The translation of an Italian procura alle liti into English: word-by-word rendering or functional translation?

Ad hoc corpora are considered effective to deliver native-like technical (and legal) translation. This paper explores how the translation of a system-specific document such as an Italian procura alle liti (power of attorney to appear in court) can be addressed. Translating this legal document may be problematic due to the fact that it is not recurrent in the common law system, which contemplates other types of special powers of attorney (PoAs). This paper explores whether and to what extent a corpus-based translation is possible when L2 parallel texts are difficult to find in native contexts. To this aim, a translation project with technical translators and lawyers is carried out. The participants compose an ad hoc corpus which they consult as a language reference tool. The paper findings report that the translators' corpus-based translation is mostly a word-by-word rendering of the source text, which may

sound unnatural to an expert of the field.The lawyers' translation partly deviates from source words but addresses the communicative function of the source text.The paper highlights the need for a balance between the two strategies, for example by focusing on textual functionality and/or paraphrasing, although at the cost of a reduced formulaicity.Keywords: corpus-based translations; legal translations; functional translations; legal documents; special power of attorney for a court case TRADUZIONE DI UNA PROCURA ALLE LITI: RESA LETTERALE O FUNZIONALE?

Introduction
Legal discourse is characterised by Latinisms, long sentences, syntactic discontinuity, preponderance of passive forms, formulae and formulaic expressions (Tiersma 1999;Coulthard and Johnsons 2010;Williams 2011), which must be tackled consistently in the translation process.As a matter of fact, legal language is hallmarked by lexicostructural items (Weber 2001: 16) which make it unique and different from other technical languages.For example, the literature suggests that before engaging in the translation of legal texts, translators should have a good grasp of comparative law (Orozco-Jutorán and Sánchez-Gijón 2011: 25;Giampieri 2016: 445).As claimed by Šarčević (1997), in fact, "the source legal system cannot be simply transposed into the target legal system" (Šarčević 1997: 13).
The absence of an exact correspondence between legal concepts of different legal systems is one of the most common challenges of comparative law (Rene and Brierley 1985: 14;Scott 2019: 52) and, hence, of the legal translator's profession.Also, translating into an L2 (second language) has become very common; in particular, court-related documents are increasingly translated into the translators' second language (Vigier Moreno 2011: 325;Vigier Moreno and Pérez-Macías 2022: 75).Therefore, legal translators may be confronted not only with the challenges of addressing a foreign language, but also of assessing degrees of equivalences (Prieto and Orozco 2015: 112).In these cases, "equivalence" does not necessarily imply a one-to-one correspondence (Harvey 2000: 358).Correspondence (or equivalence), in fact, must occur at both word and legal system level.Therefore, it may be interpreted as "acceptability", depending on word functions (Harvey 2000: 358) and on the communicative situation (Vigier Moreno and Sanchez 2017).In some contexts, a strict relationship between words of the source and target text is not possible due to gaps in the source and target cultures, or because of the different terminological or lexical development of the two languages (Nord 2002: 32).In legal texts, such discrepancies are particularly evident due to differences between the source and target legal system (Scott 2019: 52).
Given the complexity of the language of the law and in light of the differences among legal systems, scholars tend to propose a "functional adaptation" of the source text (Garzone 1999 and2000;Harvey 2002;El-Farahaty 2016: 479).According to Nord (2002: 32), a functional translation puts emphasis on the translation functions needed in the target text and, in particular, "in the context of the recipient culture" (Nord 2002: 32-34).Therefore, a functionally translated text relies on the communicative functions of words.Šarčević (1997) (also quoted in Harvey 2002: 180) defines the quality of legal translation on the basis of the equivalent legal effects produced by the source and target texts.In practice, according to this perspective, target texts should not necessarily follow the exact wording of the source text, as long as the source text intentions (and communicative functions) are transferred (Nord 2002: 42).This, of course, occurs at the cost of accuracy, as legal formulae may be lost in the process of translation.

