Interjections, or exclamations, are historically non-mainstream research topics despite their significance in people’s everyday communication. Researches specializing in interjections over the past ten years are mainly on the part of European scholars (e.g. Cuenca, 2000; Gehweiler, 2008; Sauciuc, 2006; O’Connell et al., 2005, 2007, 2010; Norrick, 2007, 2009 ). Moreover, related journals nationwide are only a few (e.g. Wang, 1999; Gao, 2001; Li, 2007 ). At the current cognitive turn in linguistics, increasing researches on interjections come out from the angle of prototype theory and (cognitive) psycholinguistics (e.g. Cuenca, 2000; Li, 2007; Cruz, 2009; O’Connell et al., 2005, 2010 ).
It is hard to define interjections in a comprehensive way since they appear not simply in the form of a single “word”, but also sometimes as “sounds”, “phrases” or even “utterances” (Norrick, 2009: p. 867) . In consideration of the heterogeneity and multi-functions that interjections could perform, it is proper in this paper to narrow down the research field only to emotive/expressive interjections, which are primarily about giving expression to private feelings or emotions and demonstrate the core pragmatic function of interjections.
In this paper, emotive interjections (the second part) are classified first for a better presentation of the later discussion. Then it comes to the arguments (the third part) with regard to the conceptual content of different types of emotive interjections as well as the contexts based on which more specific “emotions” beyond interjections’ conceptual content might be specified. Further, paralinguistic factors and other non-verbal gestures (the fourth part) are also essential to interjections owing to their strong link to both medial and conceptual orality (i.e., interjections are especially intended for spoken language). In the end, it is the conclusion.
2. A Taxonomy of Emotive Interjections
The earliest classification of interjections can be traced back to Wundt (1900/ 1911, see O’Connell & Kowal, 2008: p. 135 for overviews of research on interjections) who distinguished the categories of primary and secondary interjections and acknowledged onomatopoeic interjections. In later literature, the general division of interjections as primary and secondary is academically accepted (Ameka, 1992) while the inclusion of onomatopoeic interjections is controversial.
In the following discussion, “emotionality and expressiveness” are emphasized as the basic criteria in the determination of interjections, just ignoring interjections with other functions. Resultantly, phatic interjections (e.g. hey in some cases), conative/volitive ones (e.g. psst, oi), filler/filled pauses (e.g. uh, er, um), HA-HA laughter and onomatopoeic interjections and the like are excluded from what will be covered in the taxonomy as well as in the further elaboration.
Based on the preliminary classification of interjections as primary and secondary, the new taxonomy of “emotive interjections” is hence a subdivision based on the rough criteria of interjections’ conceptual contents.
2.1. Primary Interjections
According to Ameka (1992: p. 105) , primary interjections are called primary because “they are not used otherwise” while secondary interjections are so called because they are “forms that belong to other word classes based on their semantics and are interjections only because they can occur by themselves non-elliptically as one-word utterances.” Put another way, primary interjections are non-lexical ones. Ward (2006) even treats several of them like oh, mm, yeah in his research as “non-lexical conversational sounds”, while huh and hmm, in the eyes of Bolinger (1989; cited in Norrick, 2009 ), are mere “intonation carrier” (see Norrick, 2009: p. 869 , for overviews of research on Primary versus secondary interjections). Nenova et al. (2001) investigate by way of computational analysis the relation between the sound patterns of primary interjections and their functional realization in the discourse.
As the prototypical center of interjections (Cuenca, 2000; O’Connell & Kowal 2008: p. 136) , these vocal units can be further divided into three categories in view of their conceptual meaning, which is not as specific as many other word classes (e.g. verb, noun), though.
2.1.1. Universal Emotive Interjections
Ameka (1992) , Norrick (2009: p. 868) summarized that “That is to say, many primary interjections do not express emotions, as is often maintained of interjections generally, but rather information states”. This is true in the case of mhm, mm, um, huh and the like, which might be used in discourse as indicator of self-reflection or (dis) approval. These items, therefore, go against the “emotionality and expressiveness” criteria for defining interjections here and have to be excluded from this paper. Those with both informative and emotive functions such as oh, ah (or aah), ooh (or whooh) are at issue, but the following discussion will only deal with their emotionality.
These “vocal sounds” seem universally similar across languages by producing common monophthongs (i.e. pure vowels) which phonetically minimize articulatory effort while pragmatically maximize communication (i.e. economy principle). Moreover, thanks to their universality, the acquisition of such interjections is unnecessary or not much a problem.
