The case of question-based exclamatives: From pragmatic rhetorical function to semantic meaning

1 Introduction

Questions are used for multiple rhetorical purposes, beyond just seeking for information-bearing answers (e.g. Sadock 1971, 1974; Han 2002; Koshik 2005; Bardenstein 2018, 2020a; Bardenstein in press). Questions are used for polite requests (e.g. could you please pass the salt?, do you have the time? Hello, is mom home?, could you please repeat that?), questions are used for polite propositions (e.g. shall we dance? Why don’t you go and rest a little?), questions are used to draw the addressee’s attention and/or prepare the addressee for upcoming news, as discourse markers (e.g. wanna hear something amazing?/(are you) ready for this?).^[1] Questions are also used for discursive bias of prior knowledge/experiences and/or speaker’s expectations (e.g. didn’t we watch that movie before?, aren’t you happy with the news? respectively), as well as to confirm or authenticate (isn’t that your sister in law in the coffee shop?), express affection or to complement (who’s mommy’s little angel?, who’s the prettiest girl in the world?), to express surprise/doubt (e.g. really?!) to bring about allusions and salient quote-based questions (am I my brother’s keeper?, shall I compare thee to a summer’s day?)^[2] and lastly and most importaantly, questions are discursively used to express the speaker’s strong stance (usually that of resentment or dicontent): are you crazy?! What do you think you’re doing?(!), who do you think you are?(!) etc.

All in all, linguistic items in general and questions in particular are rhetorically-discursively recruited for five main reasons/purposes and this is why change occurs (also see Bardenstein in press):

1. Following social and cultural norms such as being polite (saving face, and linguistic hedging.

2. Stance-expressing such as exclamation and the use of negatively-biased rhetorical questions.

3. Argumentation such as the use of minimizers to rectify or strengthen a negative claim due to loose use.

4. Keeping discursive coherency and proper flow of the discourse such as the use of attention grabbing discourse markers or suspending the progressivity of the discourse for various reasons such as to varify and clarify.

5. Expressing speaker’s style and language innovation and versatility as in the case of the use of allusions, humor, sophisticated use of language etc.

However, despite these different recruited discursive-rhetorical purposes of questions, they all linguistically evolve in a srtikingly similar way. They do so in the sense that the initial core rhetorically-recruited pragmatic function motivates the change, ‘takes over’ the meaning of the question construction and persists throughout its history, even when turning into an adverbial. Thus, what used to be a question might get to a point where it doesn’t function or even be perceived as a question any more. For example, English exclamative How- X! (e.g. How lovely!) and Hebrew agreement intensifier exclamative Veo’d eix! (literally: ‘and more what’, idiomatically: ‘of course’/‘you bet!’) started out as question constructions but are not used and are no longer even perceived as a question by native speakers.^[3] Despite all of that, in both cases, it is their initially recruited rhetorical function of expressing the speaker’s strong stance that persists throughout their history of linguistics change.

In previous work (Bardenstein 2018; in press), I have divided ‘rhetorical questions’ into three types:

1. Ad-hoc Rhetorical Questions

2. Questions on the rhetorical spectrum

3. Constructionalized Rhetorical Questions (henceforth CRC).

The first type of rhetorical questions deals with ad-hoc contexts, such as asking a married woman who comes to work wearing a velvet white dress if she was getting married: “are you getting married today?”, implying that she overdressed. This type of questions is always interpreted via a Gricean pragmatic implicature (Grice 1975). The second type of questions is available due to its salience to speakers. These questions contain specific lexical material, which can be interpreted either as answer-seeking questions or as rhetorical ones (not seeking an answer) or anywhere in between. e.g. the English surprise-expressing really?! which is sometimes answered (usually with a “yes” that answers the speaker’s expectations) and sometimes not answered at all but rather simply expresses the speaker’s surprise. The third type is a case of idiomatic constructions (in these cases, the idiomatic meaning is the default interpretation (Giora 1997)). These question constructions can be treated as ‘Goldbergian Constructions’ (see Goldberg 1995), whose meaning is not composed of the sum of their parts (first and foremost because this type of questions is not used for interrogative purposes anymore and they are not followed by an answer). For example, the English who are you to tell me what to do?! Is never used as an interrogative any more.^[4] Rather, it is used to scold the addressee for telling the speaker what to do since s/he don’t have the right to do so.