Corpora for legal translations
An effective way of dealing with and finding equivalences in legal translation is by using corpora (Scott 2012;Vigier Moreno 2016).Corpora allow users to notice collocations, recurrent linguistic patterns, word usages in context and word frequencies (Jensen, Mousten and Laursen 2012: 30;Vigier Moreno 2016: 108).They help deliver fine-grained translations, which mirror the language and style of authentic texts in an L2 (Vigier Moreno 2016: 105).
Amongst the various types of corpora available (i.e., online, offline, monolingual, parallel, comparable, etc. Krüger 2012;Gallego-Hernández 2015), DIY (do-it-yourself) databases are considered particularly effective for legal translations (Giampieri 2019).They are composed by users and serve specific purposes.The terminology they contain is ad hoc (Gallego-Hernández 2015: 375-376) and highly targeted, because they are built on the basis of users' specific needs (Varantola 2002), and/or of particular translation projects (Varantola 2003).The literature argues that legal DIY corpora do not need to be large to be representative of their genres (Biel 2010;Giampieri 2019), as legal discourse is very repetitive (Bhatia, Langton and Lung 2004: 207;Biel 2010: 6).For example, powers of attorney (PoAs) are considered highly formulaic and standardized (Giordano 2019: 125).At the same time, however, they are also "locally adapted and drafted to suit specific, local realities" (ibid.).
Given the specificity of PoAs, consulting an ad hoc corpus is useful, irrespective of the amount of technical preparation it entails.As a matter of fact, the corpus compilation process is generally perceived as daunting and time-consuming (Varantola 2002: 181;Zanettin 2002: 245;Krüger 2012: 514).Also, corpus consultation skills must be developed to search for terms properly and eschew wrong interpretations or inappropriate translation options (Krüger 2012;Gallego-Hernábdez 2015: 380).These disadvantages are generally largely compensated for by several benefits.For example, corpora increase the translators' confidence (Zanettin 2002: 245;Vigier Moreno 2016: 111 and2019: 94;Giampieri 2019: 5), because they expose users to instances of "naturally occurring language" (Sinclair 1991: 171).Ad hoc corpora are highly employable in the long run, for instance by in-house translators or for specific topics or genres (Wilkinson 2006).For these reasons, it is generally accepted that the efforts of using DIY corpora do not seem so strenuous vis-àvis their benefits (Zanettin 2002: 245).

The research question
This paper addresses the corpus-based translation of an Italian procura alle liti (power of attorney to appear in court) into English.It explores whether it is possible to retrieve parallel texts from the Internet and use them to compose DIY corpora.The paper investigates discrepancies between the source and target legal systems; how they are reflected in parallel texts and, hence, in English/Italian PoAs.Also, the paper discusses possible challenges due to the difficulties of a word-by-word translation, as the language and formulae of parallel target texts may differ from the ones of the source text.
Therefore, the research questions that this paper wishes to address are the following: how is it possible to tackle the translation of an Italian procura alle liti when source and target reference documents differ greatly?How can legal translation be carried out when target parallel texts are difficult to find?
To this aim, this article describes a translation project carried out with four participants: two professional translators with sound knowledge of legal language and two lawyers with advanced knowledge of English.The participants' first language was Italian and they worked in two separate groups: the translators in the first, and the lawyers in the second.They were firstly introduced to the foundations of corpus compilation and analysis.They were given some time to practice the newly acquired skills, then they translated a procura alle liti (power of attorney to appear in court) from Italian into English via corpus consultation.
The translation observation study proposed in this paper is qualitative and tentative.
Before describing the translation project, a few words should be devoted to the challenges posed by a procura alle liti.It is an important dispute-related document on the basis of which lawyers can represent clients in a court (art.83 of the Italian code of civil procedure).Without this special power of attorney, there is no possibility for claimants/plaintiffs to start a legal proceedings, save for very few exceptions (e.g.divorce, but only in given circumstances, according to Italian Law No. 55 of 6 May 2015).
Apparently, no such document is envisaged at common law.The following sections shed light on this aspect.For the purposes of this translation observation study, the British and North-American legal systems are focused on.

