Placeholders in crosslinguistic perspective: abilities, preferences, and usage motives

1 Introduction

Naturally occurring speech is full of features absent in idealized language use, including sound stretch, cut-off, false start, and the use of interjective hesitators (e.g., uh, um). Another notable feature of spontaneous speech is the use of placeholders (PHs) (Amiridze et al. 2010; Channell 1994; Enfield 2003; Hayashi and Yoon 2006; see Section 4 for further references). Building on Seraku (2022a), Seraku et al. (2022: 390) define PHs as (1).

For illustration, consider (2)–(3).

In (2), Naomi cannot produce the target form ballpoint pen due to memory lapse, and she resorts to the PH whatchamacallit, placing it at the object slot of the sentence structure. In (3), Naomi remembers the target form fly but prefers not to verbalize it for reasons of politeness. She then uses the PH you-know-what, placing it at the subject slot of the sentence structure.

Previous studies have described the types of motives for using PHs in a wide range of languages, but it remains to be seen (i) what crosslinguistic tendencies (if any) are found with regard to the usage motives of PHs and (ii) how such tendencies are theoretically captured. I aim to answer these two questions by surveying PH data from 56 languages and presenting an analysis from a cognitive–pragmatic point of view, couched within the framework of Relevance Theory (Sperber and Wilson 1995; Wilson and Sperber 2012).

Section 2 reviews previous studies to further clarify the concept of PH and explicate the objectives of this paper. Section 3 puts forward my proposals about the usage motives of PHs, and Section 4 tests them through a crosslinguistic survey. As will be shown, the usage motives of PHs are affected by various factors (e.g., cognitive–pragmatic, lexical-semantic). In the present paper, I concentrate on cognitive–pragmatic factors by proposing an implicational hierarchy based on the relevance-theoretic concept of the speaker’s ‘ability’ and ‘preference’ and applying it to the crosslinguistic data; non-cognitive–pragmatic factors will be considered when the proposed hierarchy does not seem to account for the data. As a way of conclusion, Section 5 discusses empirical and theoretical implications for subsequent research.

2 Previous studies

Definition (1) divides PHs into two types in terms of usage motives. Following Seraku et al. (2022: 391), we distinguish between them with the abbreviations ‘PH_A’ and ‘PH_P’.

1. PH_A refers to a PH whose use is motivated by a speaker’s abilities (e.g., [2]).

2. PH_P refers to a PH whose use is motivated by a speaker’s preferences (e.g., [3]).

Definition (1) is commonly adopted in the literature (e.g., Channell 1994: 162; Cheung 2015: 275; Jucker et al. 2003: 1750). Channell (1994: 162) states that PHs are utilized when a speaker “does not know/has forgotten name/noun” or “does not wish to use name/noun”. In some studies (Amiridze et al. 2010; Hayashi and Yoon 2006), however, PHs are narrowly defined. Hayashi and Yoon (2006) examine demonstratives in diverse languages and identify their interactional uses such as the ‘placeholder use’ and the ‘avoidance use’. In our terms, the former corresponds to a PH_A, and the latter to a PH_P. That is, Hayashi and Yoon (2006) reserve the term placeholder for PH_A, but they at the same time admit that some PH_A forms serve as a PH_P too. Thus, even for those who define PHs narrowly, their usage motives are worth investigating.

A typical situation for PH_A is where a speaker cannot recall a target form (see [2]). Other situations include those where (i) a person who is reading a passage aloud encounters illegible characters (Cheung 2015: 276), (ii) a speaker just does not know the target form (Enfield 2003: 103), and (iii) an appropriate word is unavailable in a language due to a lexical gap (Hengeveld and Keizer 2011: 1965). On the other hand, a PH_P is often used as a substitute for a taboo or face-threatening expression (see [3]). In yet another situation, a speaker may withhold a target form to pleasantly surprise their addressees, as in (4)^[1]. (When an example is cited from previous work, minor amendments have been made for the sake of consistency.)

In (2)–(4), a PH substitutes for a nominal element. Whilst the prototypical category of a target form seems to be nominal (Podlesskaya 2010: 13–14), it may also be verbal, adjectival, or adverbial. In some languages, one and the same PH form may substitute for a wide array of morphosyntactic elements. In Mandarin, a PH form is created based on the distal demonstrative na ‘that’, the general classifier ge, and a wh-word (e.g., shenme ‘what’), as in na ge shenme, and it can occupy not only a nominal but also a verbal or an adjectival slot (among other things), as illustrated in (5)–(6). (Na ge drops when an adjective is substituted for, as in [6].)