In general, the above division represents a scale reflecting an ascending rate of association between the literal and the (originally) inferred. In (a) it’s possible, by relying on the ad hoc context, in (b) it’s available due to its salience to speakers and in (c) it’s already a default interpretation, sometimes even the only and obligatory interpretation. This type might also be named the ‘non-question’ type, since this type of rhetorical questions doesn’t functions as an answer-seeking question (it is not provided with an answer from addressees) and in many cases, it is not even perceived as a question at all. In writing, it is not necessarily followed by a question mark and prosodically it might not include the question contour (see, for example Hirschberg and Ward 1995).

In this paper I focus on the two latter cases of rhetorical questions. First, I show that a type B rhetorical question might turn with time into a type C rhetorical question since like any other linguistic phrase that is consistently associated with a certain discourse function, questions might gradually lose their original function (in this case: questioning) and turn into linguistic constructions a-la Goldberg (1995). In addition, I argue, this process of constructionalization is conducted via a core discursive function that persists throughout the question constructions’ history- a process that makes them “polysemous” in a cyclic manner. More specifically, in this paper I investigate the constrcutionalization (a process which is followed by semnaticization/grammaticalization) of questions into differrent types of exclamatives via the initially implicated strong speaker’s stance.

2 Theoretical background

3 Grammaticalization and the persistent cyclicity of question-form polysemies

Some questions indeed undergo a process of grammaticalization (see, for example, Herring 1991 for the grammaticalization of rhetorical questions in Tamil). However, questions which turn into CRQ demonstrate processes of constructionalization in which the mechanism is pragmatically motivated and driven and it is that constrcutionalization process that might eventually lead to grammatical changes as well.^[7] In other words, grammaticalization is the outcome and not the process itself. In addition, this linguistic change cannot come to be in clear and well defined ‘clear cut’ stages of grammatical changes per se, since this linguistic change occurs on a wide spectrum and operates up to a point where the relevant questions: (a) are never interpreted as ‘real’ (information-seeking) questions and (b) there is no way of predicting their meanings based on the meaning of their parts or based on other related constructions. This process is gradual- it does not “leap” from one stage to the other and might demonstrate more than one clear cut ‘Bridging context’ (Evans and Wilkins 2000).^[8] i.e. with usage and depending on the context, the question form becomes less and less “clear” (in the questioning sense) and more and more vague. i.e. it becomes less and less clear if the question requires an answer or not (if the speaker actually expects an answer). This process creates question-form polysemies in the sense that once recruited for rhetorical purposes, the question construction adds meanings in a cyclic manner^[9] because once constructionalized, the CRQ might evolutionize and constructionalize yet again and again.

In recent work, linguists have researched the essense of cyclic developments at the semantic/pragmatic level of change (see Hansen 2014, 2015; Ghezzi and Molinelli 2014). Hansen (2014, 2018 has focused on “semantic/pragmatic cycles” or “cycles of pragmaticalization” (also see Ghezzi and Molinelli 2014), and found that the evolution of a particular type of content-level meaning or function to one or more particular types of context-level meanings or functions takes place more than once within a given language and/or across two languages and that one is the direct descendant of the other.

This paper deals with what I dubb “persistent cycles of constructionalization” – a process which revolves around the linguistic evolution of a particular function-based construction which takes place more than once, while maintaining its initial recruited pragmatic rhetorical function.

