construction has long had both prototypical and fringe iterations that are accepted widely enough to be published. Little mention has been made here of the potential deep metaphor in let alone, but in those cases in which the scalar implicature seems to be absent from the propositions (the “underspecified” use), it could be that of the equally weighted conjuncts, one is more important, and the other, found in the fragment clause, can be cast off to the side, and in a more literal sense, “left untouched” (Cappelle et. al 2015: 79).
The presence of the afterthought uses of let alone in the historical data lends credence to Cappelle et. al’s theory that it is more than a simple misunderstanding or misuse of the canonical structure. Because two of these historical examples were the “underspecified” use, it supports Cappelle et. al’s theory that the canonical and afterthought uses are “cognitively linked via a more general constructional pattern with partially underspecified semantic and pragmatic information” (Cappelle et al. 2015: 81). This conjecture is further supported by the presence of “transitional cases” in which neither proposition is objectively more relevant, but rather both are relevant to different aspects of the discourse (example 3.1, 3.2). If a more general construction pattern than even the ones described by Fillmore et. al and Cappelle et. al can be stored in the mental lexicon, that is some significant supporting evidence that it is meaning-carrying constructions, not just individual lexemes, that are stored in the mental lexicon. The recurrence of these syntactic, semantic, and pragmatic patterns throughout history indicates that forms like let alone should not be relegated to the “Appendix” or “repository” outside of quotidian grammar, because they are not anomalies at all. They are standard constructions that interact with a particular family of expressions and phrases.
V. Conclusion
This paper has given a brief history of the study of idioms, and more specifically what has been written on the let alone construction. The review has covered the seminal work on this topic, in which Fillmore, Kay, and O’Connor argue that the only correct use of let alone is one in which one first offers two propositions with a scalar dynamic between them: by first offering a strong proposition, and then using its strength to reject the second proposition as less likely. The review has also covered the updated proposals about this construction--using Cappelle et. al’s 2015 corpus study as a guide--namely that the seemingly deviant use of let alone, in which its propositions are reversed, appearing with the weaker proposition first, and the stronger second, is a valid use of the construction, forged consciously and upon a different type of logical predication. Furthermore, it was found by a small corpus study, that historical utterances of let alone from the 1800’s patterned similarly to those found in the corpus study of modern language by Cappelle et. al. These results align nicely with the contentions of Construction Grammar, evincing the overarching claim that phrasal constructions themselves have meaning, and perhaps always have--meaning the expressions which are considered irregularities within the framework of traditional grammar, are actually evidence that the framework itself is fundamentally flawed.
References
Cappelle, B., Dugas, E., & Tobin, V. (2015). An afterthought on let alone. Journal of Pragmatics, 80, 70–85. doi: 10.1016/j.pragma.2015.02.005
Davies, Mark. (2010-) The Corpus of Historical American English (COHA): 400 million words, 1810-2009. Available online at https://www.english-corpora.org/coha/.
Fillmore, C., Kay, P., & O'Connor, M. (1988). Regularity and Idiomaticity in Grammatical Constructions: The Case of Let Alone. Language, 64(3), 501-538. doi:10.2307/414531
Francis, E. J. (n.d.). Linguistic Institute. Linguistic Institute. Lexington. Retrieved from https://www.linguisticsociety.org/e-learning/ppt-construction-grammar-idioms
Goldberg, A. (2013). Constructionist Approaches. In The Oxford Handbook of Construction Grammar. doi: 10.1093/oxfordhb/9780195396683.013.0002
Hirschberg, J. B. (1985). A Theory Of Scalar Implicature (natural Languages, Pragmatics, Inference) (Order No. 8603648). Available from ProQuest Dissertations & Theses Global: The Sciences and Engineering Collection. (303383309).
Lakoff, George. “There-Constructions.” Women, Fire, and Dangerous Things: What Categories
Reveal about the Mind, University of Chicago Press, 1987, pp. 462–585.
Nunberg, G., Sag, I., & Wasow, T. (1994). Idioms. Language, 70(3), 491-538.
doi:10.2307/416483
Roberts, J. L., & Melville, H. (1987). Moby Dick. Singapore: J. Wiley (SEA).
