Religious Formulae in Historical Lower- Class Patient Letters

This article examines the use and functions of religious formulae in historical lower-class letter writing. The data analysed are taken from the Corpus of Patient Documents (CoPaDocs), a new corpus of 19thand early 20th-century texts written by patients from German psychiatric hospitals. An illustrative investigation of the occurrence and usage of religious formulae allows us to differentiate between explicit and implicit uses and to discuss instances of variation and modification. The functional analysis exhibits predominantly argumentative functions of religious formulae, but also parallelisation, expression of a shared ethos and a text-structural function. Finally, we discuss to what degree the formulae found are corpus specific, for example, resulting from the religious delusions of some patients, or whether they could also be connected to the idea of a cross-linguistic repertoire of formulaic writing.


Introduction
Den gemeinen Mann in Deutschland hört man bei solchen Gelegenheiten, wo er sich besonders feierlich erheben möchte, häufig in den Bibelton verfallen. Bei dem Volke ist die Bibelsprache zur Sprache des täglichen Lebens, zur Sprache des innigeren Familienverkehrs, ja selbst zur Sprache der Liebe geworden, denn selbst in die Liebesbriefe des Volkes, gerade je ernsthafter es der ungeübte Schreiber meint, tritt um so leichter diese biblische Färbung ein […]. (Mundt 1844: 159-160) 1 In our paper, we discuss the results of a case study on the use and functions of religious formulae in 19th and early 20th century letters written by patients from psychiatric hospitals. In the past, the focus of research on formulaic language has mainly been on standard languages and close-to-standard varieties. As a consequence, we see a lack of studies based on text types from non-standard domains, even more so with regard to historical varieties. With the rise of new approaches to the history of language, such as the concept of a language history 'from below' (Elspaß 2005), private letters written by less-experienced writers are now regarded as a particularly valuable source of linguistic data. This is due to the writers' limited literacy: while they obviously do have a command of basic writing skills, they are usually not familiar with producing conceptually written language, or the language of distance in the sense of Koch and Oesterreicher (1985). Such letters hence provide a good opportunity to reconstruct the non-standard varieties of common people and to get "as close as we can […] to authentic oral registers" (Elspaß 2012: 45). This also holds for investigations into formulaic language. However, studies on the use of formulaic language in historical letters written by inexperienced writers are quite rare. This may be explained by the -already mentioned -dominant paradigms in the research on formulaic language, but also by a lack of available texts. Despite an increase in schooling and thus writing competence during the 19th century (Elspaß 2005: 76), lower-class people did not usually have a reason to write, either in their private lives or in their professions as farmers, craftsmen, etc. And if they ever wrote to each other, their texts would not usually be transmitted to us, but kept by their families and thrown away after some time.
Situations of separation, however, increased the amount of text production by lower-class people. Emigration played an important role in the 19th century and we have evidence of a large number of letters written by emigrants, especially emigrants to North America, sent home to their families (Elspaß 2005). Soldiers were also far away from their families and wrote home (Langer 2013). Another context of separation in which letter-writing became relevant were the psychiatric hospitals that were established systematically throughout Europe during the 19th century. After their hospitalisation, writing letters permitted the patients to continue personal communication with their spouses, relatives and acquaintances. Hence, patients at these institutions often wrote letters home, but also to the doctors of the hospitals and other recipients. These letters, however, were often not sent out but censored and put into the patients' files as proof of their mental illnesses (cf. Schiegg 2015). The research project 'Flexible Writers in Language History' in Erlangen, Germany, is currently compiling the first corpus of about 2,000 of those letters and other texts written by 19th-and early 20th-century patients from German psychiatric hospitals. 2 This corpus is the empirical basis of our study.
Taking into account the results of other studies, we expect the patients to often resort to formulaic language that, among other things, could support their text production and strengthen their arguments (Elspaß 2005: 174). Furthermore, as religion was a central part of 19th-century everyday life, particularly for rural people from the lower classes, we also assume that their texts will occasionally adopt religious elements (cf. the introductory contemporary quotation). The central purpose of our paper is thus to analyse the usage and functions of religious formulae in the CoPaDocs letters. Since research in the field of 'language in religion' generally shows a lack of studies with a pragmatic orientation (Lasch and Liebert 2014: 478), our study might also be relevant from this point of view.
