I am wild about cabbage : evaluative ‘ semantic sequences ’ and cross-linguistic ( dis ) similarities

The paper investigates the phraseology of evaluatio n in two comparable corpora of discussions from blogs in Italian and in English. S ubjectivity markers are taken as an indication of the significant role that the writer’ s ‘self’ plays in the genre, ideal territory for an analysis of the language of evaluation. Afte r considering collocates and grammar patterns of the selected markers, the analysis cent res on evaluative ‘semantic sequences’ by aligning typical recurrent surface arrangements wi h strings of prototypical meaning elements such as ‘entity or process evaluated’, ‘ev aluation’ and different ‘sources of evaluation’. Four types of sequences are identified : ‘basic’, ‘framed’, ‘dialogic’ and ‘concessive’. The results attest for substantial se mantic similarities over and above lexico-syntactic and inter-linguistic mismatches. S emantic sequences can be shown to be useful tools for cross-linguistic analysis.


Introduction
The present paper is devoted to the analysis of evaluative phraseologies as attested in two comparable corpora (1.5 million words each) consisting of English and Italian texts taken from the web genre of blogs.
Our assumption is that the 'evaluatively charged' nature of blogs' postsand comments especially -makes them ideal territory for an analysis of the language of evaluation (Herring et al. 2004;Baron 2008;Myers 2010).The investigation follows up from previous research (Bondi & Seidenari 2012) on 1 st person pronouns and adjectives (I, me, my; mi, io, me), which score at the very top of the keyword lists extracted from the two blog corpora.The pervasiveness of such 'subjectivity markers' reflects the high degree of subjectivity of blogs and can be taken as an indication of the significant role that the writer's 'self' plays in the phraseological realizations of evaluative meaning when looking at texts "in which authors report on their lives and inner thoughts and feelings" (Herring et al. 2004:6).The self-disclosure elements typically identified with personal blogs (Miller & Shephard 2004:9) have generally survived genre migration over to blogging news sites or corporate blogs (Garzone 2012:235).In other words bloggers and blog commenters, in discussing the wide range of issues that are debated on in blogs, show a strong tendency to evaluate and show (or imply) their attitude to them.
Starting from a brief presentation of the corpora analyzed and the type of analysis chosen, the paper offers a preliminary overview of subjectivity markers, before moving on to the lexico-grammar of selected items and to the typical sequences of semantic elements that characterize expressions of evaluation.

Materials and methods
The analysis is based on two comparable corpora of blog texts in English and Italian that were posted on-line between September 2008 and September 2009.The collection amounts overall to 3,042,023 running words: 1,582,448 words from 65 blog sources for the Italian corpus, and 1,459,575 words from 71 blog sources for the English corpus. 1or the corpus design, Technorati's directory (http://technorati.com/blogs/directory/)was selected as one of the most popular blog resources among those available on the Web.Each corpus is organized according to 6 macro-categories derived with minor adaptations from Technorati's content classes at the time.The macrocategories are: Business, Entertainment, Science & Technology, Lifestyle, Politics and Sport.
Furthermore, the classification features a finer-grained, second level of categorization according to 24 more micro-categories that were adapted from the original 50 available at Technorati.com.The microcategories, each represented by approximately 60,000 words from three different sources, are listed in Table 1 The analysis is based on a preliminary process of keyword identification.The keyword list is automatically generated by the 'Keywords' program which is part of the WordSmith Tools suite of corpus analysis software (Scott 2008, version 5.0).We start with an overview of the keyword lists obtained comparing the two BLOG corpora with reference corpora of general Italian and general English: the CORIS/CODIS (Rossini Favretti 2000) (CORpus di Italiano Scritto/Corpus Dinamico di Italiano Scritto -Corpus of Written Italian/Dynamic Corpus of Written Italian) and the British National Corpus (BNC), respectively.From such keyword lists we select the Salient Grammatical Words (SGWs) (Gledhill 2000;Groom 2010), featuring at the top of each keyword list, i.e. 1 st personal pronouns (EN I, me; IT mi, io, me).
For each keyword we extract a random sample of concordance lines from both the EN and IT blog corpora, aiming at reaching at least 250 hits for each keyword.It was felt to be important to cover the whole range of blogs in each corpus and therefore to collect samples at regular intervals throughout it: more specifically, we extract one concordance line in 100 for each keyword sample in the EN BLOG corpus, and one in 10 for the IT BLOG corpus.The samples we used to carry out our investigation of the 1 st person subject pronouns in the two languages (I/io) are the following: the EN BLOG corpus attests 27,529 occurrences of I, and the sample we used consists of 275 concordance lines.As regards the Italian correspondent first-person subject pronoun io, the corpus attests 4,810 occurrences of io, and the sample consists of 481 concordance lines.It was felt to be important to get higher figures in Italian, to make sure we could map an area that is hardly explored in the literature.
As regards the clitic form of the Italian 1 st person singular pronoun mi [to me], with 8,345 occurrences in the corpus, the analysis is based on a sample of 834 concordance lines.Regarding the EN personal pronoun me and the Italian correspondent me [to me], the samples consist of 387 concordance lines of EN me and 317 of IT me, respectively.Since the EN corpus attests 3,879 occurrences of me, we decided to extract one concordance line in 10 as to get a larger sample to be explored.
The analysis follows a two-step procedure.First, we perform a concordance analysis of such keywords in order to investigate the typical lexico-syntactic phraseological arrangements they are involved in (i.e.typical collocates and grammar patterns), and explore instantiations of evaluative meaning that are realized in these phraseologies.The aim is to find out whether such keywords are involved in the expression of evaluation in a way that is comparable between the two languages.
Second, following Hunston (2008) and Groom (2010), we provide a comprehensive semantic interpretation of all the structural information collected in terms of abstract evaluative "semantic sequences", i.e. "recurring sequences of words and phrases that may be very diverse in form [...] sequences of meaning elements, rather than […] formal sequences" (Hunston 2008:271) The aim here is to investigate whether the evaluative 'semantic sequences' identified for the English SGWs may be applied in the description of their Italian counterparts as well.
The phraseological arrangements identified in our analysis consist of 'prototypical sequences' and 'argumentative sequences'.The 'prototypical sequences', where the writer expresses her/his stance towards a message/object, were classified as:

Preliminary analysis: an overview of subjectivity markers in English and Italian
A preliminary overview of quantitative data in the two languages highlights bloggers' propensity for subjectivity and self-expression.As   The analysis of the 1 st person subject pronouns in the two languages confirms that both I and io have a significant role in the expression of evaluation.Indeed, for 145 occurrences of I (52.72% of the total) the sample attested an evaluative use.The share of evaluative instances for the correspondent Italian pronoun io is comparatively slightly lower, but not significantly dissimilar: 202 (41.99%) were found attesting an evaluative meaning of some sort.However, a direct comparison between the two sets of pronouns requires a qualification, since Italian, unlike English, is a pro-drop language.This means that in Italian the subject pronoun may have a 'stressed' (attached to a verb) and 'unstressed' form, and that the pronoun is not explicitly co-selected with the verb, as the subject is always signalled though morphology in the verb ending (Maiden & Robustelli 2000:93).Typical examples can be found among the top keywords generated: verb forms like credo [I believe], concordo [I agree], spero [I hope] are clearly identifiable as 1 st person forms through the -o ending which is in most un-marked cases the only 1 st person marker.In the case of credo, for example, on the basis of 200 randomly selected concordance lines, only 6% (12/200) of the occurrences are actually found with explicit subject pronoun (io credo), whereas the percentage of occurrences without it is 94% (188/200).The trend is similar with concordo and spero.The corpus attests only 5.5% occurrences of io concord (11 out of 200 concordance lines), and 10% instances of io spero (20/200).The use of stressed pronouns is often required for purposes of emphasis or contrast.
Personal pronouns in English and verb endings in Italian can be considered comparable as they perform the same function.However, for the purposes of the present analysis, we investigated only instances of explicit pronouns, leaving verb forms for further investigation.
Another point worth making here is that in Italian the lexical realization of subjective expression is not primarily realized through the personal subject pronoun io or the direct/indirect object pronoun me, the tonic forms of the pronoun.The clitic form of the first-person singular pronoun -mi [lit.me; to me] -ranks higher than I in the keyword list, with 8,345 occurrences, that is 52.7 instances per ten thousand words (hereafter, pttw).Mi is the unstressed form of the pronoun, acting as direct/indirect object (as well as reflexive pronoun) and typically precedes personal verb forms.The most prototypical expression of an evaluation in Italian has a 'clitic pronoun + verb' construction, usually rendered in English by 'I + verb': mi piace [I like].Using Halliday's (1994) terminology for the analysis of the meanings expressed in the clause, we could say that, whereas English I is grammatical subject and Senser of the mental process (of Emotion), Italian mi is grammatically indirect object, while still being semantically in the role of the Senser.Mi stands for the indirect object complementing a range of very common socalled 'impersonal' verbs (e.g.mi sembra/mi pare che... [lit. it appears to me that, I think that]), also frequently used to express an evaluation through a mental process (of Cognition).The following table shows the ten most frequent collocates occurring in R1 position.The quantitative significance of such lexical items and their distinctive role in the realization of evaluative meaning are noticeable both for English and Italian corpus data.
On the whole, similarities were attested as regards the verbal collocates of the personal pronouns I and io across the two corpora.Apart from the link-verbs (be and essere [to be]), the most substantial group accounts for verbs having to do with cognitive processes that may be broadly defined as epistemic: EN think, know, guess, believe; IT credo, penso, trovo, capisco, so.The second most large set of wordforms include verbs that may be generally categorized as verba dicendi: EN wonder; IT dico, direi.A final sub-set attests verbal word-forms conveying an attitudinal meaning of some sort: EN love, hope, like, feel; IT concordo, spero, adoro.
However, differences exist between English and Italian as regards subjectivity.As already noticed, no equivalent for the clitic pronoun mi can be found in English (cf.mi piace vs.I like).Subjectivity is also not overwhelmingly realized through verbs marked in the first person.Indeed, a number of quite common epistemic and attitudinal verbs are canonically realized through constructions that require the third person instead.An illustrative example is the very frequent Italian verb sembrare or parere [seem; look like].In a sentence such as sembra un paradosso [it looks like a paradox], the link-verb word-form sembra is morphologically marked as third person singular and there can be no other surface realization of a subject in the first person.Nonetheless, the sentence logically implies a 1 st person subject as Senser of a mental process of cognition: we can rephrase the sentence as I think that this is a paradox.Such constructions are very productive in Italian and, indeed, they are very frequently attested in the IT BLOG corpus: sembra (88 occurrences) / pare [(it) seems] (104).

