cross-linguistic semantic prosody : the case of ‘ commit ’ , ‘ signs of ’ and ‘ utterly ’ and their norwegian correspondences

This paper reports on a contrastive study of semantic prosody in English and Norwegian. Semantic prosody refers to the communicative function of extended units of meaning (Sinclair 1996; Stubbs 2013), i.e. “the semantic prosody of an item is the reason why it is chosen, over and above the semantic preferences that also characterise it” (Sinclair 1998, 20), where an item is equated with the sequence of words constituting an extended unit of meaning. The paper presents three case studies of English units with an established negative prosody containing the core items commit, signs of and utterly. The Norwegian correspondences of these items are identified on the basis of a bidirectional translation corpus, viz. the English-Norwegian Parallel Corpus. These correspondences serve as the starting point for an investigation of cross-linguistic prosodies. It is shown that while units with commit and signs of have good Norwegian matches in terms of semantic prosody, units with utterly are less stable across the two languages, underlining the importance of carrying out studies of this kind in order to improve the crosslinguistic understanding of extended units of meaning. This in turn has implications for how teachers, translators and lexicographers choose to present words in isolation or as part of larger, extended units. [1] introduct ion As pointed out by Stubbs (1995, 25) “it is becoming increasingly well documented that words may habitually collocate with other words from a definable semantic set”. Such collocations have in turn been shown to contribute towards a certain communicative purpose of a unit of meaning, i.e. its semantic prosody. This phenomenon has received some attention in English corpus linguistics over the past two decades, and several scholars have identified such prosodies for a number of items. One example is the phrasal verb set in, which primarily is seen to co-occur with subjects referring to unpleasant states of affairs, including rot, decay, and despair (Sinclair 1991, 70ff.), giving rise to a negative semantic prosody. Through a [162] signe oksefjell ebeling process of co-selection, items such as set in are said to formmulti-word units that constitute a single choice (Sinclair 1991, 111; Tognini-Bonelli 2002, 91). In a previous cross-linguistic study of an English unit with an established negative prosody – a unit including cause – and its Norwegian correspondences, it was revealed that the favoured Norwegian translation of the verb cause: få (x til å) shows a tendency towards a neutral prosody, rather than a negative prosody corresponding to that of cause (Ebeling 2013). In this paper I will use the extended fiction part of the English-Norwegian Parallel Corpus (ENPC+, henceforth)1 to further explore semantic prosodies in a contrastive perspective, taking three English items asmy starting point, which all have been said to take on negative prosodies: commit, signs of, and utterly. By studying their Norwegian correspondences I hope to gain more insight into how units with a specific semantic prosody in one language manifest themselves in another language. Do they tend to diverge as in the case of cause and its main translation correspondence in Norwegian or are they found to bemore stable in the sense that they are imbuedwith similar prosodies?2 Although I will focus mainly on correspondences found in Norwegian translations, reference will also be made to Norwegian sources that give rise to commit, signs of, and utterly. [2] background [2.1] Semantic prosody In Sinclair’s (e.g. Sinclair 1991, 1996) model of meaning (also known as his model of extended units of meaning) it is postulated that words that habitually co-occur constitute a single choice with a certain communicative purpose. As noted above such a communicative purpose, or semantic prosody, has been identified for a number of units in English, including the items under study here, viz. commit, signs of, and utterly, all of which have been said to be the core of units with a negative, or unfavourable, semantic prosody. Louw (1993, 157) defines semantic prosody as “a consistent aura of meaning with which a form is imbued by its collocates”, while Partington (1998, 68) says that semantic prosody refers to the spreading of connotational colouring beyond single word boundaries. For instance, in the case of commit, Partington (1998, 67) claims that “the unfavourable connotation can be seen to reside not simply in the word commit but over a unit consisting of commit and its collocate (offences, serious crime, foul etc.)” as illustrated in the concordance lines in figure 1 on the facing page. In other words it is the cotext of a core that determines the semantic prosody of a unit of meaning, and following Stubbs (2013, 25), underlying the choice of [1] See Section [3] and Ebeling & Ebeling (2013, 86f) for a description of the ENPC+. [2] Although it is the whole extended unit of meaning that has a semantic prosody, I will sometimes use the core item – viz. commit, signs of, utterly – as shorthand for the whole unit of meaning. OSLa volume 6(1), 2014 cross-linguistic semantic prosody [163] figure 1: Partington (1998, 67) cotext is the speaker’s motivation for uttering