The paper is structured in the following way. In section 2, we delimit our concept of a religious formula and summarise the results of existing research on religious formulaic writing by lower-class people. In section 3, we describe our corpus and the methods used to identify religious formulae in the letters. Sections 4 and 5 discuss the results of our case study. Section 4 gives an overview of the occurrence and usage of religious formulae in the corpus with regard to explicit versus implicit uses, to variation and creative modifications. In section 5, we analyse different functions of religious formulae in the CoPaDocs letters, with a focus on argumentative contexts. The paper ends with a summary of our main findings against the background of existing research on 19th century private letters.

Religious Formulaic Language in Lower-Class Letter Writing . Religious Formulaic Language
Before analysing the uses and functions of religious formulae, it is essential to define our concept of the term religious formula. Particularly when dealing with historical data, it is helpful to use a broad definition of formulaicity (Filatkina 2018: 164). We therefore take the often-cited working definition by Wray (2002: 9) as a starting point and consider formulaic any "sequence, continuous or discontinuous of words or other elements, which is, or appears to be, prefabricated: that is, stored and retrieved whole from memory at the time of use, rather than being subject to generation or analysis by the language grammar". Expanding this approach, we additionally integrate the view advocated by Stein (1995: 57-58) that pragmatically fixed one-word-utterances can also be regarded as formulaic. The scope of formulaic language thus ranges from one-word-utterances with a specific pragmatic function over multi-word units and fixed sentences to formulaic texts (Filatkina 2018: 30).
A definition as such, however, does not permit a reliable extraction of formulaic material from a text corpus (cf. Wray 2009). To identify formulaic language in texts, we also need an operationalisation of the defining features, i.e. independent criteria that mark a given sequence as prefabricated and hence formulaic. In this context, Wray and Namba (2003: 28) offer eleven diagnostic criteria that can be derived from peculiarities of formulaic units in terms of the four dimensions form, meaning, function, and provenance. Most of these criteria, however, cannot be operationalised objectively either. Rather, it is their function to assist and support the researcher's introspective judgement, to help articulate the basis of his or her intuition and to provide insight into possible biases (Wray 2009: 41). The final decision on the status of an item, however, remains subjective and based on the researcher's intuition.
The second question to be addressed is under which circumstances a given formulaic item can be regarded as religious. A variety of approaches (semantic, discursive, intertextual etc.) could be chosen here. 3 For the purpose of our study, we will focus on the provenance aspect, regarding formulaic items as religious if they are associated with a particular religious source or context. Possible sources here include texts from the scriptures, but also popular textual sources from other religious contexts such as prayers, hymns, liturgies, rites, and texts from the catechism. From a theoretical point of view, the narrowing of the aspect of provenance is certainly simplifying; this reduction, however, is a necessity for a case study like ours.
To be classified as a religious formula in our article, it is not essential that the wording used in the patient letter should be completely identical with an available source text. We can assume that writers did not usually have the particular text at hand when writing. They normally quoted from memory rather than from the original text. Consequently, we also include cases in which the formulation 3 According to Lasch and Liebert ( : ), characteristics of the religious domain are distinct objects, a reference to transcendence, and a more or less elaborated metaphysics. itself is novel but clearly derived from or associated with something that is formulaic in its own right. 4 Thus, we have a continuum of usages, ranking from word-for-word formulae, fully identical with an available version of the original text, through formulae with individual deviations, to allusions and paraphrases, i.e. intertextual references which do not try to recapitulate the original text but merely refer to the content of a certain religious formula (cf. Lange 2012: 100-101; Bartolini 2012). In all these cases, the intertextual character of the formula can either be made explicit or remain implicit (see section 4).