Evaluative 'semantic sequences' of pronouns EN I and IT io/mi
As fully grammatical and highly frequent word-forms, the 1 st personal keywords EN I and IT io and mi provide the data-base featuring the widest range of syntactical arrangements and lexical variation.As such they are the best starting point for trying to hypothesize general evaluative 'semantic sequences' that may be able to account for most of the actual evaluative expressions attested in the corpora.
The results show great cross-linguistic correspondence in the typical evaluative phraseological arrangements that both I and io/mi participate in.These were analyzed as sequences of prototypical meaning elements such as 'entity or process evaluated', 'evaluation', and different 'sources of evaluation'.From the analysis two major semantic sequences emerged: 'prototypical' and 'argumentative' sequences.
As regards the frequency of occurrence and distribution of the two basic categories above, the findings show similar trends, though with different figures.Both corpora display a heavy concentration of 'prototypical' semantic sequences (90.34% in the English corpus, 82.17% in the Italian corpus) in comparison with the 'argumentative' sequences (EN 9.66%;IT 17.82%).Another level of similarity can be observed between the types of the 'prototypical' sequences identified ('basic' and 'framed').The 'framed' sequence turned out to be the most frequently attested for all the evaluative instances found in both samples: it occurs in 54.19% of the English sample and in 75.90% of the Italian sample.

Prototypical 'framed' sequences with I and io/mi: a cross-linguistic perspective
In a framed sequence a cognitive, a speech or in general a communicative process is made explicit (most typically) immediately following the node word.This cognitive element of the sequence seems to function as a sort of marker, or 'framework', anticipating and encapsulating the actual evaluation taking place toward the end of the sequence.Moreover, the framework -most typically subjectivity marker + verb phrase -performs the primary function of unequivocally signalling the source of the evaluation, i.e. the speaker or writer taking the responsibility for the evaluation realized by the subsequent element of the sequence.In this sense, the 'framework' meaning element may be regarded as a form of self-attribution (Bondi 2010).Since the occurrence of such a framework appears to be a consistent feature of the sequence, we will provisionally refer to it as the 'framed' sequence.
In Italian the framed sequence with io is typically realized by the pattern 'io v (0/)that-clause', accounting for 19.4% of all the occurrences in the sample.The most typical verbal exponents of the pattern are epistemic verbs -such as credo [I believe, I think] and penso [I think] -and verba dicendi like dico [I say], all scoring among the first 12 most frequent R1 collocates of io overall (cf.Table 5 above (1) io penso che molti di voi di F1 ne capiscano ben poco (...) [I think that most of you don't know the least thing about F1 (...)] (2) Io credo sia stata la scelta più giusta per me [I think that was the best choice for me] As the examples show, the frame signals the source of the evaluation and the 'that-clause' typically instantiates an act of evaluation.