something; “semantic prosody has to do with why it is expressed” (ibid.). In recent years, linguists have discussed, and indeed questioned, the validity of the whole concept of semantic prosody; see for instance Whitsitt (2005); Hunston (2007); Morley & Partington (2009); Stewart (2010). Some of this criticism may be unwarranted; as pointed out by Stubbs (2013, 25): Stewart (2010), in his book on Semantic Prosody, is very sceptical of the concept, but, I think, makes the error of discussing semantic prosody independently of the model [of extended units of meaning] of which it is only one parameter. In the present paper it will be assumed, with e.g. Sinclair (1987, 1991, 1996) and Stubbs (1995, 2009), that the concept of semantic prosody is valid and that it is one of the obligatory elements of an extended unit of meaning. Taking this as our underlying assumption, we also infer that semantic prosody plays a crucial role in discourse, as it is said to represent the communicative, or discourse, function of a linguistic choice (Sinclair 1991; Stubbs 2013). According to the extended-unit-of-meaningmodel, as outlined by Sinclair (e.g. Sinclair 1996) and further developed by Stubbs (e.g. Stubbs 2013), in addition to an obligatory core, it operates with four structural categories: • Collocation (words) • Colligation (grammar) • Semantic preference (topic) • Semantic prosody (speech act / communicative purpose) In the case of commit, this can be visualised in the way shown in Table 1 on the next page. OSLa volume 6(1), 2014 [164] signe oksefjell ebeling table 1: Extended unit of meaning with commit as core (Core: commit) Collocation commit (the/a/an/Ø) murder/suicide, etc. Colligation commit NP Semantic preference word or phrase to do with an unpleasant event Semantic prosody bad/negative By collocation is meant the actual words that are found in the vicinity of the core, while colligation describes the grammatical category that is found in the immediate co-text of the core; the grammatical category is determined on the basis of themost frequent collocations; the typical colligation of commit is thus an NP. Similarly, the semantic preference of the unit is decided on the basis of shared semantic content of the collocations/colligations, i.e. “a definable semantic set” (Stubbs 1995, 25), i.e. “semantic preference has to do with what is expressed (the topic)” (Stubbs 2013, 25). In the example above, the nouns and noun phrases in the co-text of commit commonly depict an unpleasant event, e.g. murder, suicide, deception, crime. Finally, all of the above elements form part of an extended unit of meaning with a decidedly negative semantic prosody. “[T]he semantic prosody of an item is the reason why it is chosen, over and above the semantic preferences that also characterise it” (Sinclair 1998, 20), where an item is equated with the sequence of words constituting an extended unit of meaning. In this context it should be pointed out that this bottom-up approach to phraseological analysis is actually the opposite of what happens in a real-life language-producing situation. In a real life setting, the (choice of) semantic prosody is primary, in so far as a person, at a specific point in a narrative, will seek to express that something, i.e. whatever is being talked about or discussed, is difficult. (Ebeling & Ebeling 2013, 58) An analysis, as outlined above, of the four structural categories accompanying the core in an extended unit of meaning will be performed on the cross-linguistic data identified for the case studies (see further Section [3] on themethod applied). [2.2] Previous cross-linguistic research Although a few cross-linguistic studies of semantic prosody have emerged over the last few years, comparing a range of languages mainly on the basis of comparable data,3 this section will focus only on findings from a previous study of En[3] E.g. Berber Sardinha (1999, 2000); Tognini-Bonelli (2001, 2002); Partington (1998); Xiao&McEnery (2006); Dam-Jensen & Zethsen (2006). See further Ebeling (2013) for a discussion of similarities and differences between these studies. OSLa volume 6(1), 2014 cross-linguistic semantic prosody [165] glish and Norwegian (Ebeling 2013), using a parallel corpus, including both translational and comparable data (see Section [3] for a description of the corpus). Taking another English core of an extended unit of meaning with an established negative semantic prosody as its starting point (see Stubbs 1995; Ebeling 2013) explored the noun and verb uses of cause in a contrastive perspective. All instances of cause in the English-Norwegian Parallel Corpus were analysed in order to determine its semantic prosody and lexicogrammatical patterns. In each case the Norwegian

cross-linguistic semantic prosody: the case of 'commit', 'signs of' and 'utterly' and their norwegian correspondences University of Oslo a b s t r a c t This paper reports on a contrastive study of semantic prosody in English and Norwegian.Semantic prosody refers to the communicative function of extended units of meaning (Sinclair 1996;Stubbs 2013), i.e. "the semantic prosody of an item is the reason why it is chosen, over and above the semantic preferences that also characterise it" (Sinclair 1998, 20), where an item is equated with the sequence of words constituting an extended unit of meaning.