2.1.2. Specialized Emotive Interjections
Some other primary interjections which typically express certain specific emotions or feelings fall into the second category. Examples are ouch (pain-hurting), oops/whoops (mild annoyance), hurrah/hooray (triumphant excitement), crikey (surprise). With a more concrete emotive concept encoded, “specialized emotive interjections” are language-specific (not universal, as compared to the first type), and hence need acquiring in the second language learning, especially in that of spoken language.
2.1.3. Grammaticalized Primary Interjections
The above definition of primary interjections focuses on their “non-lexical” property. However, judging from a diachronic perspective, a third type of primary interjections stands out. They are fully grammaticalized elements after having undergone a semantic change (conversion) by pragmaticization of mean- ing and syntactic reanalysis and probably, a optional morphological change by a series of word formation processes (e.g. cutting, blending, borrowing).
Gehweiler (2008) investigates the case of “gee” diachronically to illustrate how certain primary interjections are derived from secondary interjections which originate in content words, or rather proper name (noun). For instance, “gee, jeeze, jeez (Louise)” have evolved from religious proper name “Jesus (Christ)” via secondary interjection “Jesus (Christ)”; similarly, “gosh, golly” from “God”; “blimey” from “God blind me” (Norrick, 2009) .
Still, one more fact cannot be denied that some of these grammaticalized interjections also serve as notional word (i.e. in addition to interjectional property, some of them also have other lexical meanings). “Gee”, for example, works as a verb like in the verbal phrase “gee sb. up” (to encourage sb. to work harder or perform better). Such a phenomenon of nationalization against grammaticalization deserves attention for a further look.
Conceptually speaking, although they are no longer lexical (barring a few exceptions) but fall into the primary category of interjections, grammaticalized primary interjections, due to its origin (religious proper name or phrases), genetically endowed with some concepts (see 2.2.1 and 3.4 for a detailed elaboration) in addition to their prototypical emotive property.
2.2. Secondary Interjections
Compared to the case in 2.1.3, secondary interjections can only be thought of as partially “grammaticalized” (in various degrees) since they retain its original lexical meanings which actually still hold the central place of the word while its interjectional use is only periphery and thereby occasional.
In general, secondary interjections have two lexical sources―nouns (including pronouns) and verbs.
2.2.1. Nominal Secondary Interjections
Secondary interjections originating from pronouns (e.g. you) and nouns (e.g. common nouns boy and nouns with roots in religion) are always “deictics”. Based on this observation, the influence of the derived word class (i.e. pronouns or nouns referring to persons) can be perceived. However, it seems that in conversational use some of these nominal secondary interjections are more referent-specific such as you or boy, while the others like God or Heavens, with the gradual process of conventionalization, turn out to have only a roughly corresponding entity (i.e., one who cries out God! does not necessarily refer to the God.).
2.2.2. Verbal Secondary Interjections
The case of verb-derived secondary interjections is slightly different, given their relatively less connection with the senses encoded from their original counterparts, namely the derived verbs. Verbal secondary interjections are simply for the (strong) emotion expressing of an action, an event or a state.
Specifically, there exist mainly two kinds of verbal interjections. One is those common verbs such as shoot, snap, boom, blast, which appear more civilized in use; the other type, such as fuck, shit, damn, often have their roots in sex and scatology and thus seem more vulgar. Such subdivision intends to point out the sociolinguistic significance of secondary verbal interjections.
3. Conceptuality and Context-Sensitivity of Emotive Interjections
This section unfolds around the following three questions: 1) Does interjections encode any conceptual content? 2) If it is an affirmative answer (it is actually is), then, to what degree do these concepts differ with regard to different types of emotive interjections as divided in the previous section? 3) From the perspective of interjections’ conceptuality, what functions does context perform in the interpretation of interjections?
The first question that whether or not interjections possess semantic content involves polemics (see Cruz, 2009: p. 242 , for overviews of research on (non-) conceptualist approaches to interjections). In the manner of structuralism, Wilkins (1992) and Wierzbicka (1992) defended the conceptual content of interjections by decomposing interjections into “primitive” deictic elements and “universal” or “near universal” concepts that capture their basic senses respectively. Wharton (2003) proposed a procedural shift in the analysis of interjections and denied their conceptual content for various reasons.
