Epistemic modality is the expression of the speaker’s stance (Biber et al. 1999) and it is often linked to philosophical approaches and to logic, as it is concerned with the concept of truth of a particular situation or proposition as perceived and expressed by the speaker (Hale 2000). Epistemic modality is organized in the form of a cline with varying degrees of likelihood. This paper deals with the expression of epistemic modality in English and its translation into Spanish, as well as with the semantic shifts affecting certain epistemic markers and how these shifts are represented in translations.

In English, epistemic modal meanings are most often associated with the expression of modal verbs, and it is true that these linguistic units have specialised in the actualization of these particular functions. However, there are a few modal adverbs that can also be used to express epistemic force, alone or in conjunction with a modal verb, to reinforce the modal function, as in examples (1) and (2):

As a matter of fact,

[t]here are very few adverbs that may occur with epistemic modals (or even modals in general). They are confined to those that themselves express judgements, and so occur largely pleonastically – possibly, perhaps, etc., for possibility, and surely, certainly, etc., for necessity.

Palmer 1979: 57

In this paper, I focus on the expression of epistemic modality by means of certain adverbs in English and their translations into Spanish.

The traditional concept of “epistemic modality” partly overlaps with more recent studies devoted to the category of “evaluation,” “the broad cover term for the expression of the speaker or writer’s attitude or stance towards, viewpoint on, or feelings about the entities or propositions that he or she is talking about” (Thompson & Hunston 2000: 5). This is a particularly complex field of study, since the degree of commitment of the author may vary a lot, as we have already seen, and may be represented linguistically in many different ways. The broad category of “evaluation” and the linguistic items that embody it have been referred to by various authors using different terms, such as “disjuncts” (Quirk et al. 1985), “evidentiality” (Chafe 1986), “stance” (Biber and Finegan 1989), “attitude” (Halliday 1994), “discourse markers” (Fraser 1999), and “pragmatic markers” (Aijmer 2005). These elements are not part of the propositional content of discourse, but rather comment of the proposition itself, and may express a variety of functions in addition to opinion, mainly the construction and maintenance of interpersonal relationships and the organization of discourse.

As Aijmer (2002, 2005) points out, a number of the traditional epistemic adverbs in English are undergoing a process of grammaticalization and developing into textual and pragmatic markers. In this process, some adverbs that have typically been associated with the expression of stance are increasing their semantic range, and can be used in addition to express emphasis on the part of the speaker or to relate the following proposition to the previous discourse or to the speaker’s attitude as in example (3):

In example (3), the adverb certainly indicates a strong commitment on the part of the speaker, but it also acts as a link between two parts of discourse, as we expect that there must have been some type of evidence mentioned previously in the text to justify the speaker’s certainty.

Where form is concerned, pragmatic markers tend to occur in sentence-marginal positions or parenthetically, as can be seen in the initial position the adverb occupies in example (3). However, many other pragmatic markers have become fused with the rest of the sentence and are therefore to be considered on the boundary between proper adverbs and discourse markers. In English, this is the case of units such as really, actually, of course, etc. This flexibility with respect to position is only apparent in some cases. I will claim in this paper that pragmatic markers tend to form collocations and patterns in text, precisely because of their textual function (see section 4.3. below).

From a functional perspective, modal adverbs with epistemic meanings contribute to the expression of interpersonal meanings, i.e., they represent the speaker’s stance with respect to the proposition. However, languages change over time and many modal adverbs are noted to have developed textual functions linking together the different parts of a text at a suprasentential level. In these cases, adverbs such as well, really, of course and others lose part of their epistemic force and should rather be treated as pragmatic markers. This shift in meaning of some epistemic adverbs has been explained as a process of grammaticalization of these linguistic units as discourse markers or pragmatic markers. This is a very subtle evolution in some cases, but clearly noticeable in others, although it affects different types of particles, in particular adverbs, in different degrees. The analysis of the translations of a number of epistemic adverbs in this paper will enable us to discriminate the various types of semantic functions that may be performed by these linguistic units by identifying translation trends.