Discrepancies between the source and target legal system
In Great Britain, there is currently no Act addressing powers of attorney to appear in court.The "Agents, appointees, attorneys, deputies and third parties: staff guide" (in short "Staff guide" 2013) quotes that "A PoA can be granted under: The Power of Attorney Act 1971 (...); The Enduring Power of Attorney Act 1985 (...); The Mental Capacity Act 2005.(...)" (Staff guide 2013: 7).Nevertheless, none of the above-mentioned Acts addresses powers of attorney for court cases.Moreover, the Acts do not provide for litigation, disputes or, more broadly, for any court-related situation.
Different is the situation in the United States of America, where a Uniform Power of Attorney Act is adopted in many states.Each Act envisages several possible scenarios, amongst which litigation in court is addressed.In particular, an attorney-in-fact can represent a client before any court in case of disputes, for example to: Demand, receive, and obtain by litigation or otherwise, money or any other thing of value to which the principal is, may become, or claims to be entitled; (...) Prosecute, defend, submit to arbitration, settle, and propose or accept a compromise with respect to, a claim existing in favor of or against the principal or intervene in litigation relating to the claim.
(Uniform Statutory Form Power of Attorney, California, Chapter 2: 4550) Assert and prosecute before a court or administrative agency a claim, claim for relief, cause of action, counterclaim, crosscomplaint, or offset, and defend against an individual, a legal entity, or government, including suits to recover property or other thing of value, to recover damages sustained by the principal, to eliminate or modify tax liability, or to seek an injunction, specific performance, or other relief.
(Uniform Statutory Form Power of Attorney, California, Chapter 2: 4559) Therefore, it is apparent that Uniform Power of Attorney Acts address PoAs issued in court-related cases.However, a closer look at the subject matters dealt with by the Uniform Power of Attorney Acts reveals that they establish the granting of powers to an agent (hence, not necessarily an attorney-at-law) to perform court-related and, mostly, non-court-related acts on the principal's behalf i .For example, The People's Law Dictionary describes a "power of attorney" as "a written document signed by a person giving another person the power to act in conducting the signer's business, including signing papers, checks, title documents, contracts, handling bank accounts and other activities in the name of the person granting the power".A similar description is provided on the uslegal.comportal, quoting that "specific types of power of attorneys include: Health Care Power of Attorney, Power of Attorney for Care and Custody of Children, Power of Attorney for Real Estate matters and Power of Attorney for the Sale of a Motor Vehicle ii ".As can be seen, no court-related situation is mentioned.By contrast, the subject matters addressed seem very different from the one of an Italian procura alle liti, which focuses only on a court-related representation.
Also, by analysing the American Notice of Entry of Appearance as Attorney or Accredited Representative, issued by the US Citizenship and Immigration Services, it is evident that attorneysin-fact are entitled to represent their clients, but only as far as immigration matters are concerned iii .Furthermore, American Contracts for Legal Services, or Attorney-Client Representation Agreements are mostly non court-related and mainly deal with the attorney's fee or compensation scheme iv .
In light of these observations, it can be claimed that neither in the UK, nor in the USA is a special power of attorney to appear in court actually prepared (and necessary) for court cases (see also Giampieri 2018: 17).Therefore, due to such divergences, finding parallel documents can be difficult, if not impossible.

Discrepancies in the content of source texts vs parallel target texts
As already mentioned, in order to deal with specific source terminology and deliver accurate translations, a translator needs representative parallel target documents.Therefore, the document retrieval process plays a key role.If powers of attorney to appear in court are searched for in North American websites, unfortunately no similar documents can be found for the reasons stated above.In addition, the LawInsider and Onecle legal document databases do not provide any example of such specific PoAs.
The challenges posed by the lack of reference materials is also reported and analysed by Scott (2019).In these cases, she posits that "the translator has to hunt down the correct, consistent, approved or authoritative terms, legal instruments and other documents, or authority to be incorporated in their translation" (Scott 2019: 66).As it is evident, in the case in question there are evident discrepancies between the source and target legal systems and between source and parallel target texts.Therefore, if confronted with the translation of a procura alle liti, translators should search for authoritative parallel documents most likely in non-English-speaking contexts.

Methodology: the case study
The lawyers' and translators' corpus-driven training was carried out over a three-hour online session via the Zoom platform.The training was administered by the author of this paper: a lecturer of legal English, court interpreter and legal translator.The lesson was divided into two parts and organised in the following manner: in the first part, the participants were firstly introduced to document retrieval from the Internet (i.e., they were explained how to carry out Google advanced searches to source parallel texts); they were then told how to convert the retrieved documents into text files, and how to upload them to the AntConc freeware software solution (Anthony 2023).Finally, they were explained how to consult the corpus by using the AntConc software interface.AntConc is an offline concordancer, i.e., it is a software solution which generates concordances from an offline corpus.It also lists collocations and shows word usages in contexts.In this way, it helps notice formulaic expressions and recurrent patterns of language.The participants were taught how to deal with simple and multiple word searches; collocational searches; use of the asterisk as a wildcard character, and lemmatisation.
The first part of the lesson was 1 ½ hour long.During the second part of the lesson, the participants were divided into two groups and worked in separate virtual rooms.They were prompted to retrieve parallel texts from the Internet, compose their corpus and carry out a corpus-based translation.The lecturer was always available to give advice on some technical aspects or issues regarding the document retrieval and/or the corpus consultation process.The participants could either submit their corpus-based translations at the end of the lesson, or after a few days from the lesson.