While the existence of semantic/pragmatic cycles is a relatively recent a discovery, even in the early twentieth century, Jespersen (1917) discussed cycles in which a verbal negative particle ‘ne’ and its strengthening minimzer ‘pas’ (the construction [ne.pas] unites and is perceived as one, to the point where the former ne is no longer explicitly needed in colloquial language and pas “takes on negation”. This was dubbed by Dahl (1978) as “The Jespersen’s cycle” and it is basically a cycle of constructionalization which is based on pragmatic inferencing (its initial stage is that of ‘ne + minimizer pas’ = absolute negation) even though usually it is referred to as a cycle of grammaticalization (‘noun’ < ‘negation operator’) or that of semanticization (‘a step’ < ‘not’).

4 Question-based exclamatives: “It’s all about the stance”

Exclamatives can be expressed by a wide variety of formal structures and constructions (inter alia Michaelis 2001), and they constitute expressive speech acts (e.g. Searle 1979; Castroviejo-Miro 2010; Chernilovskaya 2014; Nouwen and Chernilovskaya 2015). Question-based exclamatives necessarily convey the speaker’s strong stance (of amazement, appreciation etc.) and are constructions that resemble questions in containing question words (in the case of wh-exclamatives) or inverted word order (in the case of yes/no exclamatives), but differ from interrogative questions in meaning and prosody (Burstein 2005, Nouwen and Chernilovskaya 2015).

In English, there are two main wh-exclamatives: what as in (4), and how as in (5):^[10]

Whilst English mainly has what- and how-exclamatives, languages like Dutch, German Russian, Arabic and Hebrew allow for exclamatives based on other wh-words too (also see for example, Chernilovskaya and Nouwen 2012). Each language has its own set of wh-exclamatives that scope over either one constituent (such as a noun, an adjective or an adverbial) or a whole proposition (see Chernilovskaya and Nouwen 2012). In Hebrew, for example, many (though not all) wh-question words are used as exclamatives, scoping over one or more constituents for various discursive purposes.^[11] For example, the wh-question-based exclamatives can be used to adverbially intensify the constituent they scope over (either explicitly as in 6a and 6f in which the constituent yafa ‘beautiful’ and taim ‘tasty’ (respectively) is explicit) or implictly as in 6c in which the adjectival nature of the ‘day’ as good or bad is implicit). In addition, wh-question based exclamatives can also be used for discursive rejection (as in 6d). All in all, all of these wh-exclamatives convey the speaker’s strong stance (either negatively or positively, usually negatively) and this shall be the departure point of this paper:

All in all, this paper focuses on the historical process by which questions turn into exclamatives or excalmative constructions and later might turn into different kinds of adverbials whilst keeping their stance-expressing function throughout their history. Not many have dealt with this issue. Among others, Nouen and a Chernilovskaya (2015) claim that “given the occurrence of wh-expressions in exclamatives, one might wonder whether (wh-) exclamatives and questions share certain grammatical or semantic mechanisms” and that “the resemblance between these two types of clauses is arguably quite superficial” and conclude that “even if there exists some relation between questions and exclamatives, this relation is not going to be a straightforward one”. I claim that the connection is indeed not a strightforward one, but one of a gradual historic change based on pragmatic inference that turned semantic and is cyclic in nature. In Bardenstein (2020a) I have given several cases of CRQs in Hebrew and their historical change. More accurately put, I have provided evidence for four types of stance-expressing CRQs, based on their constructionalized discursive use/meaning:

1. CRQs of strong disagreement, e.g. ma pito’m! ’lit: what all of a sudden’, idiomatically: ’No way’!

2. CRQs of strong agreement, e.g. ve’od eix! ’lit: and more what’, idiomatically: ’of course’/’you bet!’

3. CRQs of adverbial upgrading, e.g. eize X! ’lit: which X’, idomatically: ’what (a) X!’/how X!

4. CRQs of adverbial dowtoning, e.g. lo mi yodea ma, lit: ’not who knows what’ idiomatically: ’not very (good/adjective)’.