The adverb certainly indicates strong agreement (or disagreement, if in a negative environment) and is therefore considered to express a high degree of likelihood. If this adverb is used, it is because the addressee may have some doubts that need to be clarified by the use of certainly, which is employed with a clearly emphatic sense. In our parallel corpus there were 91 cases of the English adverb certainly. Two of these instances were modifiers, not adverbials, and were not included in this study. Only in 14 cases did the adverb co-occur with a modal verb: in 7 cases with will, in 5 cases with would, in 1 case with could and in 1 case with can. In all of these cases, the combination was “modally harmonic” (Lyons 1977), i.e., the adverb was used to reinforce the meaning conveyed by the modal verb.

Table 2 below shows the translations found for the English adverb certainly in the P-ACTRES corpus:

The data show that over 25% of cases of the English modal adverb certainly were translated into Spanish by the prepositional phrase (PP) desde luego, as in examples (4) and (5):

This PP is highly idiomatic in Spanish for expressing a strong commitment on the part of the speaker with respect to his/her belief of the truth of the proposition. In addition, a number of other PPs with similar meanings were used with the same function, notably sin duda (10%), con certeza (5%), con seguridad (2%), sin lugar a dudas (1%), de ningún modo (1%), por supuesto (1%), por cierto (1%). In these cases, the noun in the PP is the one that carries the epistemic force, be it in the negative as in sin duda (“with no doubt”), or in the positive as in con certeza (“with certainty”) or con seguridad (“with sureness”). Altogether, in our corpus PPs account for 50% of the translational options of the English adverb certainly. Examples (6) and (7) illustrate the uses of two different PPs in Spanish:

This wide range of options in the form of a PP ties in with the well-know typological nature of the Spanish language, which tends towards an analytic structure and therefore makes heavy use of PPs for many types of adverbial meanings to the detriment of the corresponding single-word adverbs typically ending in the suffix –mente (Hoye 1997: 253). The cognate of the English form certainly, the Spanish adverb ciertamente, does appear in the list and accounts for less than 10% of the translations, as in example (8).

Three additional variants of adverbs in –mente also occur, but with just one case each: evidentemente, seguramente and verdaderamente.

Modal adverbs ending in –mente do not seem to be frequent options of translators when dealing with the English adverb certainly and we can therefore infer that they are not particularly popular among native speakers of Spanish for expressing the meanings related to evidentiality and commitment on the part of the speaker. All in all, only 12% of the translational options correspond to one of these modal adverbs in Spanish. While –ly adverbs are rather neutral style in English, Spanish adverbs ending in –mente carry strong stylistic and pragmatic connotations. As a group, they are considered rather formal and tend to be avoided by native speakers because of their length, which makes them cumbersome and less flexible than PPs, which are used much more readily to express adverbial meanings (Hoye 1997; Rabadán et al. 2006). An overuse of –mente adverbs in translated Spanish may actually be considered an indicator of translationese, i.e., instances in translated language where the influence of the source language can be detected (Mauranen 2000, 2004). Comparing data from original Spanish (CREA) and the Spanish translations in P-ACTRES, ciertamente occurs slightly more often in translated texts than in original Spanish texts: 18 cases per million words in original texts versus 26 per million words in translations. However, for this particular adverb the difference is not statistically significant (p-value = 0.096)[4].

The second option most commonly taken by Spanish translators to convey the meaning of certainly was sheer omission, with nearly 20% of cases. This is probably the most striking fact in the analysis of this particular adverb. The following are some examples:

In these cases, it seems that the translators have assumed that the degree of certainty implied by this particular adverb is virtually similar to absolute certainty, and he/she therefore uses the indicative mood and no sign of epistemic modality whatsoever in the Spanish translation. This omission points towards the idea that certainty is not a type of modality at all, but rather one of the ends of the scale of factuality. Epistemic modality is a cline going from what is extremely likely to what is extremely unlikely from the point of view of the speaker, and certainty implies that there is no room for doubt. If the speakers in these cases are certain of the truth of the proposition, we could ask ourselves why the modal adverb certainly is being used at all. The indicative mood would be the unmarked choice for expressing factuality, reality, and absolute truth. The use of an additional epistemic adverb seems somewhat superfluous and even confusing. The answer to this question seems to be that speakers of English only use the adverb certainly when they are not really certain. I claim that in the cases where certainly is omitted in the translations it does not act as an epistemic adverb, but rather carries out a different function related to the textual organization of discourse. In these cases, certainly is functioning as a pragmatic marker expressing stance only marginally and mainly being used as a link between two different parts of the text. For example, in (10) above I consider that the adverb is used to emphasize the enumeration and is, therefore, being employed here with a clearly additive meaning, on top of reinforcing the speaker’s stance and commitment with respect to the action of the verb. Similarly, in (11) the adverb certainly suggests that there is some type of contrast involved between a previous part of the discourse about which the speaker is not so certain.

Why are these cases systematically omitted in translations? The reasons may be varied: a) pragmatic discourse markers are often difficult to convey cross-linguistically because a slightly different effect may be achieved; b) on the other hand, the lack of pragmatic markers is an indicator of simplification, a feature typically identified as characterizing translated language in general (Baker 1998).

One case of omission deserves a separate comment:

Again, in (12) the adverb indicates not only a high degree of likelihood, but also functions as a pragmatic link reinforcing the relationship between the condition and the probable result if the condition is fulfilled. Here, the Spanish translator has opted for the use of the lexical verb poder in the past tense, which is used to express only a very low degree of likelihood. No additional epistemic or pragmatic particle appears in the Spanish translation and this results in a considerable downtoning of the epistemic force.

In about 7% of cases the English adverb certainly was modulated into a Spanish construction including an adjective that expresses a high degree of likelihood: seguro, claro, evidente. These constructions all follow a common structural pattern: copula + adjective + que, as in (13) and (14):

The P-ACTRES corpus included a few cases (about 7% of the total) where the meaning of the English adverb certainly was modulated into constructions in Spanish including the superlatives más and menos, for the positive and negative sense, respectively, marking the target text with highly idiomatic and somewhat colloquial expressions as in (15) and (16):

Another 4% of the translation solutions corresponded to cases where the degree of certainty was conveyed by means of a modulation into a clause containing the Spanish noun duda, “doubt” as in (17) and (18):

Finally, regarding the use of a lexical verb to convey the epistemic force of the English adverb certainly, there was just one case involving the Spanish verb esperar followed by the subjunctive mood. This construction is one typical way for expressing doubt about a future event, and therefore epistemic modality, in Spanish, although it does not seem a popular translation solution in the case of the adverb certainly, according to our parallel corpus as it accounts for only 1% of cases.

Figure 1 below shows the various translation possibilities found for the adverb certainly in the P-ACTRES corpus.

This brief analysis of the Spanish translations of the adverb certainly has revealed that PPs are clearly the preferred option for expressing the meanings conveyed by this particular adverb when expressing a degree of likelihood. Other resources such as adverbs ending in –mente, adjectival, clausal or lexical constructions occur much less frequently. The most interesting conclusion that can be derived from this analysis, though, is the fact that nearly 20% of cases of this adverb tend to be omitted in Spanish translations when there is an added pragmatic effect on top of the epistemic meaning. It appears that this omission does not generally affect the overall meaning of the translation, but simply results in a text that is less tightly woven together with textual and pragmatic links than the original.

The adverb probably is one of the most common epistemic adverbs in English (Biber et al. 1999: 563). From a semantic perspective, the use of probably implies that

…the speaker’s reasonable certainty that the propositional content of what he is saying is in fact true. […] Probably is the most common item and the strength of likelihood it conveys is best understood if it is contrasted with the much weaker adverbs conceivably and possibly.

Hoye 1997: 194

The P-ACTRES corpus contained 170 cases of probably, but 12 of these cases were not taken into account in the study as they corresponded to modifying uses in adjective phrases. In addition, another 2 cases had to be left out because the sentences in which the adverb occurred were entirely omitted in the target text. This leaves us with 156 instances of probably and their corresponding translations into Spanish. With respect to co-occurrence with modal verbs, probably appeared particularly often with the modals will (16 cases) and would (15 instances), and there was one case of co-occurrence with can and another one with should.