. Research Overview: Religious Formulaic Language in Lower-Class Letter Writing
Formulae are and have been essential parts of letters. In the nineteenth century, letter writing manuals enjoyed great popularity, and forms of address as well as closure formulae were considered the 'touchstone' (Baasner 1999: 23) of any official letter. These manuals, however, do not seem to have had significant influence on lower-class writers (Elspaß 2005: 195). 5 Nevertheless, comparative research on textual sources from different European countries has revealed that lower-class writers in particular relied heavily on textual routines and formulaic patterns acquired from model letters and other texts (Elspaß 2012: 45). Rutten and Van der Wal (2013) have identified a strong correlation between the frequency of formulae and the social class of a writer. They explain the increased use of formulae by 'ordinary people' by the fact that the use of formulae was "convenient to lesser-skilled writers" (2013: 45), for whom writing was not an everyday practice. It helped these writers solve communicative problems in the written code (Elspaß 2005: 192). Several of the formulae that research has identified in lower-class letter writing contain religious elements. While Barton (1975: 5) states that most of his analysed emigrant letters written by Swedes in America "have in fact little of interest to relate" and consist, among other things, of "religious platitudes", more recent research has identified the textual functions of religious formulae beyond the mere informational function. In her analysis of 19th-century Scottish correspondence, Dossena (2013: 57) considers the religious element in letters to be a "highly meaningful device for the expression of involvement and psychological proximity between participants". Thus, religious formulae can be "network-reinforcing strategies" (2013: 50), by which a "writer expresses shared ethos with the recipient" (2013: 54). Similarly, Rutten and Van der Wal (2012: 181) have identified the 'Christian-ritual function' as one of the main functions of epistolary formulae in 17th-and 18th-century Dutch correspondence. Such formulae "usually place the writer and/or the addressee under divine protection, thereby manifesting the writer's religiosity". The authors connect this with Coupland's (2007) sociolinguistic theory of stylisation and interpret religious formulae as stylisations of "ethical reliability" (Rutten and Van der Wal 2012: 181). Elspaß's analysis of 19thcentury German emigrants to America has revealed a large number of biblical quotations and proverbs, religious formulae and commonplaces that justify, excuse or support an argument. Writers trust in the approval of general religious 'truths' valid in their communities to avoid opposition. Thus, religious formulae can provide various aids in the formulation of letters (Elspaß 2005: 175-181).

Religious Formulaicity in the CoPaDocs Letters: Corpus and Methods
The Corpus of Patient Documents currently (March 2019) consists of about 2,000 texts written by more than 200 different writers born in the 19th century. Most of the writers can be classified as 'ordinary people' without much writing experience. They usually followed manual professions and had only attended primary school. Social data about these writers such as their professions, their provenance and family background as well as their financial circumstances can be retrieved from the patient files. The medical diagnoses given in the files, however, need to be treated with caution and can definitely not be equated with modern medical diagnoses. The great majority of their texts are letters, but in the patients' files, we also encounter other text types, such as autobiographic texts, poems, and diaries. Some patients wrote religious texts that can be identified by their title, for example: 'prayer before lunch' and 'evening prayer' (ans-34 Johann G. A. 6 ) or 'prayer against blasphemous thoughts' (kfb-80 Hans A.). There is a 'sermon', written by the baker Franz O. (kfb-518), 'confessions' by the tailor Johann V. (kfb-775) and the miller Josef W. (kfb-2058) and even a 'prophecy' leading to a supposed apocalypse in 1888 by the nailer Johann H. (kfb-789). Johann H. was suffering from religious delusions, and a large number of references to religion are found, particularly in texts by patients with such a diagnosis. Apart from one example from a diary (see 14), however, in this article we only consider religious formulae that appear in letters.
A characteristic feature of the corpus and of historical letters in general is a large number of spelling variations, not only between different writers but also within one and the same person. Inexperienced writers in particular show this kind of variation, which is influenced by their degree of education, their regional varieties and by individual stylistic factors. The absence of a consistent orthography in our texts -along with a supposedly low frequency of individual formulae -obviates the application of corpus driven methods to automatically extract formulaic language, especially since corpus compilation is still in progress and lemmatisation has not yet been conducted. In addition, the CoPaDocs corpus currently comprises about 750,000 words -a size too large to be investigated on a manual basis alone. Consequently, the methodological approach used here does not aim at an exhaustive extraction of all religious formulae. However, this is not the ambition of our study in any case. Instead, we aim at an illustrative analysis of the uses and functions of religious formulae in the CoPaDocs letters. For this purpose, we applied a mixture of different methods to extract a sufficiently large sample of relevant text passages.