Document retrieval
This section describes the document retrieval process that the participants carried out during the online lesson.To start with, the participants were informed that the Proz.comtranslators' forum suggests "power of attorney to appear in court" as the English translation of the Italian procura alle liti v .Hence, the words entered on Google search string were "power of attorney to appear in court".For the reasons previously mentioned, no domain restriction was applied.Therefore, powers of attorney to appear in court were not necessarily sourced from British or North-American websites.After this initial procedure, the two groups worked separately in two virtual rooms: one with the translators and the other one with the lawyers.The information that follows was provided orally by the participants after the document retrieval process was over.
The documents retrieved from the Internet were carefully assessed by the participants and, if considered well-written and representative of the genre, they were downloaded and converted into text documents.The procedure adopted by the participants was, hence, the following: the search string "power of attorney to appear in court" was googled.Relevant documents (either in doc, pdf or html format) were downloaded and saved in a folder.Then, they were converted into txt files (either manually or by resorting to online converters).After converting all the documents into text files, they were uploaded to AntConc.The document reliability and representativeness were determined on the basis of the institution or body issuing them (such as embassies, chambers of commerce and law firms).For example, powers of attorney found in embassies' websites were considered reliable from both a legal and language perspective.Also, their representativeness was assessed on the basis of their similarity with the Italian procura alle liti in terms of content and formulae.More details on document retrieval and conversion is described in Giampieri (2019), as well as in Vigier Moreno (2019: 98).
At the end of the retrieval process, 17 documents were collected by the translators (3,838 word types; 20,587 tokens) and 19 by the lawyers (2,577 word types; 18,974 tokens).For the purpose of the translation project, both corpora were considered satisfactory.The literature, in fact, reports that an offline corpus composed of at least 10 documents per genre suffices (Williams 1999: 516;Giampieri 2019).Both corpora were submitted to the lecturer after the lesson was over.
After composing the corpora, corpus analysis took place.Each group continued working separately, although the lecturer was always available for technical assistance.As mentioned, the software used was the AntConc freeware concordancer (Anthony 2023).While translating, each group was asked to keep a log of the word queried in the corpus, of the relevant results or hits obtained, and of the overall search process carried out.The log was to be submitted together with the corpus-assisted translation at the end of the lesson, or within a few days from it.The next sections report and comment on the participants' corpus-based translations.

The source text and the target texts
An extract of the Italian power of attorney is reported in Table 1 below.
The corresponding translations by the lawyers and the translators are reported in Table 2 below.
Table 2.The translations of the procura alle liti (power of attorney to appear in court) by the lawyers and the translators Lawyers' corpus-based translation Translators' corpus-based translation I appoint …. to represent me during every phase and judgement of the present Trial and all other proceedings relative to the aforesaid case included the execution phase, hereby GIVING AND GRANTING unto my said attorney-at-law full power and authority to do and perform any and every act and thing whatsoever requisite or necessary or proper to be done in and about the premises as fully to all intents and purposes as I might or could lawfully do if personally present, with power of substitution and revocation, and hereby ratifying and confirming all that my said attorney-at-law shall lawfully do or cause to done under and by virtue of these presents.To receive any notice, whether in writing or verbally communicated to my I delegate Ms …., lawyer, to represent and defend me in any stage of the trial and in all of the acts resulting from enforcement and objection proceedings.I grant him/her full power and authority, including the power to appoint legal representatives, resume the trial in case it is suspended, ask for the intervention of third parties for whatsoever reason, file counterclaims, settle the suit, waive and accept waivers of the proceedings, appoint experts, collect money.I will take up domicile in the lawyer's office address in (city).attorney-at-law law firm of (city), which will then constitute as notice to me.
In the following sections, the translations carried out by the lawyers and the translators are analysed and commented thoroughly.