.) [I guess something is wrong (…)]
Another frequent pattern realizing a framed sequence with mi is the pattern 'mi link-v n/adj', where the verbs sembra, pare or sa di function as link verbs, and then followed usually by an evaluative noun or adjective.Overall, it accounts for 25.2% of citations of mi taken as a sample for the present analysis.Here are some illustrative examples: (5) Mi pare un'ottima soluzione [That seems a very good option to me] (6) (...) la sua lettera mi sa tanto di falso [(...) your letter really seems like a fake to me] The Italian sample of io also attests a significant occurrence of the pattern 'io v n/adj'.Forms of the verb trovo (to find) are the most frequent with the pattern (18.5%), e.g.From a cross-linguistic perspective, it is interesting to note that remarkable similarities were attested as regards the surface realizations of the framed sequences with I. Like Italian, the English 'Iv (0/)thatclause' pattern is attested as the most frequent in the realization of the framed sequence with I (17.2%), as shown in the following examples: (11) I think that you have mental problems.
Cross-linguistic similarity is also confirmed by the prominent role of the pattern 'I v wh/if-clause', e.g.I wonder what is the most sensational in this story; I often wondered if he was secretly tone-deaf, which was also found to be significant in the IT BLOG corpus.A direct correlation is attested in the two samples between the two equivalent verbs I wonder and mi chiedo frequently realizing the sequence.'I v QUOTE' and 'I v' patterns are amply attested in the EN BLOG corpus, even if they are not the most frequent, among the patterns realizing the framed sequence in the sample extracted for I: (13) I must say that's a pretty cool puzzle.( 14) Her fans are still living in the 80's, I guess.
It is worth noting here again that the same two patterns realizing the framed sequence are frequently attested in the IT BLOG corpus as well, especially in combination with epistemic verbs (e.g.Una maestra, non può limitarsi a sapere le cose -io credo... [A teacher cannot restrict herself to just knowing things, I think...]) and verba dicendi (e.g.io dico che l'inserimento dei blog è corretto [I say that the inclusion of blogs is correct]).
In Table 6, illustrative instantiations of the 'framed' sequence are presented across the two language corpora, together with the relevant grammar patterns.5.2.Prototypical 'basic' sequences with I and io/mi: a cross-linguistic perspective In a basic sequence, the evaluation and the source of evaluation are directly linked to the entity or process evaluated, without any 'mediation' provided by a projecting framework as in the case of the framed sequence.Accordingly, as a way to unequivocally distinguish it from the previous one, we propose to refer to it as basic.The basic sequence is less frequent than the framed one in the samples selected (EN 45.80% of evaluative instances; IT 24.09%).
In Italian the pattern that most typically realizes this sequence with mi is the 'v' pattern, that is, followed by a verb alone, with no further complementation.It accounts overall for 19.7% of the citations of mi in the sample (just under one over five citations of mi in the corpus).In a cross-linguistic perspective, as was the case with io, 'v n/prep' is the most frequent pattern realizing the 'basic' sequence with I (45%).In this syntactic arrangement, I is followed by a verb which in turn takes a direct object or is complemented by a prepositional phrase, such as in I love indie games or I don't care about HIM.
Like io and mi, the 'v (0/)that-clause' pattern is attested for the basic sequence as the second most common (31.67%), typically in combination with the verb to hope (e.g.I was hoping the review process would focus on the quality of the software).
The third most frequent pattern (accounting for 23.33% of the whole sample) is 'link-v adj/n/prep'.Examples are: (24) I am wild about cabbage.Again, as shown above, the pattern was frequently attested for io as well, cf.example (21) above.The full range of patterns that I and io and mi take part in are illustrated in Table 7, together with the relevant instantiations of the 'basic' sequence.As regards the 'concessive' sequence, it may be considered as an iteration of the 'basic' one: the 'evaluation' of an entity or a process is firstly presented as a 'concession'; then, a second evaluation follows (typically introduced by adversative conjunctions) realizing a 'contradiction' element.This is an illustrative example taken from the EN BLOG corpus: The contradictory element may be more or less articulated, for example, including or not including reason for rejection.
In both corpora the incidence of argumentative sequences in the samples analyzed is altogether relatively low (EN 9.66%; IT 17.82%).The two corpora also showed similar patterns in the different frequency of the two types of argumentative sequences.The 'dialogic' sequence turned out to be the most frequently attested in both samples (EN 78.58%; IT 83.33%), while the 'concessive' sequence tended to be lower (EN 21.42%; IT 16.67%).The Italian corpus thus shows greater interest in argumentative sequences, but these are mostly dialogic, rather than concessive.
The 'dialogic' sequence with I is consistently realized in the sample by the verbal word-form agree, typically in the 'v prep that-clause' and the 'v' patterns.Examples are: (27) I don't agree with Mr. P that Twitter breeds trust and togetherness.
(28) But I agree, Oprah should have shown both sides of domestic violence during her special.
A substantial share of corpus data attest the 1 st person form of the verb to be, 'm and am, as realizing the 'dialogic' sequence in the 'link-v with' pattern.Typical instances of the sequence are the following examples taken from the sample: (29) I am totally with BK2009 on one point -the Smart is a deathtrap waiting to happen.
(31) I am with you on the celery dessert thing... not good.
The syntactic arrangement of the sequence is constant: am, functioning as a link-verb, is followed by a prepositional phrase governed by with, that consistently stands for the human entity that the writer agrees with.
Frequently, another prepositional phrase follows which is governed by on: the phrase provides a reference to the subject matter that the writer agrees on (cf.29 and 30 above).Such references are made frequently in general terms, usually by means of a pronoun or a general noun (such as thing or point).Finally, the sequence gets completed with the adjunction of a chunk of text that usually follows (but sometimes it may precede the prepositional phrase), reporting a sort of reformulation of the original statement that the writer agrees on, as in ( 29): the Smart is a deathtrap waiting to happen.
Both the semantics and the patterns ('v prep' and 'v') described above are perfectly comparable with the typical instantiations of the sequence with the personal pronoun io when followed by the 'dialogic' Italian verbs concordo [I agree] or condivido [I agree].Examples are: (32) Anch'io concordo con il post del Sig.Fabris.
[But I fully agree, I find the game great fun.] In realizing the 'dialogic' sequence, the clitic mi is frequently involved in the phraseological construction trovarsi + the adverb d'accordo as, for example, in mi trovo completamente d'accordo con la tua valutazione [I totally agree with your evaluation]).In such a case, the construction is semantically very close to concordo [I agree].
Let us move on to consider the 'concessive' sequence.The sequence is instantiated in a number of concordance lines where I is followed by the English verbs understand or know.The 'v n' and 'v that-clause' are the patterns that typically instantiate the 'concessive' sequence.Examples are: (34) So I understand the utility of over-the-top populist rhetoric, even when it's kind of silly.But Claire McCaskill isn't a candidate in a campaign.
(35) I know, but for some reason I think this is cool and kind of big.
Both the semantics and the patterns ('v n' and 'v that-clause') described above are perfectly comparable with the typical instantiations of the sequence with the Italian personal pronoun io when followed by the verb capisco [I know], and the typical occurrence of the adversative conjunction ma [but], as a way to introduce the 'contradiction' meaning element.Examples are: (36) io capisco benissimo che ci siano delle differenze, non ho detto che sia tutta roba identica.Però comunque c'è tra di esse una differenza non abissale (...) [I understand very well that there are differences, I didn't say that it's all the same stuff.But in any case there is a difference between them that is not abysmal (...)] (37) io capisco che ognuno ha i suoi gusti, ma cacchio, un pò di obbiettività.
[I understand that everyone has their own tastes, but damn, a bit of objectivity.]