The paper presents three case studies of English units with an established negative prosody containing the core items commit, signs of and utterly.The Norwegian correspondences of these items are identified on the basis of a bidirectional translation corpus, viz. the English-Norwegian Parallel Corpus.These correspondences serve as the starting point for an investigation of cross-linguistic prosodies.It is shown that while units with commit and signs of have good Norwegian matches in terms of semantic prosody, units with utterly are less stable across the two languages, underlining the importance of carrying out studies of this kind in order to improve the crosslinguistic understanding of extended units of meaning.This in turn has implications for how teachers, translators and lexicographers choose to present words in isolation or as part of larger, extended units.
[1] i n t r o d u c t i o n As pointed out by Stubbs (1995, 25) "it is becoming increasingly well documented that words may habitually collocate with other words from a definable semantic set".Such collocations have in turn been shown to contribute towards a certain communicative purpose of a unit of meaning, i.e. its semantic prosody.This phenomenon has received some attention in English corpus linguistics over the past two decades, and several scholars have identified such prosodies for a number of items.One example is the phrasal verb set in, which primarily is seen to co-occur with subjects referring to unpleasant states of affairs, including rot, decay, and despair (Sinclair 1991, 70ff.), giving rise to a negative semantic prosody.Through a process of co-selection, items such as set in are said to form multi-word units that constitute a single choice (Sinclair 1991, 111;Tognini-Bonelli 2002, 91).In a previous cross-linguistic study of an English unit with an established negative prosody -a unit including cause -and its Norwegian correspondences, it was revealed that the favoured Norwegian translation of the verb cause: få (x til å) shows a tendency towards a neutral prosody, rather than a negative prosody corresponding to that of cause (Ebeling 2013).In this paper I will use the extended fiction part of the English-Norwegian Parallel Corpus (ENPC+, henceforth) 1 to further explore semantic prosodies in a contrastive perspective, taking three English items as my starting point, which all have been said to take on negative prosodies: commit, signs of, and utterly.By studying their Norwegian correspondences I hope to gain more insight into how units with a specific semantic prosody in one language manifest themselves in another language.Do they tend to diverge as in the case of cause and its main translation correspondence in Norwegian or are they found to be more stable in the sense that they are imbued with similar prosodies? 2 Although I will focus mainly on correspondences found in Norwegian translations, reference will also be made to Norwegian sources that give rise to commit, signs of, and utterly.
[2] b a c k g r o u n d [2.1] Semantic prosody In Sinclair's (e.g. Sinclair 1991, 1996) model of meaning (also known as his model of extended units of meaning) it is postulated that words that habitually co-occur constitute a single choice with a certain communicative purpose.As noted above such a communicative purpose, or semantic prosody, has been identified for a number of units in English, including the items under study here, viz.commit, signs of, and utterly, all of which have been said to be the core of units with a negative, or unfavourable, semantic prosody.Louw (1993, 157) defines semantic prosody as "a consistent aura of meaning with which a form is imbued by its collocates", while Partington (1998, 68) says that semantic prosody refers to the spreading of connotational colouring beyond single word boundaries.For instance, in the case of commit, Partington (1998, 67) claims that "the unfavourable connotation can be seen to reside not simply in the word commit but over a unit consisting of commit and its collocate (offences, serious crime, foul etc.)" as illustrated in the concordance lines in figure 1 on the facing page.
In other words it is the cotext of a core that determines the semantic prosody of a unit of meaning, and following Stubbs (2013, 25), underlying the choice of [1] See Section [3] and Ebeling & Ebeling (2013, 86f) for a description of the ENPC+.[2] Although it is the whole extended unit of meaning that has a semantic prosody, I will sometimes use the core item -viz.commit, signs of, utterly -as shorthand for the whole unit of meaning.
OSLa volume 6(1), 2014 cross-linguistic semantic prosody [163] figure 1: Partington (1998, 67) cotext is the speaker's motivation for uttering something; "semantic prosody has to do with why it is expressed" (ibid.).In recent years, linguists have discussed, and indeed questioned, the validity of the whole concept of semantic prosody; see for instance Whitsitt (2005); Hunston (2007); Morley & Partington (2009); Stewart (2010).Some of this criticism may be unwarranted; as pointed out by Stubbs (2013, 25): Stewart (2010), in his book on Semantic Prosody, is very sceptical of the concept, but, I think, makes the error of discussing semantic prosody independently of the model [of extended units of meaning] of which it is only one parameter.