Since interjections covered in this paper have been restricted to emotive ones, it is relatively easy to tell, at least, some sort of core meaning all of them are taking along with―“emotion”―very general, vague or schematic emotion, though. Or, in the words of Sperber and Wilson (1997, see Cruz, 2009: p. 257 ), such concept are labeled as “pro-concept”, i.e. “a concept that is not fully fledged but needs to be fleshed out into a full concept”. In the case of emotive interjections, their conceptual content would amount to so-called “pro-concept” that have something to do with emotions and could be specified by means of context.
The second and third questions concerning the subtle relations between the conceptuality and context-sensitivity of interjections will be elaborated in the order of interjections’ five subcategories.
3.1. Primary Interjections
3.1.1. Universal Primary Interjections
Interjections in the first subclass are relatively stable in the inventory of interjections across languages. It does not indicate the stability of their forms, but that of their phonetic representations. Therefore, these interjections are deemed more as human vocal sounds or sometimes sound combinations than words or phrases. Conceptually, “emotions” behind them is optional in such a wide range that one item (e.g. oh) would have various interpretations in different situations. Context, as a result, is strongly bound with these vocal units.
It is just because of their phonetic universality and great sensitivity to context that this kind of interjections is helpful in a way for the general human communication (i.e. across languages).
Suppose one scene in an international corporation (i.e. a multi-lingual community), a Chinese colleague broke a vase by accident and cried a free-standing utterance “哦!”―a spontaneous emotion outburst and a sound symbol of the crier’s inner state of mind―when an English who knows little about Chinese passed by. Could the English man discern what specific emotion the Chinese intended to convey by the interjection “哦!” (or “Oh!” as its counterpart in English)? The answer is quite probably affirmative: for one thing “哦!” or “Oh”-like primary interjections are highly context-dependent, and for another the context in the above illustration is informative enough. For sure, a smattering knowledge of several “universal” interjections is of not much use for further interactions; but to say the least, there do exist such interjections as “emotive pro-concept” -carrier that could contextually strike a chord in the minds of both addresser and addressee in spite of language gaps.
3.1.2. Specialized Primary Interjections
Cruz (2009) expounds on the existence of a continuum of more and less conceptual items for interjections. Interjections such as ouch (for “pain”), oops (for “disgust”) and hurrah (for “triumphant excitement”) are defined as “specialized” just because they encode some concepts that are related to more specific types of emotions, beyond the core pro-concept (i.e. “emotion”) shared by all the emotive interjections. But it seems that they still contain some ineffable meanings which are not lexicalized because individuals lack stable concepts for them, so these meanings might only be encodable by means of phrasal or sentential structures. Therefore, ouch could mean “intense pains”, “unbearable pains” or “a more intense pain than the one you experience when you just scratch your arm” or some other kinds of “pain”. In a spoken discourse, all these nuances are quite possibly hidden behind the context, which, in this example of ouch, helps specialize the “pain” emotion.
Unlike universal interjections which benefit to some extent the human communication, specialized primary interjections which are always phonologically regular constitute obstacles to cross-cultural communication, because for non- English speaker nor learner ouch is just as strange as a totally new word. In other words, a hearer of these interjections may not even be able to distinguish them as emotion indicators, let alone conceptually and contextually specify them.
Thus, it is reasonable to claim the specifying functions that context performs are, in this case, restricted by hearers’ mastery of the interjections’ specific concept.
3.1.3. Grammaticalized Primary Interjections
As compared to the first two “unadulterated” primary interjections, grammaticalized primary interjections with roots in lexical words (esp., nouns) are in nature the final results of grammaticalization process which proceeds all the way from content words to primary interjections via secondary ones. For this reason, related discussions concerning this type of interjections are found in 3.2.
3.2. Secondary Interjections
Secondary interjections derive from other content words or lexical categories. From the angle of semantic relation, interjectional sense of that particular word, together with its other meaning(s), jointly asserts the polysemic spectrum of the word, though interjectional uses is periphery while lexical ones are more central. (This is different from “grammaticalized primary interjections”, the prototypical concepts of which are interjectional.)
Conceptually speaking, “since they are originated in other lexical categories such as nouns and verbs as mentioned, which do indeed have a semantic content, these interjections may preserve part of the initial conceptual content that the words that originated them had or, at least, some conceptual content related to it as a result of metaphorical or metonymic process” (Cruz, 2009: p. 245) . This general statement on the concepts encoded in secondary interjections points out the existence of levels of conceptuality, that is to say, for secondary interjections, there always exists a problem of grammaticalization―to what extent do they grammaticalized? The more grammaticalized they are, the less conceptual connection they keep with the derived content form, hence the more dependence on the context for meaning specification.