The slot-occupying property of PHs distinguishes them from interjective hesitators such as uh and um, which do not occupy any substantial grammatical slot. Thus, in Can you hand me that, um, ballpoint pen?, the hesitator um occupies neither a nominal nor a predicative slot.

Despite a wealth of previous studies on PHs in individual languages, crosslinguistic work is scarce. Vogel (2020) surveys PHs in 29 languages, but her concern is morphological, and no sentential examples are discussed. Hayashi and Yoon (2006) analyze PHs in 12 languages, but their analysis is limited to demonstrative-related PHs, and they do not consider crosslinguistic tendencies of usage motives. Podlesskaya (2010) is noteworthy in making several typological claims, but the issues of usage motives are not systematically addressed.

The principal aim of this paper is to fill this lacuna in the literature. More specifically, I will make several predictions about the usage motives of PHs from a cognitive–pragmatic point of view (Section 3) and evaluate these predictions against a wide range of examples, including new data from Spanish and Finnish (Section 4).

3 Proposal

Not much theoretical attention has been paid to the semantics and pragmatics of PHs. Channell (1994: 163) sketches a Gricean account of English PHs, and Cheung (2015) presents a formal-semantic account of Mandarin PHs. Recently, Seraku (2022a) has analyzed Japanese PHs from the perspective of Relevance Theory (RT). The present paper builds on Seraku’s account.

In RT, a stimulus (e.g., utterance) is ‘relevant’ iff it yields cognitive effects (e.g., deriving a new contextual assumption). Relevance is a matter of degree: other things being equal, the more cognitive effects a stimulus yields, the more relevant it is. There is another factor affecting the degree of relevance: other things being equal, the less processing effort a stimulus demands, the more relevant it is. According to Seraku (2022a), a hearer presumes that the speaker uttered a PH because it is the most relevant stimulus modulo the speaker’s abilities or preferences. This claim follows from the second clause of ‘presumption of optimal relevance’.

The RT notions of ability and preference encompass a wider range of situations than their non-theoretical analogues. First, a speaker, if drunk, may be unable to enunciate words or may fail to recall a word. Also, a speaker reading a letter aloud may encounter illegibly blotted words. These situations are all concerned with a speaker’s abilities, though, with the ordinary word of ability, one could say, e.g., that a drunken person has the ability to enunciate words but alcohol temporarily makes her slur words. Second, a speaker as a lawyer may avoid disclosing certain information due to confidentiality even if she knows that it is what the hearer needs. In another situation, a speaker may decide not to utter a certain word if she feels it is socially inappropriate. In these situations, a speaker could, but prefers not to, produce a certain expression.

Seraku (2022a) analyzes the demonstrative-related PHs in Japanese.^[2] He shows how their encoded meaning (‘procedural meaning’) interacts with contextual assumptions and pragmatic principles, so that the referent of a PH is recovered and various contextual implications emerge. With example (8), I only illustrate part of his analysis pertinent to the present study.

The hearer presumes that the speaker uttered the PH ano due to his abilities or preferences. As indicated by the interjective hesitators uun and anoo, it is highly likely that the speaker produced ano because of her abilities (e.g., memory lapse), rather than her preferences. After K-Gakuin is produced, the hearer supposes that the PH stands for the name of a university and makes use of the concept encoded in this target form to construct the proposition expressed and derive various contextual implications.

‘Ability’ is a more fundamental notion than ‘preference’. If one is unable to do something, she is not in the position to prefer to do it or not; she simply cannot. On the other hand, if one is unwilling to do something, she could, but choose not to, do it. This point is illustrated in (8). As stated above, the speaker utters ano for her ability reasons. The target form seems to become accessible to her later, but she produces it only partially, as in K-Gakuin, presumably due to her preferences. She understands that her speech is recorded for the creation of a corpus and will be publicly available once the corpus is published; she may thus have felt it was inappropriate to mention a particular school name. That is, at the time of uttering ano, she could not provide the target form, but at the time of uttering K-Gakuin, she could, but chose not to, mention it.

I suggest that this ability/preference asymmetry puts a cognitive–pragmatic constraint on the uses of PHs. More specifically, I propose the implicational hierarchy in (9).

The intended reading of (9) is: (i) if an expression has a PH_P use, it has a PH_A use too, but (ii) if an expression has a PH_A use, its PH_P use is not guaranteed.^[3] The former entails (10a), and the latter (10b)–(10c). We also predict (10d), with the assumption that in an implicational hierarchy x > y, x is as likely or more likely to be attested as compared to y (Corbett 2010: Section 2).