While type a shifts from expressing strong negative stance inferencially to semantically conveying such stance, type b takes an additional step and develops a strong positive stance (following the previously negative one). Type d CRQs are extremely interesting since their negative bias (negative inference) gradually took over and became explicit and the construction became a NPI (Ladusaw 1996). Such is the case of lo mi yodea ma lit: ’not who knows what’ that functions as a downtoner (originally stemming from the negative biased question: mi yode’a ’who knows?’.^[12]

While types a, b and c are considered excalamtive constructions, type d isn’t, but historically it used to be an exclamative before fixating under the explicit negation therefore becoming an adverbial NPI downtoner, leaving its exclamative use behind. All four types of CRQ evolutionize as shown in Figure 1.

Figure 1:

The linguistic evolution of question constructions exclamatives.

Figure 1 demonstrates the different stages of linguistic change that question constructions undergo. In terms of the semantics/pragmatics decision, this change can be outlined as demonstrated in Figure 2.

Figure 2:

Four types of exclamative CRQ.

All of the above types a–d exclamative CRQ evolutionize via the speaker’s negative stance that is initially pragmatically deducted and then fixates and becomes semantic. Type b and c evolutionaize further and the its semantic negative stance becomes positive. Therefore, while type a rhetorical questions demonstrate a two-stage constrcutionalization process via speaker’s stance, type b and c demonstrate a three-stage constrcutionalization process via speaker’s stance and type d demonstrate a four-stage constrcutionalization process via speaker’s stance. In a nutshell, no matter how far the cyclic evolution of the CRQ goes, the speaker’s strong stance persists (it is initially pragmatic and then becomes semantic). This evolution shifts from one stage to the next via processes (one or more) of constrcutionalization.

In Bardenstein (2020a) I have elaboratively demonstrated the historical evolution of the four types of CRQ, from biblical Hebrew to contemporary Hebrew. Type a was demonstrated with the historical evolution of the exclamative ma pito’m (literally: ’what all of a sudden’, idiomaticaly: ’no way’!). Initially the question ma pito’m?(!) was used to convey the speaker’s strong resentment towards an accessible discursive assumption or claim and later, ma pit’om constructionalized as an exclamative whose meaning/function is the rejection of an accessible assumption or claim. This CRQ is often not even perceived as a question construction by speakers any more.^[13]

Type b was demonstrated with the historical evolution of the exclamative: ela ma (’what else’). Initially the alternative seeking question ela ma? was used to pragamtically convey the speaker’s strong rejection of the existence of alternatives to the accessible assumption or claim and later, ela ma constructionalized as an exclamative whose semantic meaning/function is the rejection of an accessible assumption or claim.

Hebrew ela ma! (literally: ’but what?’ idiomatically: ’of course!/’naturally’) can be compared to the English what else?!. Even though these two question constructions are not identical, they can be claimed to have evolved based on the same mechanism: the pragmatically drawn inference becoming semantic. This will be demonstrated in the following examples.

In the above example, what else is used an interrogative question (that awaits an answer). It is even justified by the speaker (“I feel like you’re not telling me stuff that I need to know”). The speaker here is waiting for his addressee to give him more/other important information.

Here, what else? Can be interpreted to mean ’of course/naturally’. Hypothetically, there could be other names to the “cow’s knees” but here the speaker uses a sarcastic tone for giving it its “primitive” name and not an alternative one, as if there are no other alternatives to that name. In this case, the only answer to the rheotical question what else? is what follows that question (here: a “cow’s knee”). The speaker answers himself in a rhetorical manner.

Lastly, examine 8:

(8)

[http://coffeechannel.blogspot.com/2012/05/nespresso-what-else.html/]

Here, What else? Is still followed by a question mark but is not discursively regarded as a question at all (i.e. the speaker does not expect an answer to his/her question). This advertisement conveys a message that there isn’t actually any alternative (coffee) to Nespresso, thus making Nespresso the only (=best) coffee.

This difference between (6)–(8) in regards to their modern discursive uses is srinkingly similar to Hebrew ela ma! and thus, it can be claimed to have undergone a similar path of change.