The translational options found in the parallel corpus for the English adverb probably are the following:

The table shows that over half the occurrences of probably (55% of cases) have been translated by the cognate Spanish adverb probablemente as in (19) and (20):

This indicates that Spanish translators rely very heavily on this particular adverb for the expression of the intermediate degree of likelihood expressed by the English adverb probably. However, as mentioned in the analysis of certainly, –mente adverbs carry a strong stylistic connotation and generally tend to be avoided, although usage varies a lot from one adverb to another. Intralinguistic comparable data have shown that there is a significant overuse of probablemente in translated texts with respect to texts written originally in Spanish: 53 instances per million words in original Spanish texts versus 176 instances per million words in translated Spanish. This appears to be a clear case of translationese, since the difference is statistically significant (p-value = 0). A similar feature has been found for this particular English adverb in Swedish translations (Aijmer 2001).

The next most common translational option (16.66% of cases) has been another adverb ending in –mente with a very similar meaning, although slightly more colloquial: seguramente, as in (21). It may be claimed that the degree of certainty implied by this adverb is somewhat higher than the one implied by the English adverb probably.

Another two adverbs ending in –mente have been found as translation equivalents of probably: posiblemente and presumiblemente, with one case each. Here, the degree of likelihood appears to be somewhat lower than the one expressed in the English source text. This variation indicates the difficulty in conveying in a different language a semantic nuance that is so elusive as a particular degree of probability. The boundaries between semantic categories are always fuzzy, and probability can be said to be a cline rather than a clearly defined separate grouping. However, the relationship between meaning and the form used to expressed it is something that is quantifiable to a certain extent and in this case the analysis has shown that, all in all, over 73% of the translational options for this particular adverb correspond to –mente adverbs.

The next most frequently used resource has been the modulation of the semantic content of probably into a construction including a copula followed by an adjective indicating the degree of probability – probable (12 cases), posible (2 cases), previsible (1 case) – and followed by the particle “que” plus the verb in the subjunctive mood, as in (22) and (23) below:

The subjunctive mood is a very common resource employed by native speakers of Spanish to indicate non-factuality, futurity and various degrees of probability and possibility. The modal meaning expressed exclusively by adverbial means in the English text is thus divided up into two separate parts in Spanish: a lexical part in the form of an adjective, and a further reinforcement of the uncertainty in the form of verbal mood.

A further 6 cases of probably were translated into a variant of this adjectival resource, but in the form of a highly idiomatic superlative construction in Spanish including the adjective probable and the subjunctive mood, as in (24) and (25):

Here the epistemic meaning is conveyed by the grammatical construction as a whole, which follows a fixed order. Again, as in the case of the adverb seguramente, I consider that there is an increase in the degree of certainty conveyed in the Spanish translation with respect to the English original. Superlatives are used here for epistemic modality, as is the case of certainly.

There were 7 cases in our corpus (4.48% of the total) where probably was translated by a lexical verb or a verbal periphrasis indicating probability: deber de, suponer, creer and poder, as in examples (26) and (27):

In this last example, the degree of likelihood is expressed in the lexical verb creer followed by the subjunctive mood in the past tense reinforcing the attitude of the speaker with regard to the proposition and referring to the future time. In addition, the form of the original is greatly modulated with a change in polarity, from positive in the original to negative in the translation.

In 6 cases, the adverb probably was omitted in the Spanish translation, as in (28):

Considering the size of our sample (156 instances), this is an indicator of the fact that this particular adverb does not tend to be omitted very often (about 4% of cases), its meaning being considered of importance for the transfer into the target language. In (28) it may be noted that the source text includes the lexical verb suppose, which indicates epistemic modality itself. The cognate lexical verb is used in Spanish, supongo, so the reinforcing modal meaning of probably is left out but the epistemic meaning is preserved to a certain extent in the form of the lexical verb. Again, this could be considered a downtoning of the epistemic force present in the English example. In none of these cases can probably be considered to have any pragmatic effect whatsoever on the meaning of the source text and the few omissions there are occur when other indicators of epistemic meanings are present and attenuate these omissions.