Our first approach was lexically based. We searched the corpus for single lexical items typical in religious contexts, assuming that these items should also frequently appear in formulae of religious provenance. Such key word searches were executed for the expressions Amen and Kreuz ('cross'). In both cases, we integrated a large number of spelling variants. The resulting hits were manually reviewed in order to identify those examples representing religious formulae in the sense of this paper. Afterwards, we also conducted key word searches for the items Bibel ('bible'), sagen, and sprechen ('say', including their morphological variants). The rationale was that some writers often use source-indicating routine expressions based on verba dicendi. By searching for the central constituents of such formulae, we were indeed able to identify a large number of explicit references to religious intertexts.
Besides these semi-automatic approaches, we examined the letters of one writer with regard to religious formulae using a traditional paper and pencil strategy: Martin B. (kfb-1621). Being a day-labourer and suffering from delusions, he is a prototypical example of an inexperienced writer who uses a lot of religious language in his letters. What is particularly interesting here is that Martin B. writes not only the relatively large number of 14 letters, but also letters to different addressees (his wife, his aunt, his brother-in-law, the mayor of his home town) in different social groups. Hence, in his letters he uses different registers (cf. Schiegg 2018), while the general function of his letters is relatively constant: appeals to be released from the hospital. We thus have plenty of intraindividual variation here that also appears in his formulaic language.
A difficulty in comparing the formulaic items with their possible sources results from the large number of Bible versions and other religious texts that were common in the 19th century -and of course the writers' own variations and modifications. It was not possible to find one 'best' Bible version for all cases, so we decided to quote different versions to achieve a maximum degree of accordance with the wording in the letter. With regard to the English translations of the formulae, we quote from the online-version of the Standard King James Bible (KJV).

. Explicit vs. Implicit Usage
In this section, we will illustrate different ways of how religious formulae appear and are integrated in the patients' letters. We can make a basic distinction between formulations which are explicitly marked as a religious formula on the one hand and implicit references to religious sources on the other.
In the letters, we find a variety of strategies used to explicitly mark the status of a text passage as a (religious) formula. In a number of cases, formulae are set apart from the rest of the text. Sometimes, this is realised by the arrangement of the page, e.g. by paragraphing and visible spaces between the body of the text and the formula and/or by explicit indications of the source text. This is illustrated by examples (1) to (4).
The farmer's daughter Magdalena S. (kfb-450) was suffering from religious delusions. Both of her letters, one to her sister (22.11.1857) and one to her friends (Christmas Day 1857; see example 1 and figure 1), begin with a biblical quote directly after the date and address, which span over 4 and 3 lines respectively and close with an identification of their sources. In each letter, the formula appears in an individual paragraph. The letter to her friends even has an empty line after the formula. Magdalena S. presumably copied both formulae from a Bible or another religious text that she had access to. A hint for a printed model text rather than her memory is the full capitalisation of 'HERRN' ('the Lord'; line 3), a common practice for nomina sacra in printing (Nübling et al. 2017: 265): The complexity and length of this quotation, the adherence to contemporary writing conventions (the capitalisation of HERRN 'the Lord') and the correct quotation of the source makes us assume that Magdalena S. did not cite this sentence by heart, but copied it from a written source. A search through the texts between 1800 and 1857 available as digital versions in the catalogue of the Bavarian State Library produces hundreds of results for this formula. None of them, however, has the exact wording used in the letter, so we do not actually know Magdalena S.'s textual version. Nevertheless, a fairly good correspondence can be found, for example, in a 48-page religious consolation book for the sick (Kindler 1848). 8 Both formulae are only slightly connected to the content of the letters, in which Magdalena S. complains about her situation in the psychiatric hospital, while the formulae convey a more positive mood: the temporality of all misery.
Other strategies to highlight formulaic language are punctuation (especially quotation marks), indications of the source, metalinguistic introductions and any combinations of these features. Such forms of highlighting, however, are rather rare in the CoPaDocs corpus, which consists mainly of texts by inexperienced writers. The rare examples we do find in the corpus generally come from more educated writers and mostly appear in more recent texts (beginning of the 20th century) at a time when orthographic standardisation had begun. Example (2) illustrates these strategies. In a letter to her mother, the deaconess Katharina S. uses a direct quote from the Psalms: (2) Deaconess Katharina S. (kfb-3085), letter to her mother (01.05.1925): Ds. Wort des Psalmisten: "Freuet euch mit Zittern", möchte ich Dir zurufen.