In the present paper it will be assumed, with e.g.Sinclair (1987Sinclair ( , 1991Sinclair ( , 1996) ) and Stubbs (1995Stubbs ( , 2009)), that the concept of semantic prosody is valid and that it is one of the obligatory elements of an extended unit of meaning.Taking this as our underlying assumption, we also infer that semantic prosody plays a crucial role in discourse, as it is said to represent the communicative, or discourse, function of a linguistic choice (Sinclair 1991;Stubbs 2013).
According to the extended-unit-of-meaning model, as outlined by Sinclair (e.g.Sinclair 1996) and further developed by Stubbs (e.g. Stubbs 2013), in addition to an obligatory core, it operates with four structural categories: • Collocation (words) • Colligation (grammar) • Semantic preference (topic) • Semantic prosody (speech act / communicative purpose) In the case of commit, this can be visualised in the way shown in By collocation is meant the actual words that are found in the vicinity of the core, while colligation describes the grammatical category that is found in the immediate co-text of the core; the grammatical category is determined on the basis of the most frequent collocations; the typical colligation of commit is thus an NP.Similarly, the semantic preference of the unit is decided on the basis of shared semantic content of the collocations/colligations, i.e. "a definable semantic set" (Stubbs 1995, 25), i.e. "semantic preference has to do with what is expressed (the topic)" (Stubbs 2013, 25).In the example above, the nouns and noun phrases in the co-text of commit commonly depict an unpleasant event, e.g.murder, suicide, deception, crime.Finally, all of the above elements form part of an extended unit of meaning with a decidedly negative semantic prosody."[T]he semantic prosody of an item is the reason why it is chosen, over and above the semantic preferences that also characterise it" (Sinclair 1998, 20), where an item is equated with the sequence of words constituting an extended unit of meaning.In this context it should be pointed out that this bottom-up approach to phraseological analysis is actually the opposite of what happens in a real-life language-producing situation.In a real life setting, the (choice of) semantic prosody is primary, in so far as a person, at a specific point in a narrative, will seek to express that something, i.e. whatever is being talked about or discussed, is difficult.(Ebeling & Ebeling 2013, 58) An analysis, as outlined above, of the four structural categories accompanying the core in an extended unit of meaning will be performed on the cross-linguistic data identified for the case studies (see further Section [3] on the method applied).

[2.2] Previous cross-linguistic research
Although a few cross-linguistic studies of semantic prosody have emerged over the last few years, comparing a range of languages mainly on the basis of comparable data, 3 this section will focus only on findings from a previous study of  E.g.Berber Sardinha (1999,2000); Tognini-Bonelli (2001, 2002); Partington (1998); Xiao & McEnery (2006); Dam-Jensen & Zethsen (2006).See further Ebeling (2013) for a discussion of similarities and differences between these studies.
Taking another English core of an extended unit of meaning with an established negative semantic prosody as its starting point (see Stubbs 1995;Ebeling 2013) explored the noun and verb uses of cause in a contrastive perspective.All instances of cause in the English-Norwegian Parallel Corpus were analysed in order to determine its semantic prosody and lexicogrammatical patterns.In each case the Norwegian translations were recorded, serving as translational mirrors in a similar analysis going from the Norwegian source texts into English translations.In this way it was possible to establish to what extent the most commonly used Norwegian correspondences (translations and sources) share the negative semantic prosody of cause.
The investigation revealed that there is no Norwegian correspondence that matches cause in terms of negative semantic prosody.For instance, the most commonly used verb translation of få (x til å) ('get (x to)') is typically used in neutral contexts in original texts, as in example (1).Although the third-most common verb correspondence føre til 'lead to' has a preference for negative contexts (see example ( 2)), it is not used in such environments to the same extent as cause.Furthermore, føre til is most commonly translated into lead to and not cause, suggesting that føre til and cause have different semantic prosodies. (1) En lyd fra lønnetreet fikk henne til å se opp.(THA1) A sound from the maple tree made her look up.
(2) Det var ilden og blåsingen som førte til at de så så stygge ut.(KAL1) It was the fire and all the glassblowing that made them look so ugly.
With regard to the Norwegian correspondences of the noun use of cause, it was concluded that "[t]he closest Norwegian equivalent of cause in terms of semantic prosody is årsak 'cause'; årsak is also the most commonly used correspondence in the translations" (Ebeling 2013, Section 4.2), as shown in example (3).
(3) In the top right-hand corner was the legend "Opposite Hotel Wolcott", whose bar was the cause of so much of the trouble.(RF1) Øverst i høyre hjørne sto det "Vis à vis Hotel Wolcott", der den baren lå som var årsak til så mange vanskeligheter.