Type c exclamative CRQs was demonstrated in Bardenstein (2020a) with two types of Hebrew exclamative adverbials: eize-X! and ma-ze-X!. First, I have claimed that Eize-X! (lit: ’which-X! idiomatically: ’what (a)-X!/How-X!’ evolutionized as an exclamative upgrading adverbial via embedding constructions, conveying the speaker’s strong stance- usually one of resentment or discontent. After which the embedded excalamtive became independent. I hereby argue that this can be compared to the English what (a)-X! possible evolution, seen in Figure 3.

Figure 3:

The four stages of what (a)-X! possible evolution.

Secondly, I have argued that the construction [noun/pronoun] ma-ze X! evolutionized as an exclamative upgrading adverbial first due to the use of the rhetorical Question ma ze X?(!) as conveying rejection/resentment towards an accessible assumption or claim, followed by a rectifying Y to the rejected X : ma ze X(?), Y! (Y upgrades X). Then, the shortened upgrading construction was left (via ellipsis and hybridity): X, ma ze X! (X denotes Y). Lastly, the constrcution shortened again (via ellipsis) and became ma-ze X! in which construction, X performs upgrading to itself (replaces Y). See an example in Figure 4.

Figure 4:

The evolution of ma ze ’what is this/so’.

This above process resembles English “Yes/No-inversions” (as in “boy, was it good!!”). Even though these two constructions seem different from the wh-exclamatives previously discussed in this paper, I claim that the process via which a question construction turns into an exclamative also applies to this phenomenon in English since it has to do with strong speaker’s strong stance motivating and taking over the meaning of the construction, thus bearing an altogether different meaning of its parts. Thus, even though the yes-no inversion exclamative are typically claimed to perform grammatical substitution of yes/no questions (see, inter alia, Tanigucci 2018), I argue that this is not a grammatical exchange of places, but rather a process that is similar to the evolution of Hebrew ma ze-X! in the sense that an initial claim is rejected and replaced/rectified with an upgraded claim. It is a process whose initial stage is that of echoing the addressee’s prior question in order to express his/her strong rejection regarding the inappropriateness of the first speaker’s claim (as the speaker’s claim was not high enough on the contextual scale). Later, the echo fixates and becomes semantic- the construction becomes an upgrading exclamative, as in Figure 5.

Figure 5:

Yes/no inversion.

In this example, Speaker A asks Speaker B a yes/no question. In response, Speaker B reiterates the question of Speaker A in order to express his displeasure with the question, since the answer should have been clear to Speaker B. Speaker B uses an upgrading rectification in his response (’not only was it good, it was excellent’! = not (only) X, Y!) And then, we witness its constructionalization as an upgrading construction denoting Y: a high level on the contextual scale.^[14]

As already mentioned, there are other wh-exclamatives in Hebrew such as biblical and formal Hebrew ma ’what’ [ma-X! (X = adjective/verb)], as in ma nexmad! (=how lovely!) as well as kama (she-) X! ’how much (that-) X!’, as in kama yafe!/kama she-hi yafa! (’how beautiful!/How beautiful she is!’), eix (she-)X! ’how (that-)’ X! as in the colloquial eiz ta’im! (’how tasty!’) and eix she-ahavti ota! (How I loved her! = I loved her a lot). All in all, these wh-exclamatives have undergone semanticization/gramamticalization that is similar to the English ones due to the constructionalization of questions conveying pragmatic inferencing. In a nutsell, I claim, it is the persistent pragmatic use/function that became the semantic one.