'Semantic sequences' with I and io/mi: cross-linguistic (dis)similarities
The results of the analysis carried out so far seem to attest remarkable cross-linguistic similarities in the surface realizations of the 'semantic sequences' identified in the two corpora.Indeed, the most frequent grammar patterns instantiating the 'framed' and the 'basic' sequences were found to be, with some significant exceptions, essentially the same.As shown above, the prominent role of 'v that-clause' pattern (e.g.I think you have mental problems;... e io non penso che questo faccia bene al PD [... and I don't think that this will do any good to the Democratic Party...]) and 'v wh/ifclause' pattern (e.g.I wonder what is the most sensational in this story; mi chiedo a cosa possa servire [I wonder what's the use of it]) in the realization of the 'framed' sequence was confirmed in both samples.The patterns realizing the 'basic' sequence were found to be comparable to a considerable extent across the two corpora.The 'v n/prep' pattern (e.g.I love indie games; io approvo tutte le battaglie di Beppe [I approve all of Beppe's demonstrations]), and the 'v that-clause' pattern (e.g.I wish that I had more opportunity to spend more time actually doing research than simply commenting on it; mi dispiace che Berlusconi non abbia adempiuto agli impegni presi [I am disappointed that Berlusconi did not fulfil his commitments]) were all found to be significantly attested both in the English and in the Italian samples.
However, significant differences can be noticed as well.For example, the 'v-link adj/n/prep' pattern is dominant in the realization of the 'basic' sequence in the English sample (22.9%) (e.g.I'm also all for Glen's idea about Cannonball Run), but it is much less frequent as regards Italian (2.2%) (e.g.io non ne sarei così orgoglioso [I wouldn't be so proud of that]).Conversely, the same pattern very frequently realizes the 'framed' sequence in the Italian corpus (25.2%) (e.g.mi pare un'ottima opzione [that seems a very good option to me]).By comparison, it does so much less commonly in the English sample (1.1%) (e.g. this one seems obvious to me).
Other notable dissimilarities have to do with differences in the patterning of lexical items that nonetheless are frequently found in both samples.For example, the considerable share of instances attested for the 'v' pattern in Italian that is mostly realized though the verbal construction mi piace (e.g.Mi piace molto il tuo modo di scrivere [I really like the way you write]) is perfectly replicated in English through the 'v n' pattern (e.g.I like the idea of this plum kuchen).
Again, the instances of 'm and am in the 'v-link with n/on n' pattern realize the 'dialogic' sequence in precisely the same way as Italian verbs such as concord or condivido instantiate it.All these instances replicate Italian instantiations of the sequences through lexical and syntactic choices that are very different in the two languages.
All in all, we can argue that the degree of syntactic continuity across the sequences realized in the two corpora is considerable, although some significant discontinuities have emerged.Semantically and pragmatically, however, the kinds of evaluative meanings that are frequently conveyed in the English and Italian blog texts are essentially the same.