Moreover, grunn 'reason', which was the second-most used translational correspondence of the noun cause, was shown to be predominantly neutral, but was also found in negative contexts.Worth noticing is that "in 31 of the 87 instances with a negative bias, grunn has been translated into reason (i.e.36%) and only 3 times into cause.The same tendency is seen for årsak (most commonly translated into reason)" (ibid.).
As to potential cross-linguistic implications, some observations were made, suggesting that even if the same degree of negative prosody is not present in the main correspondences of cause, data from the Norwegian original texts show that negative environments are not unnatural for e.g.føre til or årsak.Thus, it can be argued that the prosody is naturally preserved in translation.However, "it is important that translators are made aware of the more neutral use of e.g.få (x til å) and grunn as compared to cause (x to) and cause" (ibid.).This follows from what Dam-Jensen & Zethsen (2006, 1620) note for Danish medføre 'lead to' where the negative implications "are not as devastating as are those of forårsage", or indeed as those of cause.
Choosing a word or expression with a mismatched prosody in translation may also have further implications, as pointed out by Berber Sardinha (2000, 96): [...] the shift in semantic prosody quality may be unintentional.When this happens in translation, it gives rise to a different connotation to that intended by the writer or speaker of the source language text.This should be avoided in translation because it may create misunderstandings with respect to tone or content of the original message.
[3] m at e r i a l a n d m e t h o d The main source of data for the case studies is the ENPC+.The ENPC+ is a balanced corpus and contains around 1.3 million words of English original (source) texts with translations into Norwegian and around 1.3 million words of Norwegian original (source) texts with translations into English.Like the original ENPC it is bidirectional in structure, including comparable and translated texts in two languages (Johansson 2007, 11-12).
The objects of study are, as pointed out earlier, three items that have been found to function as cores of extended units of meaning with a negative prosody; Partington (1998) andBerber Sardinha (2000) in the case of commit, Stubbs (2001) in the case of signs of, and Louw (1993) in the case of utterly.The distribution of the three items under study in the English original texts is as follows: the lemma commit (53), signs of (38), and utterly (23).
The method applied in the contrastive analysis of semantic prosody can be briefly outlined as follows: 4 The first step is to establish the Norwegian (translational) correspondences of the English items on the basis of the ENPC+.In some cases, and where relevant, reference will also be made to their Norwegian sources, i.e. the Norwegian items that give rise to commit, signs of, or utterly in the English translations.Exploiting the bidirectional structure of the corpus, I will proceed to analyse the most frequently used Norwegian correspondences in the Norwegian original texts.By studying these in their natural environment (i.e. as they are [4] See Ebeling (2013) for a more detailed description of the procedure.
OSLa volume 6(1), 2014 cross-linguistic semantic prosody [167] used in source texts), I will be better equipped to determine how their semantic prosody relates to that of their English correspondences.
"Correspondence" is used rather than "equivalence" as the latter is a more loaded term that may suggest that items are equivalent at all levels of analysis.Correspondence, on the other hand, merely refers to observed correspondences in a corpus.As pointed out by Johansson: What we observe in the corpus are correspondences, and we use these as evidence of cross-linguistic similarity or difference or as evidence of features conditioned by the translation process.Analysing the correspondences we may eventually arrive at a clearer notion of what counts as equivalent across languages (Johansson 2007, 5).
[4] c a s e s t u d i e s Partington (1998, 66-67) uses commit as an example of a core of an extended unit of meaning with a predominantly negative semantic prosody, showing how it collocates with "items of an unpleasant nature".Moreover, Berber Sardinha's (2000, 98) cross-linguistic study of commit and its Portuguese counterpart cometer, based on comparable data, suggests similar prosodies for the units across the two languages.This tendency is confirmed also in the bidirectional translation data drawn from the ENPC+ where monotransitive commit with the meaning of "carry out" is exclusively found in the vicinity of negatively loaded events such as murder, crime, suicide, as illustrated in the concordance lines from the ENPC+ in figure 2 on the following page. 5(excluding reflexive commit oneself (to) and commit (to) 'pledge' or 'bind').
The lemma commit occurs 53 times in the English original texts in the ENPC+.40 of these have been translated by a form of begå, while the remaining 13 include four zero (Ø) correspondences, three instances of the verb utføre 'carry out' / 'perform', two of the verb ta 'take' and four translations with one occurrence each, as shown in Table 2 on the next page. [5] Commit seems to have an overwhelming bias for unpleasant collocates; however, a Google search for "commit a good deed" returned more than 30,000 hits.On the other hand, the 400-million-word Corpus of Contemporary American English returned no hits for this, suggesting that the hits on the web may reflect non-native use or "experimental" use...?