The formula is syntactically integrated into the context, but optically separated by two quotation marks, a colon at the start, and a comma at the end. In addition, the writer names the source, though not the exact place in the book (Psalms 2:11). She probably quotes the short sentence by heart, remembering it because of her religious profession. Different languages and/or scripts, e.g. blackletter vs. Latin script (cf. Schiegg and Sowada 2019), can also be used to highlight religious formulae. This practice mainly occurs in contexts of code switching when religious formulae appear in foreign languages, especially, but not only (see example 4), in Latin. Lower class writers did not usually have any knowledge of foreign languages, so they had to transcribe phonetically from memory. The farmer Georg W. provides such an example (3; see figure 2) in a letter to his parents, where he closes with a Latin sentence that he probably remembered from the Catholic church services held in Latin before the Second Vatican Council in the 1960s. After his signature, Johann Georg W., he finishes his letter by reciting the priest's absolution together with the response Amen, a common closure formula that we often found in patient letters (see section 5.2). Given that the writer does not have any command of the Latin language, the quality of his phonetic transcription is all the more remarkable -and a strong indication of a high language competence, which is generally characteristic of his letters (cf. Schiegg 2015: 155-177  Quoting Christ's final words, Maria R. draws a parallel between the last hours of Christ and her own poor health. She dies in the psychiatric hospital ten months after writing this letter. Using a religious formula to equate a biblical element with the patient's own situation is a practice that we commonly find in patient letters, particularly related to the passion of Christ. The words following (line 2) are also written in Latin script, but do not seem to be connected to the rest of the formula. We can recognise some elements of the Latin language (sum -'I am'; nattum 'born (neuter)'; est 'he/she/it is') that Maria R. transcribes phonetically from memory (nattum usually has only one t), probably without knowing their precise meaning. These words do not seem to make much sense and Maria R. presumably inserts some text in the religious language of Latin to place emphasis on her previous formula.
While the examples cited so far provide explicit indications for the status as a (religious) formula, in the majority of cases in our corpus the use of the religious formula remains implicit. Such an example is the following: ). As part of the prayer, this formulation is extremely fixed and clearly an instance of formulaic language. In the letter, it is completely integrated into its syntactic context and is not marked in any way as a formula. This also holds for the other fixed phrases in the same passage: weist niht wi noh was ('he neither knows how nor what') and dan Baist di Maus Eher kainen Zwirn niht ab (literally: 'then, the mouse rather does not bite off no thread'). It is obvious that such unmarked uses of religious formulae are rather difficult to identify in texts. At the same time, the fact that the writers often see no need to explicitly mark religious formulae can be regarded as an indication of their status as highly familiar and commonly used units.

. Variation and Modification
As already mentioned in section 3, formulaic language in the CoPaDocs letters reveals a high degree of variation. Example (5) illustrates peculiarities in several linguistic areas that can be explained as diastratic, diatopic and individual variation. In the (non-religious) formula dan Baist di Maus Eher kainen Zwirn niht ab, for example, we find a double negation in kainen … niht, which is a syntactic structure that was highly stigmatised in the 19th century and only used in the written language of less-schooled writers (Elspaß 2005: 276). This construction is also diatopically marked, as it appears predominantly in southern German (Elspaß 2005: 281). In the same formula, we see the lexeme Zwirn ('twine'), while the formula usually has the near-synonym Faden ('thread') (Duden 2013: 498). Individual variation (cf. Barz 1995) also appears at the orthographic level, where we find individual characteristics for this writer, for example the phonetic spelling <ai> for /ai/ (Baist, kainen) instead of the conventional <ei> (see also examples 6a and 6b). 10 Individual variation can particularly be observed when writers repeatedly use formulae. As Martin B.'s letters all have the same intention -being released from the hospital -their contents overlap and some of the (religious) formulae appear more than once:
(ELB) ('God judgeth the righteous, and God is angry with the wicked […]'). All three instances have a common general inventory of lexical material, but at the 10 The <ai> spelling for the former /ei/ diphthong was a typical Upper German spelling in the 16th and 17th centuries (Schmidt 2013: 370). Afterwards, however, Upper German texts followed those from the rest of the German speaking area and switched to <ei> (name spellings remained <ai>). Thus, we can classify the <ai> spelling in 19th-and 20th-century letters as individual rather than diatopic variation. same time show variation. Lexical variation can for example be observed in the particle sehr ('very') that intensifies the modifying adjective streng ('severe'), but only appears in the two letters to his wife (a and b) and not to the mayor (c). Similarly, the spelling variant <ai> instead of the conventional <ei> in ain and Gaist appears in all the five possible positions in the letters to his wife, but in neither of the two ein to the mayor. These differences between the first two formulae may appear by chance or because of the gap of almost ten years between Martin B.'s letters to his wife and that to the mayor. It may also result from the different addressees. Martin B. in general shows less conventional spellings in his private letters (cf. Schiegg 2018), and -as we will see in section 5.1 -the formulae to his wife have a more intense and threatening character, which is indicated by the intensifier sehr. Therefore this kind of variation may also be explained by his application of different linguistic registers. The phonetic spelling schtringer in (6a) supports that hypothesis. It appears only in one of the letters to his wife, and shows the palatalisation of /s/ in initial position before a consonant that is usually spelt <s> and not <sch>, as well as an assimilation of /e/ to /i/ resulting from the following velar consonant /ŋ/, which Martin B. writes as an <i>. If we assume that Martin B.'s variation here is intentional and has a communicative purpose, we approach the interface between variation and modification of formulaic language.
In phraseological research, modifications are usually defined as occasional, intentional and context-bound variations of formulaic items (Pfeiffer 2018: 51). The tendency to creatively modify formulaic language is often regarded as a characteristic development of media texts in the second half of the 20th century (e.g. von Polenz 1999: 381). Our corpus of letters by inexperienced writers at the turn of the 19th century, however, shows that this practice can also be found earlier and in other text types. A variety of religious formulae are indeed modified for their specific context. This will be illustrated in examples (7) to (9).
At the end of the following text excerpt (example 7), the tailor Pius G. uses the canonical form of the religious formula ans Kreuz mit ihm ('crucify him'), recorded in all four gospels (Mt 27:22-23; Mk 15:13-14; Lk 23:21; John 19:6, EU). These words are shouted by the crowds to demand that Jesus be crucified after his trial in the praetorium before Pontius Pilate. What is interesting here is the context of the formula, namely the continuous equation of the situation of the writer with the passion of Jesus Christ. A couple of lines before, the writer reports his own trial before a Bavarian court of justice. In his view, the trial was nothing but a show trial, ending with his condemnation and a sentence of acquittal for two fellow defendants. Already from a content point of view, the parallels to the passion of Jesus are evident. However, the parallelisation is also realised and even intensified by linguistic means, namely by modifying the formulaic item ans Kreuz mit ihm to ins Narrenhaus Haus mit Ihm. The writer substitutes the constituent Kreuz with Narrenhaus ('madhouse'), thus not only comparing his situation with the passion of Jesus but also establishing a link between the concepts of cross and 'madhouse'. This conceptual identification is quite frequent in the letters investigated (see section 5.1.2).  (Duden 2017: 382). Again, the writer does not use the formula in any canonical form, but rather adapts it to the specific context. In his letters, Martin B. often claims that his wife is possessed by an evil spirit, who has pushed aside her once good character. In his view, it is this evil spirit who prevents her from getting him out of the hospital. To express this view, he modifies the formula in a twofold manner: firstly, he substitutes the component Gott ('God') with der Böse Geist ('the evil spirit'). Secondly, he expands the formula with the comparative structure oft mer als der gute Geist ('often more than the good spirit'). Examples like this show that even less-experienced writers produce quite complex modifications to support their communicative goals. The modification in (9) also adapts a biblical quotation to the context of the letter. The highlighted passage refers to John 16:20: ihr werdet traurig sein, aber eure Trauer wird sich in Freude verwandeln (EU) ('and ye shall be sorrowful, but your sorrow shall be turned into joy'). By substituting the possessive article (eure vs. meine), inserting the relative clause (welche ich gehabt habe) and permutating the constituents joy and sorrow, the formula is adapted to the writer's situation: shortly before, he had been released from the psychiatric hospital -only to find himself rehospitalised after a couple of days. So his joy about being released has indeed turned into sorrow again -the biblical promise of salvation has not come true but has rather been turned upside down for him.