There were 4 cases of other epistemic adverbs not ending in –mente that have been chosen to express probability: tal vez and quizá(s) (2.56% of the total), as in examples (29) and (30):

Finally, the translational options for the English adverb probably included 4 instances of PPs, which indicates that this is not a particularly popular resource in Spanish for expressing the type of meanings conveyed by probably. Again the head word in the NP that follows the preposition is the one containing the epistemic force, as in examples (31) and (32):

The latter case again tends towards a higher degree of certainty in the translation than the one implied by the original English text, a trend that has already been noted in many other cases of the translation of this particular adverb. In fact, the PP sin duda was one of the options taken in the case of the English adverb certainly, which indicates a higher degree of likelihood than probably.

Figure 3 below summarises graphically the translational options found in the P-ACTRES corpus for the adverb probably.

The vast majority of cases (73%) have been translated using –mente adverbs in Spanish, predominantly probablemente, followed by seguramente, with two additional instances of one single adverb each. Indeed, there is a statistically significant overuse of probablemente in Spanish translations. Adjectival constructions of various types account for about 13% of cases and the remaining translational options, although highly idiomatic in Spanish, account for less than 5% of cases each: lexical verbs, PPs and other adverbs not ending in –mente. It is striking that the adverb probably is omitted only in 4% of cases in Spanish translations, which seems to indicate that the meaning of epistemic modality conveyed by this adverb is considered as clearly relevant and important in the original and is therefore preserved in the translations. This is supported by the fact that the omissions occurred where other modal resources supplied part of the lost epistemic meaning (28). This result seems to indicate that probably is not an adverb that is likely to grammaticalize into a pragmatic marker in English.

The English adverb possibly is considered to express a lower degree of likelihood on the part of the speaker than the one conveyed by certainly and probably. The epistemic meaning conveyed by this adverb expresses a weak commitment on the part of the speaker with respect to the truth of the proposition. Quirk et al. (1985: 600) define the use of possibly as a downtoner or minimizer, on top of the expression of doubt. Biber et al. (1999: 972), consider that many stance markers are difficult to classify in one singular meaning group and consider that possibly may indicate not only doubt, but also precision or limitation.

The P-ACTRES corpus contained 43 instances of the English adverb possibly with their corresponding Spanish translations, but in 13 of these cases possibly was not functioning as a clause adverbial but as a modifier in an adjective phrase, and these cases were not considered for the study. Thus, 30 cases were finally analysed. Nearly half the occurrences showed possibly co-occurring with modal verbs: in 10 cases with could and in 4 cases with can. Both are pleonastic combinations, i.e., harmonic verb-adverb combinations. This high frequency of co-occurrence with modal verbs when compared with the frequency of co-occurrence of the other two adverbs analyzed previously seems to indicate that the semantic role fulfilled by possibly is not always considered to be strong enough to convey epistemic meanings on its own, but is rather used as a reinforcement of some epistemic modal verbs. Consequently, possibly seems to have an emphatic force and is, therefore, a good candidate for grammaticalization.

The translations found in Spanish for the adverb possibly are shown in Table 4 below:

The data show that only about half of the occurrences of possibly (17 instances) have actually been translated into Spanish indicating epistemic modality. In almost half the cases (13 instances) the reinforcing status of the English modal adverb is completely lost in the Spanish translation. There were 2 cases of total omission of the adverb and the epistemic force involved in the original, as in (33):

The meaning conveyed by the English adverb possibly cannot be considered to indicate only a low degree of likelihood, but rather a strong emphasis and personal involvement on the part of the speaker. It may even be argued that the particular combination found in the example closely resembles a collocation or at least a highly idiomatic expression in English often associated with exclamations and spoken conversation. The modal is fused together with a modally harmonic adverb forming something that resembles a single lexical unit used with an emphasizing purpose. Possibly is being used here mainly as a pragmatic marker. The adverb is omitted in the translation and the total omission of the semantic content of possibly is more idiomatic in Spanish than its translation would have been, i.e., less is more in this case. I will use the term “idiomatic omission” to refer to this phenomenon.