OSLa volume 6(1), 2014 In order to investigate whether the prosody of extended units of meaning with begå matches that of commit, the lemma begå in Norwegian original texts in the ENPC+ were analysed, of which there were 33 occurrences.The results clearly point to a similar tendency in terms of semantic prosody for units with begå, as evidenced by the negatively loaded collocations in all but one of the occurrences.In this case, begå is also monotransitive but has the meaning of write, as in be- [6] The four instances in the 'other' category occur once each and include translations such as skje 'happen' and gjøre 'do', as in (i) and (ii): (i) There was no crime committed.(MiWa1E) Det skjedde ikke noe kriminelt.
(ii) ... they 'd tried to express their hopeless resistance by committing sabotage, the night they'd got caught ... (ABR1) ... de forsøkte å gi uttrykk for den håpløse motstanden sin ved å gjøre sabotasje, den natten de ble fakket ... In comparison, the second-most frequent translation of commit -utføre -does not demonstrate an invariably negative prosody similar to that of commit or begå.Although it is found in combination with some of the same unpleasant collocates, the more varied semantic environment attested for utføre allows for a more mod- [7] This expression is in fact established in Norwegian, as evidenced by entries in dictionaries (Bokmålsordboka).The question is whether the expression has come about as the result of a "violation of prosody" in order to create irony?(see further Louw (1993)). [170] signe oksefjell ebeling erate view on what grounds units may be available as translations of each other.
As suggested by Tognini-Bonelli (2001 and2002), only begå would constitute a unit on a par with commit, while utføre -with a more mixed, or neutral, prosody overall -would not be considered as part of a comparable unit.This view on what constitutes a sound translational choice seems to be too strict in this case, as utføre clearly has the potential of entering environments that are more typical of commit (e.g.concordance line 4 below: uftøre drapene 'commit the murders').The more varied cotext of utføre is shown in the concordance lines in figure 4.
figure 4: utføre in the ENPC+ As regards the question of equivalence in the strictest sense, it seems as if begå is closer to commit than utføre is, as begå and commit are seen to share not only collocational but also communicative properties.Moreover, the opposite direction of translation lends evidence to this conclusion, as begå is found to be the source of commit in 21 out of the 33 instances attested in the ENPC+, thus reaching a mutual correspondence of around 70%. Mutual correspondence, as defined by Altenberg (1999), is a way of measuring the degree of correspondence between items or structures in a bidirectional corpus.It is the number of times the compared items are translated by each other divided by the total number of occurrences of the compared items in the corpus source texts, multiplied by 100.While 100% would suggest that the items are always used as translations of each other, 0% means that they are never used.Thus, 70% suggests a relatively strong mutual correspondence.
In terms of colligational patterns and semantic preference, both commit and begå seem to select a noun describing a decidedly bad act, as illustrated in Tables 3  and 4, where the nouns that co-occur with commit and begå are listed according to their frequency in the ENPC+.Stubbs (2001) gives examples of various units with marked prosodies, including signs of.After establishing that the extended phrase show signs of "has a very negative discourse prosody" (ibid.: 458), he also makes a case for the core signs of being "usually negative".This observation is confirmed in the 38 occurrences of signs of in the English original texts in the ENPC+.The following concordance lines bear witness to this.
figure 5: signs of in the ENPC+ [8] tabbe 'mistake', utilgivelig gjerning 'unforgiveable act', overgrep 'offence'/'harassment', voldshandling 'act of violence', straffbar handling 'punishable offence', dumhet 'stupidity', helligbrøde 'sacrilege'. [172] signe oksefjell ebeling Although signs of followed by clearly negative collocations does not occur with the overwhelming frequency that was noted for commit, it is still the predominant environment for signs of, triggered also by collocations to the left of the core, as in showing little or no signs of life in the last concordance line above.In this example, based on the immediate right collocate life, we would draw the conclusion that this is a positive environment; however, to the left of the core we find little or no, giving it a negative prosody after all.Thus, in the material studied here, signs of is associated with negativity in about 70-80% of the cases.