Functions of Religious Formulae in Patient Letters . Argumentative Functions
From a text pragmatic perspective, the patients' letters are polyfunctional, serving appellative, contact-orientated and also informative functions. The dominant function of most letters is directive-appellative: the patients request the addressees to do something for them -often their release from the hospital, but also a visit or a parcel from home. This request is usually supported by the use of argumentative textualisation patterns, shaping major text parts and even whole letters. From the perspective of this paper, it is interesting to investigate which role religious formulae play in these contexts.
The typical argumentation in patient letters is deontic, with the thesis/conclusion being I [the writer] should be released from hospital and/or you [the addressee] should help me be released from hospital. To support this, the writers often claim that they were mistakenly hospitalised in the first place since they are not actually ill. The addressees are regarded as responsible for the writer's hospitalisation, so, from the writer's perspective, not only are they capable of effecting their release, but it is also their moral responsibility to do so. A possible strategy here is to WARN and THREATEN the addressees of what will happen if they do not fulfil this moral obligation. These acts are often realised or supported by the use of religious formulae. The day-labourer Martin B., for instance, uses this pattern in very similar ways in five of his letters: The cited passages contain another interesting type of formulaic variation. While the concrete lexical realisation of the pattern varies, the individual units of argumentation and their sequence are remarkably constant. In a first step, the writer reminds the addressee of his or her own death, actualizing the famous vanity topos of memento mori. This topos can be traced back to Psalm 90:12: Lehre uns bedenken, daß wir sterben müssen (LU, 'So teach us to number our days'). In all five cases, the memento mori is followed by an explicit warning of the Last Judgement, where the addressees will have to account for their misconduct, especially for not supporting the writer's release.

. . Topical Argumentation
Referring to God and other religious authorities is an argumentative strategy the writers repeatedly use to prove and force their point. In these contexts, religious formulae are of particular relevance due to their status as endoxa, whose truth or validity is widely taken for granted as an effect of their source. Such source-based authority provides a specific potential for argumentation. In conventional argumentative contexts, speakers are pragmatically committed to proving the truth or validity of their thesis statement. Referring to the authority of a certain person or source, however, can obviate the need for actual argumentation. In these cases, we come across instances of topical argumentation, the general mechanisms of which were already described by Aristotle (cf. Aristoteles,Topik,116a). Using the authority topos is probably the most typical form of topical argumentation. According to Klein (2001Klein ( : 1317, the underlying pattern and conversational effect of the authority topos can be described as follows: if a person or entity associated with high authority produces an utterance U or performs an action A, it is likely that U is true or valid / that A is appropriate. In such cases, it is superfluous to support one's positions in a truly argumentative way. Our key word search for sagen ('say'; and its morphological variants) yielded the following examples (11) to (14)  In the context of the first passage, plumber Friedrich Wilhelm S. sharply criticises the militarisation of Germany in the 1930s and the common practice of religiously motivating oaths of loyalty and allegiance. Instead, he pleas for demilitarisation and a pacifist attitude from the church and society. To support this, he refers to a famous quote from Jesus, uttered when one of his disciples violently tried to prevent his arrest after the betrayal of Judas Iscariot: wer das Schwert nimmt, der soll durchs Schwert umkommen (Mt 26:52, LU, 'all they that take the sword shall perish with the sword'). By quoting Jesus and the scriptures, the writer has recourse to the authority of the source. At least for fellow Christians, it is not easy to take the opposite view.