On the lexico-grammar and typical semantic sequences of the personal pronoun EN me and IT me
The EN personal pronoun me ranks 47 th in the keyword list, with 3,879 occurrences.The IT equivalent me [to me], used as the tonic accusative form of the first-person pronoun io, ranks comparatively slightly higher in the keyword list: it scores respectively as the 17 th (with 3,178 occurrences).
As was the case with I and io in the previous section, evaluative instances turn out to make up for a good deal of the overall citations of the pronoun (EN 49.45%; IT 50.18%).
In both corpora, the EN pronoun me and IT me frequently take part in prepositional phrases.In particular, the prepositions to and for are attested respectively as the first and the second most frequent L1 collocates of me.A very similar picture may be described for the collocational profile of the Italian pronoun me: the prepositions a [to], secondo [according to] and per [for] rank respectively as the 1 st , 2 nd and 3 rd most frequent collocates overall for me in the IT BLOG corpus.
Both EN me and its correspondent IT me frequently co-occur with copular constructions and have a distinctive role in the realization of the 'framed' sequence.As regards EN me, the to me pattern is the most frequent (52.55%): typically such prepositional phrases act syntactically as a sentence adverbial, as in To me, this is a sign that Apple is rotten and scared of competition.Alternatively, to me is frequently used in combination with verbs such as seem or sound that act syntactically as link verbs, in patterns such as 'link-v adj to me'(e.g. that just seems silly to me) or 'link-v like n to me'(e.g.Sounds like a menopause problem to me).The prepositional pattern with for is the second most frequent (29.92%): again, the prepositional phrase is typically employed as a sentence adverbial (e.g. they were both miserable experiences for me).Finally, the sample attests the significant occurrence of another prepositional pattern, with me (17.51%), in instances such as everything he says above rings true with me or This touches home with me.
From a cross-linguistic perspective, it is interesting to note that the instances of EN me exemplified above are both syntactically and semantically very similar to those found for the Italian equivalent me realizing 'framed' sequences, for example, in cases where the prepositional phrases with IT me act as sentence adverbials, in a pattern such as secondo me/per me [to me], as shown in the following examples: (38) per me è il miglior gioco di questa generazione.
[to me, that is the best game of this generation.] (39) secondo me sono bellissime in queste foto.
[to me they are extremely beautiful in these pictures.] There is also a strong resemblance between the two pronouns in cases where the IT pronoun combines with some of its most frequent verbal collocates, such as sembra [seem], or pare [seem], in patterns such as 'a me link-v n/adj' or 'me v (0/)that-clause'.Some illustrative examples are: (40) A me è sembrata un'ottima idea.[That seemed a very good idea to me.] (41) (...) a me pare del tutto normale (...) [(...) that seems completely normal to me (...)] As we have seen in section 5.1., the patterns are a mere variant of the corresponding patterns employing mi, cf.mi sembra la soluzione più logica [to me it seems the most practical solution]; Mi pare chiaro che la colpa è stata del cliente (...) [It seems clear to me that it was the customer's fault (...)].As a result, it would be reasonable to hypothesize that both the Italian and the English pronouns seem to contribute to the phraseological realizations of the 'framed' sequence in ways that are very similar in the two languages.Interestingly, the EN sample attests the occurrence of a fairly crystallized construction, is it (just) me, that consistently functions as the main sentence to which another direct question is coordinated through the disjunctive conjunction or, as in Is it just me or does she get more annoyingly self-righteous every week?.Within the 'framed' sequence, is it (just) me ...? acts as the 'framework' element introducing the rhetorical question that follows, which in turn constantly implies a negative evaluation on the part of the writer.
Table 8 exemplifies the full range of patterns of EN me and IT me realizing the 'framed' sequence.The data show that both the EN and IT pronouns have a consistent involvement in the realization of the 'framed' sequence.However, the two language samples attest an equally common involvement of the pronouns in the realization of the 'basic' sequence.
As regards EN me, it is found to occur preferably within verbal patterns.Rather interestingly, the verbs realizing these patterns within the 'basic' sequence seem to share a semantic trait: most of them are verbs whose meanings imply an emotional or cognitive response on the part of the writer.Typical examples of such verbal word-forms are annoy, bore, bother, crack up, haunt, impress, interest, puzzle, strike, surprise etc. Accordingly, when the 'basic' sequence is selected in combination with the pronoun me, the evaluation of an entity or a process is frequently presented in terms of the emotive or cognitive response it provokes in the writer.Here are some illustrative examples: (42) Ice Age pleasantly surprised me.
(43) That is the first thing that struck me.The range of verbs described above realize a multiplicity of interrelated patterns.The most frequent is the 'v n' pattern, where EN me acts syntactically as the direct object of the verb (e.g.I got rid of it, it creeped me out).
The next most frequent pattern is 'it v me to-inf/wh/that-clause', where me, again, stands as the direct object of the verb which is in turn further complemented by a range of clause types.The most frequent of such clausal complements is a that-clause, as in It saddens me that you have blamed yourself for so long.
A substantial portion of instances also attests a number of pattern types realizing a causative construction, predominantly with the verb make followed by a bare infinitive ('v me inf' pattern).In these instances the emotional or cognitive reaction that instantiates the 'evaluation' element is introduced by the causative, as in the following example: (44) Just thinking about someone doing that to a kid makes me want to vomit.
As was noted for the 'framed' sequence, there is a good deal of similarity across the two languages for the 'basic' sequence as well.For example, Italian instantiations of the sequence are frequently realized by the Italian clitic mi as variant of the word-form me.As mentioned in section 3.2.2above, causative constructions account for a good deal of the instances of the clitic mi: their perfect comparability with the EN me is apparent.
Consider for example, Questo mi fa stare male [That makes me feel bad] with You just make me laugh! or questo mi rende ancora più triste [that makes me even more sad] with And Kim's constant maniacal grinning makes me uncomfortable.
The same holds for instances in the 'v n' pattern, as in La vicenda delle dimissioni di Spalletti mi colpisce, ma non mi sorprende.[Spalletti's resignation strikes me, but it doesn't surprise me], which is very similar to That is the first thing that struck me.
Instances of other very frequent Italian constructions described in section 5.2 above -such as mi fa piacere/schifo/rabbia -behave semantically in exactly the same way, even though there are no direct surface equivalents for them attested in the sample.Compare for example mi fa piacere discuterlo con tutti voi [lit. it pleases me to discuss it with all of you; I'm happy to discuss it...] with and it helped me to articulate why I was so repelled.Again, the comparison seems to show that Italian and English contribute in a very similar way to the overall evaluative meaning of the sequence, even if their syntactic arrangements do not always overlap.
Table 9 provides examples of the full range of grammar patterns involved in the realization of the 'basic' sequence with EN me and IT me/mi.What emerges from the analysis of my was that its highly frequent occurrence in the corpus is not prominently connected with evaluative expressions per se.Rather, it may be better described as a reflection of the frequent reference made in blogs to facts, people and states of affairs pertaining to the everyday personal life of bloggers.Such a common tendency is attested by the frequent collocation of my with nouns such as life, husband, son, daughter, wife, parents, kids, friend, mom, children, father, family, brother, dad, house, sister etc. Cross-linguistically, it  From the collocational profile of both the EN and IT possessives, one might argue that they have to do more with 'subjectivity' intended as frequent reference to personal matters than with 'subjectivity' intended as emotive and highly opinionated involvement of writers in discussing typical blog content.
However, the sample extracted for the analysis of my (213 lines, i.e. one concordance line out of 30) attests nonetheless a relatively high incidence of evaluative instances for the node word (63 concordance lines, i.e. 29.57% of the total).In turn, this is corroborated by the relatively substantial co-occurrence of my with collocates such as favo(u)rite, opinion, mind, point, guess etc. that clearly indicate the adoption of an evaluative stance on the part of the writer (e.g.it's my current favourite song; My point is that everybody likes Greatest Hits albums).Again, a comparable set of collocates is attested for the Italian possessives: opinione, avviso, parere, pensiero [opinion], esperienza As regards the evaluative instances of my, the corpus data attest a prevalence of the 'basic' sequence (65.07%).In the most frequent pattern realizing this sequence, '(v)(prep)my n', my usually modifies a noun that acts syntactically as the verb object (or as the verb prepositional complement), as in his comments had nearly wrecked my life.
Two more patterns frequently realize the 'basic' sequence, both found predominantly in predicative constructions.In the first, '(link-v)my adj n', my introduces a predicative complement.The possessive is usually followed by an evaluative adjective (frequently favourite) which in turn modifies a noun, as in When I was a teenager I might have identified money as my most valuable resource.Alternatively, my may directly modify a noun or a noun phrase whose evaluative charge is apparent: not infrequently such noun phrases are idiomatic such as in So iPhone + Skype isn't my cup of tea.In the second pattern, conversely, my typically introduces the subject of the copular constructions.The predicate, in turn, is frequently realized by a 'v (0/)that-clause'.The pattern, 'my n (link-v)', may be exemplified as follows: My biggest complaint is that the show has a problem with wanting every storyline to be overly dramatic).
At closer inspection of the concordance lines of my, one might note that, when the 'basic' sequence is selected, an entity or process is evaluated in terms of how it affects, so to speak, the writer's 'self'.This 'self', in turn, is typically conceptualized through lexical items indicating general aspects of the personality, cognition or personal experience of the writer, such as life, expectations, attention, psyche, memories, as illustrated by ( 45)-( 46): (45) The library of music they have created is a constant source of inspiration and happiness in my life.
Alternatively, the 'self' is conceptualized metaphorically by lexical choices indicating in general body parts such as stomach, heart, face, eyes, head, etc.Some examples are: (47) i still have that anger in my head and that disgusting feeling in my stomach about this.
(48) oh, this just leaves a bad taste in my mouth (...) It is interesting to note that such typical collocations are almost entirely absent in the Italian corpus and that therefore the 'basic' sequence is very rarely attested there.This may depend on a general tendency of Italian in the use of other determiners: body parts are not usually identified by possessives, but rather by the definite article (it goes to my head vs. mi dà alla testa).
As compared to the 'basic' sequence, the 'framed' sequence is attested much less frequently in the English sample (34.92%).The 'my n (link-v)' pattern described above, where my modifies a noun acting syntactically as the subject of a copular construction, is attested as the most typical syntactic arrangement of the sequence.In this case, the predicate is constantly instantiated by a '(0/)that-clause', as in My personal view (...) is that Nuke plants can be well run (...).The only other pattern realizing the 'framed' sequence fairly frequently is 'in my n', where my modifies a noun governed by the preposition in: the whole phrase acts regularly as a sentence adverbial (e.g.Sebastian Vettel is already a superstar, in my opinion).
Perfectly comparable realizations of the 'framed' sequence are attested in the IT BLOG corpus.The sequence is instantiated through the same syntactic arrangements: the 'my n (link-v)' pattern, as in La mia opinione è invece che i lettori sono più intelligenti [My opinion instead is that readers are much smarter] and the 'prep my n' pattern, as in Ha esagerato col trucco e il vestito è pessimo a mio avviso [Her make-up is too much and the dress is terrible in my opinion].
Examples of the 'framed' sequence with my, mioand miaextracted from both the EN and IT BLOG corpora are reported in Table 10.As shown by the examples in Table 10, when selecting the 'framed' sequence, both the English and the Italian possessives collocate preferably with nouns indicating an epistemic status (e.g. view, guess, opinion, estimation, impression [impression], opinione, parere, avviso [opinion] etc.) or in general referring to the semantic sphere of cognition (e.g.mind, experience, pensiero [thought], esperienza, [experience] etc.).From a semantic point of view, therefore, the noun phrases that they participate in are perfectly equivalent in the two languages to many verb or prepositional phrases that realize 'framework' meaning elements.Relevant examples would be epistemic verbs such as think, guess, wonder, believe, credo [I believe], penso [I think], or prepositional phrases such as to me, for me, secondo/per me [to me, for me].This is an argument in favour of the concept of 'semantic sequence' as a valuable descriptive device, both in terms of its cross-linguistic comparability, and in terms of its potential for abstraction over surface variation.