The Norwegian translations are practically devoid of variation in their rendering of signs of.All but three include a form of tegn 'sign(s)'; 33 out of the 38 instances have tegn til (lit.: sign(s) to) or tegn(ene) på (lit.: (the) sign(s) on), while three have a compound including tegn, e.g.sykdomstegn 'signs of disease' (lit.: sickness-sign).Two of the occurrences that do not include tegn have spor av/etter 'traces of' instead, while one has been rewritten, as shown in the first concordance line (from AH1), where signs of death's approach has been rendered as sto overfor døden 'faced death' (lit.stood opposite death).
If we focus on the congruent correspondences, namely the lemma tegn til/på, when moving on to look at the Norwegian original texts, it can be concluded that these also have a negative bias, as illustrated in the concordance lines in figure 6 (with collocates such as sammenbrudd 'breakdown', galskap 'madness', hastverk 'haste'/'hurry', etc.) and as quantified in Table 5 on the next page.5, 41 out of a total of 51 occurrences of tegn på/til in the original texts in the ENPC+ have a negative bias.On the basis of these figures, it can be concluded that signs of and tegn til/på are not only formally congruent; they are also well matched cross-linguistically in terms of communicative function, i.e. semantic prosody.Nevertheless, it should be mentioned that with regard to colligations, signs of and tegn på/til differ slightly.While both may take nouns or noun phrases as their right colligations, the Norwegian core also occurs regularly with a non-finite clause.This use is not noted for signs of in the ENPC+ material; however, a look in the British National Corpus (BNC) shows a fair amount of nonfinite -ing clauses following signs of as well, e.g.Whenever these arguments showed signs of running out of steam, ... (BNC: ABU 423).Stubbs (2001, 458) also suggests that there is a wider pattern with show, i.e. show signs of, an observation which will not be pursued further in the present study.
[4.3] Utterly Louw (1993, 160) claims that "there are few 'good' collocates" of utterly in the 18 million word corpus at Cobuild.The overwhelmingly negative, or "bad" in Louw's terms, prosody of utterly is also confirmed in the ENPC+ data; 20 out of 24 instances in the English original texts are clearly associated with an unfavourable right-collocate, as attested in the concordance lines in figure 7.
figure 7: utterly in the ENPC+ Although we have relatively little contrastive data to go by, the Norwegian correspondences offer interesting insights.While the Norwegian translations of the 20 instances with a negative prosody reveal a number of zero (Ø) correspondences (9) in addition to fullstendig 'completely' (5) and helt 'completely' (lit.: wholly) (4), the Norwegian sources of utterly have aldeles 'altogether' in addition to helt, fullstendig and Ø.The main observation to draw attention to with regard to overt correspondences seems to be that there is a division of labour in Norwe-gian between fullstendig and helt (and to some extent aldeles) as correspondences of utterly.This is clearly different from the case of both commit and signs of where there was one main Norwegian correspondence with a fairly high mutual correspondence and a similarly negative prosody.
As the scenario for utterly and its correspondences is different at the outset, we will investigate to what extent any of the three main correspondences has a prosody that matches that of units with utterly.
Fullstendig and aldeles are quite straightforward in terms of function and word class, i.e. they are adverbs of the type that Quirk et al. (1985, 590) call amplifiers/maximizers and occur 86 and 39 times, respectively, in the Norwegian original texts.Helt, on the other hand, is more versatile and is found as a noun, an adverb and an adjective.Thus, for the purpose of this study only the premodifying adverb uses of the 1,023 occurrences of helt in the ENPC+ are explored.
In the material studied, fullstendig seems to match utterly fairly well in terms of semantic environment and prosody, while aldeles and helt do not.Although the main impression of aldeles is that it is often associated with undesirable states, such as hjelpeløs 'helpless', teit 'stupid' and unødvendig 'unnecessary', the concordances show a more mixed scenario, with aldeles being used in positive contexts (see concordances AnHo2N, OEL1 and EG2) in figure 8.The picture that emerges for helt is that, of the three main overt Norwegian correspondences of utterly, this is the item that is least tied to a particular semantic prosody.It can be said to be a more general adverb than the other two in that it is not strongly associated with any particular discourse function (cf.Sinclair's "open choice" in Sinclair 1991).Particularly when compared to utterly and fullstendig, we can observe that helt has a different collocational distribution, ranging from negative (feil 'wrong'), through neutral (greit 'ok') to positive (klart 'clear').[9] This is an interesting combination, as vill/wild will not always be associated with a negative semantic prosody.However, it lies outside the scope of this study to explore this further.