The most interesting fact, however, is that in 11 cases (36.66% of the total) the content of the adverb is omitted but the epistemic force indicating possibility is still present in the translation, although greatly diminished, in the form of the Spanish lexical verb poder in various tenses. Again, in all of these cases, the English adverb possibly was used pleonastically to support a modal verb, can or could, as in example (34):

In my opinion, these cases should not be considered simply as omissions, but rather as idiomatic omissions. Even though the exact semantic content of the adverb is not present in the Spanish texts, I claim that this is not only an appropriate translational solution, but actually the best option, since the Spanish verb poder does convey sufficient epistemic force and does not collocate with any Spanish adverb in the way can/could collocate with possibly to convey an emphatic meaning. Omitting the content of the adverb possibly results in a higher degree of idiomaticity in Spanish, as can be seen in examples (35) and (36):

I claim that these omissions in the translations are a clear sign of the use of the English adverb possibly as a discourse marker in the corresponding source texts. In these cases the adverb is void of actual epistemic meaning of its own, but is used merely as an intensifier indicating great surprise on the part of the speaker and complementing the meaning of the modal verb it reinforces. This semantic shift is most clearly seen in spoken conversation or dialogue, as in examples (34) and (36).

The idiomatic omission was the most common translational option adopted by the Spanish translators, and this option corresponded to cases where the adverb possibly was used as an emphasizer rather than as epistemic adverb. This implies that, at least in our sample, which is admittedly limited, nearly half the instances of possibly were of this discourse marker type. Is this a sign of the grammaticalization of the English adverb, as found by Aijmer (2002)? I argue that the idiomatic omission in translations of certain traditionally so-called modal adverbs can be used as an indicator of the shift these adverbs have undergone in meaning and of their newly acquired uses as pragmatic and textual markers.

In the cases where possibly was not omitted, the translational options observed closely resemble the ones found in the two previous cases. About 20% of cases corresponded to the cognate Spanish adverb posiblemente, marking, in addition to a low degree of commitment, some type of limitation, as in example (37):

As in the two previous cases, I have carried out a comparative study between the frequency of this adverb in original Spanish texts and in translations. Again, I found an overuse of possibly, which occurs 26 times per million words in original texts (CREA) and 40 times per million words in translated texts (P-ACTRES). However, this difference is not statistically significant, with a p-value of 0.015.

There were 4 cases where the meaning of the English adverb possibly was modulated into an adjectival construction in Spanish using the corresponding adjective, as in example (38), where the translation reflects the epistemic meaning but not the pragmatic mark of the English adverb:

In 4 cases, the adverb possibly was translated by single-word adverbs that were not the cognate adverb in –mente, all of them clearly indicating possibility: acaso, quizá(s), tal vez. In all cases the adverbs occurred with the subjunctive mood in Spanish, thus indicating a combination of two linguistic resources for expressing the epistemic meaning: a lexical one (the adverb) and a morphological one (mood). Example (39) illustrates this option:

There were 2 cases of a translation into the PP en modo alguno, which is rather formal in Spanish. In both cases the English adverb co-occurred with the modal could and was used as a modal reinforcement. The emphasizing effect is maintained in Spanish PP together with the lexical verb poder in the Spanish translations to reinforce the modal meaning, as in example (40):

Finally, the English adverb possibly was modulated into the Spanish noun posibilidad in one case, example (41):

Figure 3 below summarizes all the translational options found in the P-ACTRES parallel corpus for the English adverb possibly:

Even though the number of cases that have been analysed is relatively small, the distribution of the translational options can be considered sufficiently representative for our purposes here. Figure 3 shows that in nearly half of the cases (44%) the English adverb possibly is omitted in Spanish translations. When the adverb is translated, the translational options are similar to the ones found for the adverbs analysed in the previous sections: 20% of the instances were translated with adverbs ending in –mente, 13% of cases were translated using adjectival constructions and the same percentage corresponds to translations by means of other adverbs not ending in –mente. Finally, PPs and transposition into nouns show percentages under 10%.