For a further look into the conceptuality and context-sensitivity of secondary interjections, the two subclasses are respectively discussed.
3.2.1. Nominal Secondary Interjections
Noun- and pronoun-derived interjections are in most cases personal deictics, though not necessarily as specific as when they are in lexical use. Due to such indexical nature (i.e. conceptually related to “person”), another concept encoded by nominal interjections (in addition to “emotion” pro-concept) is therefore in the category of “person” or sometimes “personalized object” (e.g. a master shouted to his pet dog biting the carpet “Boy! What are you doing?” Here, personal “boy” is for a dog). Put it another way, this type of interjections is sensitive to context in such a degree as to seek out what specific emotions are involved and to whom.
However, the case of religion-related interjections such as “God” and “Heavens” is slightly different. As proper names, these lexical terms have conceptualized entities to refer to, and it might also be true for the derived interjections at their very advent. People resort to external entities (deity, heavens and earth) to express human’s inner mental activities (helplessness, disappointment, excitement etc.). But with the time going, the process of grammaticalization gradually works. They have become conventionalized with the result that they are routinely heard not in their religion-related referring sense but in their interjectional message―an outburst of emotions. When in the end they are fully grammaticalized as “primary interjections” (i.e. “grammaticalized primary interjections” illustrated as in 2.1.3), their original concept of the divine referents has already subconsciously disappeared from addresser’s mind, and a vague pro-concept of strong “emotions” (the core meaning of all emotive interjections) takes the place instead. As for their context-sensitivity, they are not as contextual as deictic interjections (e.g. boy or you), since the entities they originally point at have been pragmaticized into emotive expressions, unless the addresser himself is meant to make reference to, for example the God or Jesus Christ. Hence, in most cases, context is needed to specify the general emotions.
3.2.2. Verbal Secondary Interjections
Possibly owing to the features of verbs, emotions expressed by verb-derived secondary interjections are always concerned with an event, an act or a state― this differs from noun-originated ones which have in one way or another subtle senses of the referred “person” or “entity”.
Conceptuality of verbal interjections is not specific, but broad pro-concept of “emotion”. It seems that senses of verbs from which the interjections originate have little influence on the conceptual content of the derived interjections. Still, one inherited concept could be perceived in the derived interjections, namely, the emotional coloring. To be specific, common verb-derived interjections (e.g. shoot, snap, boom, blast) are used in a neutrally civilized way, while others with their roots in sex and scatology (e.g. fuck, shit, damn; so called “four-letter words”) are deemed as deprecatory and impolite. Apart from this stylistic connection with original content words, not much conceptual relations are found. In short, verbal secondary interjections are indicators of strong emotions, though socially different, and context is therefore still used to specify what the strong emotion is.
To make a summary, emotive interjections in this section are investigated from the combined perspective of semantics (their “conceptual content”) and pragmatics (the functions of “context”), in an attempt to answer the three questions raised at the beginning.
It is too absolute to assert a non-conceptualist approach to interjections. The basic conceptual content or called pro-concept encoded by interjections is “emotion”, and there exist degrees of conceptual specification for different types of interjections. Specifically, “emotions” conveyed by universal primary interjections are superordinately broader than specialized ones, the concept of which is always related to a comparatively more specific emotional category (e.g. ouch for “pain”, though still very general). For secondary interjections, whether or not lexical meanings of the original nouns or verbs have influences on those derived interjections depends on the degrees of grammaticalization. Nominal secondary interjections are always personally deictic―to/for whom the emotions are expressed, but the referring function of religion-related interjections like God or gee is routinely weak. Verbal interjections appear to be more thoroughly grammaticalized, since they bear little conceptual connections with their origins except for certain emotional coloring differentiation between interjections like shoot and fuck. On the whole, they all conceptually express “emotions”.
Emotive interjections are actually attempts to lexicalize “emotion” in spoken discourse. However, based on the above elaboration, it is found that the conceptuality of interjections is mostly “crude” and “elementary”. Many specific emotions have not yet been lexicalized as interjections. Such ineffable embarrassment of interjections calls for the help of context. The major role that context plays is to embody interjectional “emotions”. Further, for different classes of interjections with such conceptual nuances as discussed above, context functions in a slightly different way. Specification of “emotions” is a cognitively level- by-level process (i.e. from the broadest emotion to less broad one, then to more specific ... and finally to the most specific―the exact emotion the utterer means in that context.) Thus, for primary interjections, specialized ones are more easily cognized than universal ones, since a cognitive process/level from a pro-concept of “emotion” to less broad concept of “pain” (to take ouch for example) is left out. As far as nominal secondary interjections are concerned, context first helps identify the “target” to whom the emotions are expressed, then performs the function of specification. Verbal interjections keep not much conceptual heritage from their derived verbs except for some emotional coloring, and rely on context to a great extent for clarification.