Analysis of the translations
In this section, a detailed analysis of the translators and lawyers' different translation strategies is accounted for.This section also tries to outline the reasons why the target texts may differ.To this aim, the source text is divided into different parts.As far as the first paragraph of the power of attorney is concerned (i.e., delego a rappresentarmi e difendermi in ogni fase e grado del presente giudizio), it is quite evident that both translations are similar.It can only be noticed that the verb "delegate", proposed by the translators, is a literal translation of the Italian delego.The verb chosen by the lawyers, i.e., "appoint", is more appropriate (see Mason and Atkins 2007: 25).In both DIY corpora, for example, "delegate" occurs 3 times (lawyers' corpus) and 5 times (translators' corpus), whereas "appoint" is used 36 and 23 times in each respective corpus.
Also, targeted web searches confirm that "appoint" is more recurrent than "delegate" (Google search string: "delegate|appoint * lawyer|attorney|solicitor" site:.govand "delegate|appoint * lawyer|attorney|solicitor" site:.gov.uk).Table 4 below reports the second paragraph of the Italian procura alle liti.In Table 4 above it can be noted that the lawyers' target words deviate from the source words, since not all source terms and phrases are present in the target text (i.e., procedimento (…) di opposizione).This was probably due to a lack of corpus evidence, as no renderings of procedimento di esecuzione or opposizione can be sourced from both corpora.The lawyers, however, proposed "execution" as a calque of esecuzione.Hence, they wrote "all other proceedings (…) including the execution phase" to generally translate tutti gli atti conseguenti nel procedimento d i esecuzione e di opposizione.It could be stated that the communicative function of the source text is somehow maintained, as "all other proceedings including" would, to some extent, comprehend procedimento di esecuzione e di opposizione.The People's Law Dictionary explains "execution" as "the act of getting an officer of the court to take possession of the property of a losing party in a lawsuit (judgment debtor) on behalf of the winner (judgment creditor), sell it and use the proceeds to pay the judgment", which is adherent to the Italian definition of procedimento di esecuzione (Art.483 of the Italian Civil Code).For these reasons, the lawyers' translation was consistent, although they specifically omitted the translation of (procedimento) di opposizione.By contrast, the translators proposed word-by-word rendering.To do so, they resorted to external sources, such as the multilingual Eur-Lex platform (this piece of information was reported in the translators' log file).Hence, to render the Italian procedimento di esecuzione e di opposizione, they accessed the online multilingual Eur-Lex platform and, on their own initiative, proposed "enforcement and objection proceedings".This phrase, however, is not found in any legal English dictionary and/or on any domain (Google search query:"enforcement and objection proceedings").The third paragraph of the PoA is rendered quite in the same way by both groups of participants.However, the verb phrase "I grant", proposed by the translators, is only mentioned once in their corpus and it is not present in the lawyers' corpus.By contrast, the formula "hereby giving and granting", chosen by the lawyers, is used four times in their corpus and three times in the translators'.These data are confirmed by targeted Internet searches.For example, by querying "I grant (…) power and authority" and "giving and granting (…) power and authority" in the Onecle database, no hits are found in the first case, whereas a few hits are obtained in the second one (sample search query: "giving and granting * power and authority" site:.onecle.com).The same occurs if the LawInsider platform is queried.Therefore, the lawyers' terms are apparently slightly more frequent.Table 6 reports the fourth and fifth paragraphs of the PoA.It is remarkable that the source content of sections 4) and 5) are actually neglected by the lawyers, who include other corpus-driven formulae.The translators' version, by contrast, is more adherent to the source text words.It may, nonetheless, be debatable as the formulae proposed are not always coherent with the context.For example, chiamare terzi in causa is rendered as "ask for the intervention of third parties", as the translators found "intervention of third parties" in the corpus.By reading the concordance more carefully, however, it is noted that "intervention" is not the right target term, as the concordance reads: "authorize the said Attorney to (…) withdraw as a consequence of intervention of third parties".Therefore, the target context is different from the source one.In this respect, an equivalent of chiamare terzi in causa is "to summon third parties".In the Casemine US case-law database, for example, it is possible to read "it says that the court 'may' summon third parties and 'may' notify interested persons to appear" and "the scope of an arbitrator's authority to summon third parties to appear and provide evidence at arbitration" (search query: "to third parties"/1).Therefore, although the translators tried to render the words appropriately, misinterpretation of the source/target words occurred.The lawyers, as can be seen from Table 6 above, do not translate the source text, but propose entire different words and phrases.They probably found them relevant as they frequently occurred in their corpus.For example, the expressions "do and perform", "as I might or could (lawfully) do", "hereby ratifying and confirming", and "do or cause to be done" are particularly recurrent in the lawyers' corpus.Nonetheless, these parts are redundant and deviate from the source text.Table 7 sheds light on the last paragraph of the PoA.

