Hedges show the interesting distributional feature of frequently clustering together in ‘harmonic combinations’, as Halliday (1970) explains. Furthermore, Lyons (1977: 807- 808) employs the term “modally harmonic” to explain the “kind of concord running through the clause, which results in the double realization of a single modality. In addition, Hyland mentions the presence of “non-harmonic combinations” such as may+certain (hedge+booster), which were rarely used in his corpus of RAs. (cf. Hyland 1998a: 152) The author argues that items of an equal value such as may+probably (less tentative) and might+suggest/seem (more speculative) tend to cluster more often together. (cf. Hyland 1998a: 153) Concerning clustering of hedges or "harmonic combinations", Hyland’s (cf. 1998a: 150ff) work gives a valuable insight of their context-based features. It also presents the notable frequency of clusters of hedges and shows in a detailed manner which the most frequently occurring hedges are. This will be discussed at the end of this subsection. However, there is, of course, no reference to harmonic clusters of boosters, which are presented in my study.
In a subsequent paper from the same year, Hyland (cf. 1998b: 350) mentions Lyons (1977), while treating the clustering of hedges with other hedges or boosters with other boosters, particularly in the same clause complex. Hyland adds that hedges and boosters can occur together in order to create different rhetorical effects and differentiate between propositional information that the author considers as factual and what he or she sees as more questionable. (cf. Hyland 1998b: 350) In the given examples, Hyland clearly portrays the author’s epistemic moves between what is safely accepted and the novelty of tentative conclusions. Yet, he treats the factor from a general perspective, and this can be seen in his choice of words “stretches of discourse”. Below, an extract (3.8) from an RA in Philosophy including a cluster of boosters (and hedges) is presented. I have underlined several lexemes that could have also been considered by Hyland (1998b).
3.8 […] A few interesting exceptions we obviously do not see a static image as moving. Suppose, then, that we say that static images only depict instants. This too creates problems, for it suggests that we have a recognitional capacity for instants, and this seems highly dubi- ous. (Hyland 1998b: 350)
In this extract, Hyland only focuses on the cluster of obviously with highly, which belong to two different clause complexes. He does not concentrate on the co-occurrence of hedges and boosters in the same clause and span, for example: seem and dubious together with highly. It is clear that the author of the text has used hedges such as suggest and suppose to lower the effect of the propositional information with which he or she does not agree, as well as boosters like obviously and highly to strengthen his or her argument. A noticeable controversial choice of the author is the use of the hedge
seem next to the booster highly. My research shows the strategic use of hedges and boosters within the clause complex (with a particular focus on scope) to increase and decrease their effect, as well as the erroneous use of these devices. Hyland (1998b: 351) effectively presents the co-occurrence of hedges and boosters in his subsequent discussion, but he maintains a focus on the text section or paragraph level and not on the clause complex level. The author provides a more detailed view of (harmonic) clustering in his previously mentioned influential work, Hedging in Scientific Research Articles. (cf. Hyland 1998a: 150-153)
The following extract 3.9 from my corpus shows the harmonic clustering of seem with hint and the non-harmonic clustering of both with fact. As the example infers, both hint and fact fall within the span of seems and this makes this cluster relevant for my analysis. The whole second part […] and in this way she seems to hint at the fact that […] is quite tentative due to its rather interpretative and assumptive nature. The odd addition here is fact.
3.9 When Liz is asked if she likes them, she immediately replies ‘‘of course’’, and in this way she seems to hint at the fact that she is not surprised at all to find this kind of pastry in a Roman setting. (CIAOE15FBL_58)
An interesting feature of Albanian, presented by Çeliku et al. (cf. 2002: 375), is the clustering of the same hedge/booster right next to each other such as ndoshta, ndoshta, (probably, probably), and sigurisht, sigurisht (surely, surely). However, this particular co-occurrence did not appear in my corpora of academic writing. These examples showed how hedges and boosters cluster together within the same clause or clause complex, with a particular focus on their span.
This paragraph shows some overarching quantitative results on the clustering of different word classes and specific lexemes. Hyland (cf. 1998a: 150-153) investigates the frequency of clustering of hedges within the same clause complex. Out of 1568 occurrences of hedges in his corpus, 668 or 42.6% co-occurred in the same clause complex. In terms of word class, the category that clustered more often with other hedges were modal auxiliaries (147) and lexical verbs (188). The most frequently occurring clusters were lexical verbs-adjectives, modal auxiliaries-adverbs and adjectives-adverbs. (Hyland 1998a: 150). In addition, lexical verbs were part of 28% of the total clusters and they mostly belonged to the initial part of the cluster, while adverbs were less likely to be placed initially. (Hyland 1998a: 150). The numbers of hedges within clusters are interesting as well. Clusters with three or more hedges were rare in RAs (they occurred only 43 times). The most frequently occurring combinations with more than two hedges were modal auxiliary-lexical verb together with an adverb (six occurrences), a strategy (seven occurrences) or an adjective (twelve occurrences). (Hyland 1998a: 151). There are interesting results from the comparison of different lexemes. The hedging lexemes which tended to cluster more were suggest, possible, would and indicate. The most frequently employed pairs were may-suggest (eight), may-possible(ly) (seven), could-possible(ly) (five) and may-indicate (five) (Hyland 1998a: 150-151). These different angles of observing clustering are very useful for an
analysis of hedges and boosters. My study also focuses on different types of clusters betweenthe selected lexemes.
This subsection showed relevant examples of clustering and a previous quantitative analysis of this phenomenon. It is a relevant part of the literature review because it gives a clear idea of how I will proceed in my analysis on clustering and what I will focus on. While Hyland refers to the previously mentioned co-occurrences as clustering within the same clause complex (not necessarily within the same span, e.g. if divided by but), the examples he provides are always intertwined within the same span (“This would probably suggest that”) (D2: 265). (Hyland 1998a: 152) Harmonic and non-harmonic types of clustering within the same span will be analyzed in subsections 5.1.3.4 and 5.2.5.4. Span or scope, which are defined in the following subsection, appear to be inseparable from clustering.