As for type d, in Bardenstein (in press) I have elaboratively demonstrated the linguistic historical path of type d CRQ. Here the analysis will be given in a summative manner, starting out as a question, being recruited and mobilized for implicated negative stance, turning into an intensifying adverbial and lastly, the implicated negation became explicit and the question construction turned into a NPI downtoner. Throughout its history, it has persisted as expressing the speaker’s strong stance:

Stage I : Biblical mi yodea (ma) ’who knows (what)’ pragmatically conveying ’nobody knows (what)’:

While there are various examples in the bible for mi yodea pragmatically conveying ’negative bias’, there is only one biblical occurrence of mi yodea ’who knows’ followed (compositionally) by ma ’what’, in the Late Biblical book of Ecclesiastes (example 7), where mi yodea ma is composed of this negatively biased rhetorical question and the question word ma ’what’, conveying something like ’nobody knows what’:

In this example, ’negatively biased’ mi yodea ma is still not considered a construction. It is compositional, that is, its meaning is the sum meaning of its parts (pragmatically negatively biased mi yodea + ma).

Literally: For who knows, what is good for a man in life, the number of his days of nothingness are like a shadow, that who shall say to a man what will be after him under the sun?

Interpretation: For who knows what is good for a person in life, during the few and meaningless days they pass through like a shadow? Who can tell them what will happen under the sun after they are gone?^[15]

Later in the history of Hebrew, from second century to mid-nineteenth century, Hebrew was not spoken and was mostly used in Rabbinic writings and correspondence. In Early Rabbinic Hebrew which includes Mishnaic Hebrew (∼third century) and later the Babylonian or Jerusalem Talmud (∼fifth century) mi yodea is not present in those corpora and occurs only later (once) in the esoteric literature of Ma’aseh Merkabah (“Work of the Chariot”) or as often referred to: Ma’aseh Merkava (∼6 century) and then appears (once) in the historical book Jossipon (∼10 century), However, those use are still ’compositional’- they are still composed of the negatively biased ’mi yodea ma’ = ’nobody knows what’.

Stage II : In Revival Hebrew (mid-nineteenth century), mi yodea ma- X is used as a ’hypothetical-scenario CRQ, along with lule (’if not/unless’), conveying that the actual state-of-affairs is the desired one and that it is fortunate that this is the case. This use is similar to the English ’god (only) knows what would have happened if- X’. This is exemplified in (10):

Interpretation: ’And if it weren’t for some important home-owning that he had, who knows what had been his ending (of him)’ [The Travels of Benjamin III, Hi Jacob Abramovich, p. 6, 1896].

In the above example, mi yodea ma-X is not used to convey pragmatically negative bias. Mi yodea ma became a CRQ whose function/meaning is to hypothetically provide an opposite scenario (one which is unfortunate) to the actual state of affairs (that is explicitly negated by lule ’unless/if not’. In other words, we see here that mi yodea ma is restricted to misfortunate states-of-affairs, bearing a negative function of its own. This can be defined as a bridging context for mi yodea ma turning into semantically negative, since similarly to English god knows what (would have happened if-X), mi yodea ma’s meaning is no longer simply ’nobody knows (what would have happened)’. Instead, it functions as an implicit adjectival which means something like: ’something negatively unexpected’ (that could have happened if-X’).

Stage III : Twentieth century mi yodea ma conveying large/high quantities/qualities.

In the second half of the twentieth century, an additional development occurred. Mi yodea ma is used as an independent construction (not scoping over anything explicit) and functioned as an intensifying adverbial conveying ‘high quality/quantity’, as in the following:

Literally: I was willing to give who knows what in order to not look at the face of the device anymore.

Interpretation: I was willing to give anything in order to not look at that device anymore [The power of the telephone, 14.9.1962].

In (11), on the one hand mi yodea ma can be regarded as conveying something like: ’I was willing to give even ’something unexpected’/’even something that is unpleasant (for me to give)’: mi yodea ma functions here in an even-like fashion and this can be regarded as a bridging context since mi yodea ma also pragmatically conveys large quantity or quality of what the speaker was willing to give in order to be able not to look at the phone again.

During the same period, mi yodea ma we can be witnessed as denoting high quality or quantity, as in the following example:

In the above example, mi yodea ma ’who knows what’ is used and not mi yodea matay (’who know when’) and therefore we evidence the independent constructionalized mi yodea ma (Traugott and Trousdale 2013) as purely denoting large quantities (in this case of generations/periods/years). Here, Mi yodea ma in (12) significantly extends the long-term ancestral possession of the yarmulka to an extreme value on a scale.