Concluding remarks
The results of the analysis confirmed how deeply evaluative the language of blogs can be, which supports seeing them as virtual places for "you to have your say" (Baron 2008:99).The preliminary overview highlighted that comparable subjectivity characterized both blog corpora against general language corpora.
The close link between subjectivity and opinionated discourse in blogs was confirmed by the study of collocational profiles of first-person pronouns and possessives.The analysis of collocates showed similar trends in semantic preference.Subject pronouns (I/io), for one thing, show a similar tendency to co-select epistemic and attitudinal verbs often acting as projecting frameworks that introduce the evaluated element and the evaluation.Oblique pronouns occurring in verbal patterns show a preference for meanings implying an emotional or cognitive response on the part of the writer, so that the evaluation of an entity or a process is frequently presented in terms of the emotive or cognitive response it provokes in the writer.Possessive adjectives and pronouns are used in both corpora more often with reference to 'subjectivity' intended as reference to personal matters than with 'subjectivity' intended as opinionated involvement in typical blog content.When used evaluatively, they often construct 'framed' sequences by collocating with nouns referring to the sphere of cognition The analysis of syntactic patterns and semantic sequences highlighted (dis)similarities between the two languages.The clear asymmetry in the pronominal system, with Italian use of clitic pronouns in unstressed positions, creates divergences in key expressions (mi piace vs.I like) leading to a considerable share of instances attested for the 'v' pattern in Italian as against a preference for the 'v n' pattern in English.On the whole, however, remarkable cross-linguistic similarities were noticed in the surface realizations of the 'semantic sequences' identified in the two corpora: the most frequent patterns ('v that-clause', 'v wh/ifclause', 'v n/prep') are essentially the same.Similarly, an asymmetry in the use of determiners in the two languages could explain the different frequency and collocational profiles of the possessives found in evaluative sequences.In both languages, possessives often construct evaluation in terms of how an entity or process affects the writer's 'self' or body.However, Italian expresses possessiveness in a different grammatical way, preferring a simple definite article instead of using the explicit possessive.But apart from that, both corpora rely on a common core evaluative phraseology.
It is also important to note that some of the differences that emerge have no explanation in systematic language asymmetries, but rather reveal culture-specific phraseology and rhetorical patterns.This is the case, for example, when they highlight an Italian preference for framed sequences as against bare sequences with linking verbs, or higher prominence of dialogic as against concessive argumentative sequences.
All in all, a semantic perspective on phraseological arrangements reveals a considerable amount of lexical and syntactic variation across the Italian and the English samples, but also substantial semantic continuities in the kinds of evaluative meanings that characterize English and Italian blogs.Semantic sequences can thus be shown to be useful tools for cross-linguistic analysis of specific areas of meaning and for contrastive analysis at different levels.
), are the most typical verbal exponents of the pattern.See the following examples: (25) I am a fan of Meryl Streep & Amy Adams.(26) I am sorry to read about your friends.