6)eleggendo domicilio presso il suo studio in (città).
To receive any notice, whether in writing or verbally communicated, at my attorney-at-law law firm of (city), which will then constitute as notice to me.
I will take up domicile in the lawyer's office address in (city).The lawyer's rendering of eleggere domicilio (i.e., "serve as notice to me") is found once in the lawyers' corpus, but it is not present in the translators' corpus.The exact opposite occurs to the phrase "take up domicile".However, if the phrase "as notice to me" is searched for in targeted web domains, it is possible to notice that it is used in a large number of documents (search query: "as notice to me" site:.gov),where the verbs "constitute" or "serve" precede it.In addition, the phrase "as notice to" clearly brings to the fore a client-attorney relationship (sample phrase: "and consent that an Order served upon my attorney of my suspension shall serve as actual notice to me").On the contrary, the formula "take up domicile" is mentioned much fewer times in targeted domains (search syntax: "take up domicile" site:.gov)and the phrases obtained do not refer to any client-attorney relationship (results phrase: "testamentary declaration of United States domicile evidences a lack of intent to take up domicile in Ethiopia").
To confirm this, searches in the Casemine portal are carried out.If the phrases "as notice to" and "power of attorney" are queried, the following interesting phrase is obtained: "with a broad power of attorney, registered notice to the attorney or accountant may also serve as notice to the taxpayer".In the same line, it is also possible to read "a taxpayer may also designate the address of his representative as that to which any deficiency notice should be sent".Therefore, it can be inferred that "notice to the attorney may/shall serve as notice to" and/or "designate the address of (my attorney) as that to which any notice should be sent" express the same meaning of the source phrase eleggere domicilio presso il suo studio.By contrast, the expression "take up domicile" is mentioned only in a context unrelated to a power of attorney (Casemine sample phrase: "to establish a residence or to take up domicile in the State of Nevada").The same conclusions are drawn if the Onecle database is consulted (search queries: "serve|constitute as notice to" site:.onecle.comand "take up domicile" site:.onecle.com).In the first case, a few hits are obtained (e.g., "this letter will serve as notice to you"), whereas in the second, no results are found.

Results
As far as the two corpora are concerned, it is evident that the translations differ substantially in content and length.In particular, as highlighted in the tables above, the translators tried to remain as adherent as possible to the source text.Also, they performed online searches when confronted with expressions or formulae they could not find in the corpus.It can be noticed that the translators mainly provided a word-by-word rendering of the source text, at the risk, however, of proposing unnaturally sounding language or wrong rendering (e.g."ask for the intervention of third parties" to translate di chiamare terzi in causa).The translators naturally focussed on each individual word, perhaps without considering the overall discourse function (Harvey 2000).
The consequence the translators would face, should their translation be proposed to a client, is that of proposing a text which is not naturally sounding and, if read by a native speaker of the target language or a competent lawyer in the target legal system, may sound awkward in some part.
As regards the lawyers, it is apparent that they decided to deviate from the source text to propose a document more adherent to the style and contents of the powers of attorney composing the corpus.As a matter of fact, they wrote formulae which are not present in the source text, such as "to be done in and about the premises as fully to all intents and purposes as I might or could lawfully do if personally present", and "hereby ratifying and confirming all that my said attorney-at-law shall lawfully do or cause to done under and by virtue of these presents" (see Table 6 above).These phrases are evidently redundant.Therefore, they are non-adherent to the source text wording.Also, the lawyers omitted the translation of some portions of the source text (see Table 4 and Table 6).Therefore, in the lawyers' work, the translation intent shifted from "equivalence" to "acceptability" (Harvey 2000: 358), as they tried to tackle the specific communicative situation (Vigier Moreno and Sanchez 2017), although too broadly.
The consequence the lawyers would face, should they submit their translation to a client, is that they would propose a text which is not completely adherent to the source text, but presents some "inventive" parts.Therefore, their translation might be rejected.
In light of these results, a right balance between word-byword rendering and "creative" translations should be found.