Stage IV: The implicit-explicit negative cycle arises: NPI mi yodea ma: lo mi yodea ma ’not of high quality/quantity’.

Once mi yodea fixated as denoting ’high quality/quantity’, it can be and actually is negated. At this stage, after negated, lo mi yodea ma becomes a down-toner due to the mitigating role ascribed to negation, especially when it modifies a high or even extreme value (see Giora et al. 2010).

The characteristic of the predicate scoped over by mi yodea ma (in 13, the quality of the chorus) is conveyed as being ’not of high quality’. The speaker wishes to mitigate his/her utterance and since it is very difficult to negate a strongly high-scale word or phrase without implying that a less positive one is to some extent true, our negator becomes a CRQ down-toner. This is exemplified in (13):

Literally: Yael is saying that there is a chorus in the kibbutz “not who knows what” that performs on holidays and special occasions.

Interpretation: Yael is saying that there is a chorus in the kibbutz that is “not the best” that performs on holidays and special occasions. [Davar, 21.6.1966]

In (13) the quality of the chorus (or, actually, the quality of the singing) is down-toned as to mean something like ’a not very good/impressive chorus’.

Now, even though I have presented the four stages of constructionalization of the d-type CRQ, further changes can be witnessed. These are not changes in the negative/positive pragmatic/semantic meaning, but rather those of an implicit adjective scoped over contextually by lo mi yodea ma turning into an explicit one. This is exemplified in (14):

Literally: The handicraft teacher is not who knows what good , and all in all, we are bored there.

Interpretation: The handicraft teacher is not very good and all in all we are bored there. [Ma’ariv, 5.10.1989]

From a synchronic point of view, some would claim for the explicit preceding the implicit, i.e. explicit high quality/quantity adjective (tova ‘good’ in example 12) preceding implicit high quality/quantity adjective (as in example 13), but history tells us a different story. It was the constructionalization of lo mi yodea ma as implicitly conveying an adjective of high quality/quantity which preceded the explicit version. This is not a case of ellipsis, but rather of ’explicitation’ (making the implicit explicit) of the ’high quality/quantity’ scoped over by lo mi yodea ma. Just like what we witnessed in the above elaborated question-based exclamative cycle, it turns out that processes of historical change are based on the implicit losing its implicitness.

Lastly, I claim, even though this paper focused on Hebrew and English, CRQs are evident in many other languages that demonstrate polysemous question constructions. Examine the following:

The above examples demonstrate the fact that this paper’s analysis can be considered cross-linguistic as the above show that a question construction can be polysemous and used both as an interrogative and an exclamative (sometimes with a special addition as in 3–4). In addition, Hindi, Arabic, German, Russian and many other languages also demonstrate similar use of question constructions. Naturally, a more in depth into the history of those languages should take place; However, this paper shows that despite those languages’ differences (in many aspects), the mechanism which drives their evolutionary change is strikingly similar.

5 Summary

In this paper, I have analyzed the cyclic linguistic evolution of questions. I have first defined rhetorical questions and their typologies, I have then focused on processes of constrcutionalization of rhetorical questions in Hebrew and in English and compared the two. Lastly, I have focused on the grammaticalization of rhetorical questions throughout their history and have claimed for a gradual and well-established mechanism that applies to different languages. Lastly, I argued that similarly to the ’persistence principle’ of argumentative discourse markers (Bardenstein 2020b), it is the question’s initial rhetorically-recruited function of expressing the speaker’s strong stance that not only motivates its semantic change but also persists throughout its history, making it a polysemous construction. Although there are different types of wh-exclamatives and yes/no exclamatives, they demonstrate a similar path of linguistic evolution. All in all, it is clear that what motivates and persists throughout historical discursive (rhetorical or argumentative) linguistic changes of linguistic expressions is their initial core function.