Table 1 .
: Macro and micro-categories adapted from Technorati.com . Phraseology is thus identified by aligning typical recurrent surface arrangements with strings of prototypical meaning elements such as:

Table 2
shows, what is immediately noticeable in both the EN and IT keyword lists is the appearance in the highest positions of 1 st person pronouns and possessives and related verbal word-forms.More specifically, 1 st person subject pronouns score at the very top of both EN and IT keyword lists within the 50 positions (EN I -1 st position; IT mi -7 th position; io -12 th position), followed by possessive adjectives (EN my -4 th position), personal object pronouns (EN me -47 th ; IT me -17 th ), and 1 st person verbs (EN like -10 th , think -15 th , love -16 th , agree -45 th , guess -48 th ; IT credo [I believe] -29 th , concordo [I agree] -41 st , spero [I hope] -46 th

Table 2 .
First-person keywords in the EN and IT BLOG corpora (within the first 50 positions)

Table 3 .
The ten most frequent R1 collocates for mi as subjective expression

Table 6 .
'Framed' sequence with I and io/mi across the EN and IT BLOG corpora The most typical verb exponents for this pattern are piace [like] and fa [lit.It makes].Note that fa, when following mi, takes part consistently in constructions of the type 'mi + fa+ noun', often having negative meaning, e.g.mi fa schifo [it makes me sick]; mi fa rabbia [it makes me angry], or in causative constructions such as mi fa ridere [it makes me laugh].Here are some illustrative examples: (15) Sì, mi piace questo post (...) [Yes, I like this post (...)] (16) Questa è un Italia che mi fa schifo [This is the kind of Italy that makes me sick]The data also attest this pattern with mi, but in only 3.9% of the whole sample, e.g.mi dispiace moltissimo di queste brute novità [I am really sorry for the bad news].The 'v (0/)that-clause' pattern, which was dominant in the 'framed' sequence, is also attested for the 'basic' sequence both with io and mi (although accounting for only 2.4% instances of the sequence).

Table 7 .
'Basic' sequence with I and io/mi across the EN and IT BLOG corpora We have called 'argumentative' the different types of sequences that express forms of agreement and disagreement with other participants.As outlined in section 2, argumentative sequences can be categorized as 'dialogic' and 'concessive'.A 'dialogic' sequence blends together the distinctive features of the two prototypical sequences, 'basic' and 'framed'.In a nutshell the sequence is used to directly evaluate a statement for which a different source is responsible by signalling the writer's dis/alignment with it and, at the same time, to frame the writer's own reformulation or contradiction of the original statement.Here is an example taken from the IT BLOG corpus:

Table 8 .
'Framed' sequence with EN me and IT me across the corpora

Table 9 .
'Basic' sequence with EN me and IT me/mi The 1 st person possessive adjective my ranks 4 th in the keyword list.The EN BLOG corpus attests 6,395 occurrences of the word-form (43.8 pttw).Such a frequency is remarkable, especially if compared with that of the corresponding Italian possessives mio and mia (the masculine and the feminine 1 st person singular possessive adjectives), which are shown to be much less frequently represented in the corpus: there are 1,832 instances of mio (11.5 pttw) and 1,434 of mia (9.06 pttw).The other two Italian forms miei and mie, in the plural, are less frequent but nonetheless attested: 485 instances of miei (3.6 pttw) and 259 of mie (1.6 pttw).The figures show that even the aggregate frequency of occurrence does not add up to the overall English occurrences of my(25.7 vs. 43.8).

Table 10 .
'Framed' sequence with my, mio and mia across the corpora