OSLa volume 6(1), 2014 cross-linguistic semantic prosody [175] In connection with the cross-linguistic instability of the prosody of utterly and its Norwegian correspondences that has emerged it is important to note that a search in the 100 million word monolingual BNC tends to give a less stable picture of the claimed negative prosody of utterly.Nevertheless, its association with mainly negative states is quite obvious also in the BNC, albeit perhaps not to the same extent as for commit and signs of.This may be a contributing factor to the cross-linguistic result that utterly has several and not clearly delimited correspondences in terms of prosody, including quite a few zero correspondences.The last observation also suggests that since there is not one "perfect" and relatively stable counterpart of utterly in Norwegian, translators may more easily choose the strategy of leaving it out rather than using a more or less close overt match.
[5] c o n c l u d i n g r e m a r k s The case studies offer some new insights into the nature of semantic prosody across languages.In the case of units with commit and signs of there are Norwegian correspondences that may be said to be equivalent at most levels of discourse, i.e. the units of meaning in both languages seem to be functionally complete along the same dimensions (cf.Tognini-Bonelli 2002).Utterly, on the other hand, resembles cause in many ways, in that it does not seem to have a clear Norwegian correspondence that matches the prosody of units with utterly.One of the research questions this paper set out to answer can now be rephrased in the following manner: Does the divergence in prosody noted for cause and now also for utterly and their correspondences in Norwegian have wider cross-linguistic implications?
On the basis of the present findings, the answer to this would be "possibly", at least if we want to be absolutely sure to avoid misunderstandings "with respect to tone or content of the original message" (Berber Sardinha 2000, 96).Moreover, as pointed out by Gries: [...] it [semantic prosody] also has implications for [...] foreign language teaching since, for example, if a foreign language learner uses a word w without being aware of w's semantic prosody, this may result in comical situations or, more seriously, communicating unwanted implications.(Gries 2009, 16) However, even if the Norwegian correspondences are not always seen to match the negative prosody of their English counterparts, data from Norwegian original texts show that it is not an unnatural environment for e.g.helt (utterly) or utføre (commit).In these cases it can be argued that the prosody is naturally preserved in translation, although the units themselves may not be said to be completely equivalent.Thus, Tognini-Bonelli's claim that "only when units are functionally complete [i.e. they constitute a single choice including a certain semantic prosody] do they become available as translation equivalents or as comparable units of meaning between two languages" (Tognini-Bonelli 2002, 91-92) seems to be too categorical.
Moreover, translators, lexicographers and teachers alike should be aware not only of the clear-cut cases of differing prosodies across languages, but also the more subtle ones.
The present study has but focused on three items functioning as cores of extended units of meaning, all of which had been assigned a negative prosody in previous studies.Future research would benefit from more studies of this kind and should also explore other dimensions where, for example, potential cores that are negative in nature, may possibly be seen to have positive semantic prosodies, e.g.awfully.
The present study has shown that the contribution of bidirectional translation corpora in the discussion of semantic prosody is not a trivial one.Although it in some cases may reach the same conclusions as other previous studies of crosslinguistic prosodies, it also offers a less biased approach to the actual correspondences across languages.(1994)(1995)(1996)(1997), Dept. of British and American Studies, University of Oslo.Compiled by Stig Johansson (project leader), Knut Hofland (project leader), Jarle Ebeling (research assistant), Signe Oksefjell (research assistant). http://www.hf.uio.no/ilos/forskning/forskningsprosjekter/enpc/index.html The extended version of the ENPC (ENPC+) has been described in Ebeling & Ebeling (2013).
figure 2: commit in the ENPC+

figure 8 :
figure 8: aldeles in the ENPC+ a c k n o w l e d g m e n t s I would like to thank Susana Afonso and Diana Santos for their instructive comments on a previous version of this paper.c o r p u s enpc -The English-Norwegian Parallel Corpus r e f e r e n c e s Altenberg,Bengt.1999.Adverbial connectors in English and Swedish: Semantic and lexical correspondences.In H. Hasselgård & S. Oksefjell (eds.),Out of Corpora: Studies in Honour of Stig Johansson, 249-268.Amsterdam/Atlanta: Rodopi.Berber Sardinha, Tony.1999.Padrões lexicais e colocações do português.Paper presented at the symposium Processamento Computacional do Português, 9, InPLA, PUCSP, Brazil.http://www2.lael.pucsp.br/~tony/temp/publications/1999padroes_inpla.pdf.

table 4 :
Noun collocations of begå in the Norwegian original texts.(Neg stands for "Contributing to a negative prosody", and Neut stands for "Contributing to a neutral prosody".) [4.2] Signs of In an article discussing how pragmatic meaning is encoded in phraseological units,

table 5 :
The negative bias of (variant forms of) the core tegn på/til in the Norwegian original texts