4. Paralinguistic Factors and Gestures
Furthermore, in view of its oral nature, the understanding and interpretation of interjections cannot be solelylinguistically-based. Some paralinguistic factors should also be considered so as to better elaborate on the conceptuality and context-sensitivity of interjections.
O’Connell et al. (2007, 2008, 2010) have done empirical research comparing interjections in written and spoken discourse. Their results of several case studies indicate that interjections have strong links to both medial and conceptual orality. The number of interjections in written dialogues of a novel (conceptual orality but medial literacy) is much less than that in the novel-based film or play (both medial and conceptual orality). Such association with spontaneous spoken discourse accounts for the oral nature of interjections. Since there are no utterances or speech signals that lack the presence of a voice (all the properties of which are called “paralanguage”), paralinguistic factors (e.g. pitch, volume) accompanying the uttered interjections should not be neglected in the research.
Other supersegmental and prosodic factors are also informative for the proper interpretation of interjections. Intonation, especially, has been emphasized by Fraser (1996) as imposing a pragmatic meaning on interjections in addition to their invariant conceptual meaning. Bolinger (1989; cited in Norrick, 2009 ) even goes so far as to say that some interjections like huh and hm function as mere “intonation carrier”.
Besides these voice-related elements, gestures have also been related to interjections. gestures or other non-verbal expressions can substitute for interjections (Ameka, 1992; cited in O’Connell & Kowal, 2008: p. 136 ) in view of the involvement of the whole human body in expressive spontaneous spoken discourse.
In short, the broad concepts of emotive interjections would have to be not only contextually adjusted, but also take into consideration paralinguistic and gestural factors.
In conclusion, interjections encode conceptual content and are context-sensitive. Conceptual nuances among different types of interjections are described above as comprehensively as possible. For primary interjections, the case is notably different between universal and specialized ones. Universal primary interjections, which benefit the human communication, express “emotions” that are conceptually optional in such a wide range that one item would have various interpretations in different contexts; while specialized primary interjections constitute obstacles to cross-cultural communication, because they are always phonologically regular, and the specific functions that context performs are restricted by hearers’ mastery of the interjections’ specific concept. For secondary interjections, there always exists a problem of grammaticalization, which results in various levels of conceptuality. The more grammaticalized those secondary interjections are, the less conceptual connection they keep with the derived content form, hence the more dependence on the context for meaning specification. Although the contextual functions differ in some measure for different conceptual contents that interjections have, the major role of context is to embody interjectional pro-concept “emotion”. In addition to the above discussion, the oral nature of interjections suggests that paralinguistic and gestural factors should also be taken into consideration in the conceptual and contextual interpretation of interjections.
Still, improvement in the following three aspects is possible. Firstly, since the heterogeneous and multi-functional property of interjections makes it hard to define and classify them in an absolute and clear way, it is quite possible to find fuzzy boundaries among different types as divided in the taxonomy. So, the cases of exceptions, if there are, could be the subject in later study. Secondly, examples in the paper are based on everyday experiences and observations, and are thought of as representative and specific to illustrate related ideas. So, corpus- based data are required for a more persuasive argument. Lastly, the discussion of context in this paper is not sufficient enough, and therefore, a more detailed study of the contextual effect on interjections is possible and it is hypothesized that the case of interjections might be supportive of radical contextualism in view of their orality (i.e. interjections are used mostly in spontaneous spoken language).
As a widely used word class with so many research topics concerned, interjections are indeed a research field worthy of further study.
 Nenova, N., Joue, G., Reilly, R., & Carson-Berndsen, J. (2001). Sound and Function Regularities in Interjections. In: Proceedings of Disfluency in Spontaneous Speech (pp. 49-52). Edinburgh, Scotland.
 O’Connell, D. C., & Kowal, S. (2008). Communicating with One Another—Toward a Psychology of Spontaneous Spoken Discourse (pp. 133-141). Berlin/Heidelberg: Springer.
 O’Connell, D. C., & Kowal, S. (2010). Interjections in the Performance of Jane Austen’s Pride and Prejudice. Journal of Psycholinguistic Research, 39, 285-304.