These predictions will be explored from a crosslinguistic perspective in the next section. As for (10d), statistical evidence cannot be presented, but suggestive data will be discussed.

4 Crosslinguistic survey

This section tests predictions (10), considering PH data in 56 languages. For ease of exposition, the target languages are classified into four groups (see the Appendix for the list of languages). This classification is just for the sake of presentation, with no theoretical significance intended.^[4]

1. Indo-European (Section 4.1)

2. Austronesian (Section 4.2)

3. Other Asian languages (Section 4.3)

4. Other non-Asian languages (Section 4.4)

The data come from two kinds of sources. Spanish and Finnish examples were gathered through consultations with a Spanish speaker (female, born in Lleida, Spain) and a Finnish speaker (female, born in Kotka, Finland). Examples in the other languages rely on past studies, except for German and Russian, for which supplementary examples were offered by a German speaker (male, born in Bremen, Germany) and a Russian speaker (female, born in Moscow, Russia). For each language family, I considered every PH-related work that caught my attention, though a few studies that present only sporadic data in multiple languages were excluded. For previous studies that do not adopt the term placeholder, I searched for other key terms such as filler and dummy as well as English translations such as whatchamacallit, whatshisname, and thingy.

Before proceeding, a caveat is in order. As mentioned in Section 1, I do not claim that the usage motives of PHs are fully governed by the hierarchy in (9). My contention is that whilst various types of factors (e.g., lexical-semantic, cultural) are at work, the cognitive–pragmatic constraint in (9) acts as one of the significant factors. In this paper, I focus on the hierarchy in (9) and explore to what extent predictions (10) hold crosslinguistically. This enquiry will show not only the usefulness of the hierarchy in capturing the usage-motive patterns of PHs but also its limitation. As will be noted, there are problematic cases for the hierarchy (or the predictions drawn from the hierarchy), and I will consider non-cognitive–pragmatic factors for such cases.

5 Implications

We have crosslinguistically investigated the form–function relations of PHs and have revealed several usage-motive patterns, which are partially accounted for by the cognitive–pragmatic notions of ‘ability’ and ‘preference’ in Relevance Theory (Sperber and Wilson 1995; Wilson and Sperber 2012). This outcome may be taken to enhance the validity of the framework because a communicator’s abilities and preferences form part of the fundamental relevance-theoretic concept of ‘optimal relevance’ (see Section 3).

As emphasized at the outset of Section 4, various factors are pertinent to the usage-motive patterns of PHs. The present paper has concentrated on the cognitive–pragmatic factor in (9), from which the predictions in (10) were derived and tested crosslinguistically. In future work, non-cognitive–pragmatic factors (e.g., cultural, lexical-semantic, morphosyntactic) should also be carefully examined to obtain a fuller picture.

It has been pointed out in Section 2 that only a couple of crosslinguistic studies have been conducted thus far (Hayashi and Yoon 2006; Podlesskaya 2010; Vogel 2020). The present study fills this gap in the literature, comparing a variety of PH data (including new data from Spanish and Finnish) in a principled way. Our survey encompasses diverse language families and areas, but its coverage is far from comprehensive. Most notably, languages from all ‘macro-areas’ (i.e., Africa, Australia, Eurasia, North America, Papunesia, South America; see the Appendix) are included, but our target languages mostly belong to Eurasia or Papunesia. Further data need to be collected from a wider range of languages to arrive at more robust generalizations.

Another, more specific implication of the present study springs from our hierarchy: ‘PH_A use > PH_P use’. Based on this hierarchy, we drew predictions (10) in Section 3 (see also footnote 3), but it would be profitable to consider whether the hierarchy is construed in different ways, possibly in other linguistic fields. A promising domain is that of language acquisition. As stated in Bowerman (2010: 614), in a hierarchy ‘x > y’, the presence of y in a child’s language implies that the child has acquired (or has started acquiring) x. With this insight, it may be tempting to derive a prediction like (30).

This prediction relates to first language acquisition. Implicational hierarchies have been used in applied linguistics and language teaching too; see, e.g., the Projection Model (Zobl 1983 and references thereafter). Distinct predictions could thus be formulated in these fields. In the area of language loss, a prediction like (31) might be derived in a manner converse to (30), provided that the more basic use/function (i.e., PH_A use) is retained longer than the less basic use/function (i.e., PH_P use).

These predictions are only suggestive and speculative, and more work is required to substantiate them and to make further predictions in broader fields.^[12] As long as they are formulated in explicit terms and present non-trivial claims, they should be fruitfully testable crosslinguistically and may open up new questions.