Discussion
The paper raised questions concerning the difficulties encountered when dealing with different legal systems and with legal documents which are almost non-existent in the target language and legal system.It was aimed at exploring the quality of a corpus-based translation of an Italian procura alle liti (power of attorney to appear in court), when no target language documents are available, due to differences in legal traditions.
As discussed, the translators performed the task by addressing individual words and by consulting the web to tackle concepts or words which were not present in the corpus.The lawyers apparently focused on the communicative function of the text as a whole, and delivered a translation which deviated from the source text, sometimes with substantial omissions.However, the functional aspects of the source text were rendered (albeit not mirroring the exact wording of the source text).This occurred at the cost of accuracy, as many legal formulae were lost in the process of translation.
In light of the two approaches followed, it could be argued that a right balance should be found between the two translation strategies.For example, a functional translation (Garzone 1999 and2000;Nord 2002) as well as paraphrasing (Šarčević 1997: 231ff) could be considered as valid alternatives to the strict rendering of the source words (at the cost of sounding unnatural) and copycatting parallel target phrases (at the cost of adding unrelated parts or omitting others).This could be particularly useful when legal principles differ and no equivalences are found.For instance, the phrase "to receive any notice (…) to my attorney-at-law law firm of (city), which will then constitute as notice to me" was considered more adherent than "I will take up domicile in the lawyer's office address in (city)" to render eleggendo domicilio presso il suo studio.The lawyers' word choice, in fact, was more recurrent in targeted domains and in the case-law.In this way, the functional aspect of the source text was addressed.Another example of paraphrasing or using a functional approach, was translating tutti gli atti conseguenti nel procedimento d i esecuzione e di opposizione as "all other proceedings (…) including the execution phase".Given that lawyers did not probably find any rendering of procedimento di opposizione, they decided to resort to vagueness (i.e., "all other proceedings including") as a strategy.In this way, the communicative intent of the source text was somehow conveyed.
Finally, this study highlights the fact that other ad hoc resources must be consulted when DIY corpora are particularly small and/or the subject-matter is challenging due to discrepancies in the legal systems or traditions.Such additional language tools could be, for example, monolingual dictionaries, the case-law and targeted domains via advanced search techniques.

Conclusions
The intent of this paper was to highlight how, owing to different legal systems and traditions (Rene and Brierley 1985), legal translations are sometimes difficult to tackle and native parallel texts are hard to find.In these cases, users should resort to reliable non-native texts.The aim of this paper was also to shed light on the importance of delivering functional translations and of paraphrasing (Šarčević 1997: 231ff) when the source and target legal systems are different, as well as when discrepancies between the source and target texts do not allow wordby-word rendering.In this respect, this paper presented different approaches in translating a system-specific document, and showed that keeping the original functional communication can sometimes be opted for instead of a literal translation.This strategy, however, can be followed provided that neither omissions of source phrases occur, nor redundant target phrases are proposed.Also, it should be reminded that this approach may inevitably lead to reduced formulaicity, as specific language conventions may remain unaddressed (Mauranen 2007: 97).Therefore, translators should try to find the right balance between the need for precision and authenticity.In addition, if corpora are particularly small, consulting referenced materials, such as the case-law or targeted domains could be helpful.
In answering the first research question ("how is it possible to tackle the translation of an Italian procura alle liti when source and target reference documents differ greatly?"), this paper showed that it is possible, as long as translators do not focus their translation work on exact words, but on the general meaning and functionality of the phrases and, hence, of the text as a whole.In addressing the second question ("how can legal translation be carried out when target parallel texts are difficult to find?"), the paper highlighted that retrieving reliable (i.e., institutional) documents written by non-native speakers of the L2 can be an acceptable solution.In addition and in support of the document collection process, authoritative L2 native speaker contents could be accessed, such as case-law platforms and/or contract and agreement databases.
The limits of this paper lie in the reduced size of the corpora.Further research could explore whether a similar study could be carried out with subject matters whose parallel target texts are more accessible.
So far, academic papers focusing on corpus-based translations of procura alle liti have not been carried out.Therefore, this paper presents fresh insights into the field of translation studies.Hopefully, this initial analysis will be followed by further or similar investigations in the same field and/or in other genres.For example, native speakers of English could be prompted to translate another power of attorney for a court case by consulting an ad hoc corpus.It would be interesting to verify to what extent their translation solutions differ from those proposed by the translators and lawyers of the present study.

Table 3 .
Table 3 below reports the first.Analysis of the source text and of the related translations (first

Table 4 .
Analysis of the source text and of the related translations

Table 5 .
Analysis of the source text and of the related translations

Table 6 .
Analysis of the source text and of the related translations (fourth and fifth parts)

Table 7 .
Analysis of the source text and of the related translations (sixth part)