This function can also be found in example (12). In a letter addressed to all Bavarian bishops, mill labourer Georg Sch. basically argues that he cannot be insane since he always acts according to Jesus's model. So if the writer were insane, Jesus would necessarily have been insane, too. The letter contains theoretical reflections on a number of theological issues, particularly on confessional schism. In the passage quoted, he requests the bishops not to excommunicate religious sceptics since Jesus had not expelled sceptics among his disciples either. The request is substantiated with reference to two biblical formulae: the well-known quote from the apostle Thomas ("doubting Thomas") wenn ich meinen Finger nicht in das Mal der Nägel und meine Hand nicht in seine Seite lege, glaube ich nicht (John 20:25,EU,'except I shall […] put my finger into the print of the nails, and thrust my hand into his side, I will not believe') and Jesus's verdict that the tares should not be rooted out to protect the wheat in the parable of the Wheat and Tares (Mt 13: 24-30).
A specific variation of the authority topos is characteristic for writers suffering from religious delusions. These patients often claim that they act in a particular way because God -or other persons or forces in his name -has explicitly told them to do so. The action is thus justified as God's will. In these contexts, too, religious formulae play an important role since the religious authority often uses words from the scriptures. This may be illustrated by the following examples: In both passages, the use of the religious formula not only motivates and justifies a particular action of the patient's, but also establishes a parallel between the writers and the biblical persons who were the original addressees of the respective words (Abraham and Joseph) (see section 5.1.2).
In example (13), it is God himself who speaks to Magdalena S. with the same words he once used to call Abram (Und der HERR sprach zu Abram: Gehe aus deinem Vaterlande und von deiner Freundschaft und aus deines Vaters Hause in ein Land, das ich dir zeigen will Und ich will dich zum großen Volk machen; 1. Mose 12:1-2, LU -'Now the LORD had said unto Abram, Get thee out of thy country, and from thy kindred, and from thy father's house, unto a land that I will shew thee: And I will make of thee a great nation').
While the wording in (13) widely conforms to the biblical source, the formulae in (14) exhibit more adaptions to the context. In her diary, written in 1852 in the hospital in Augsburg before she had been transferred to the psychiatric hospital in Irsee, Anna K. refers to Archangel Gabriel's Annunciation, whereby she herself is being addressed as Virgin Mary, who will give birth to Jesus (cf. Lk 1:31). However, the angel does not appear to the writer herself but to the then-king Maximilian II. of Bavaria, who forwards the angel's prophecy. The religious authority of the original source is enriched with the earthly authority of the king as the initial bearer and receiver of the message -an interesting "double" use of the authority topos. In a further instance of topical argumentation, Anna K. then recites a voice from inside her, adopting the words of Mt 2:13: Steh auf, nimm das Kind und seine Mutter und flieh nach Ägypten; […] denn Herodes wird das Kind suchen, um es zu töten (EU, 'Arise, and take the young child and his mother, and flee into Egypt, […] for Herod will seek the young child to destroy him'). These are the words of an angel of the Lord who appears to Joseph in a dream after Jesus's birth, and Anna K. interprets this as an instruction to leave with Dr Enzler, one of the hospital's doctors.

. . Affirmation by Parallelisation
As we have seen, the patient texts contain a large number of cases in which the writer parallelises his or her own situation with a religious event, person or topos.
Our key word search for Kreuz ('cross') in CoPaDocs yielded several examples in which the writer's situation (being in the hospital) is metaphorically conceptualised as the cross he or she has to bear. 12 The farmer's daughter Marie V., for example, wrote the following letter (cited in full) from a psychiatric hospital in Lower Bavaria:  (Köpf 1996: 728). In several religious publications, we find the story of Veronica and her veil, as well as the formula Jesus nimmt das Kreuz auf seine Schultern ('Jesus carries the cross on his shoulders.'), for example in a 19th-century meditation book on the Via Dolorosa for use in church in Passau, Lower Bavaria (Schmid 1855: 6). Marie V.'s use of Christian imagery and religious language allows her to draw parallels both to her state of health and to the doctors' and nurses' benignity.

. . Expression of a Shared Ethos
It has often been noticed that formulaic language plays an important role in addressing one's audience (cf. e.g. Fleischer 1997: 218). In argumentative contexts, for instance, writers often resort to formulaic language to express shared knowledge, assessments and values (Pfeiffer 2016: 227-234). In the CoPaDocs letters, too, writers allude to religious formulae to establish a connection and create a social community with their addressees. While Martin B. predominantly uses religious formulae to warn and threaten his addressees, mainly his wife -see the five letters cited in example (10) -his formulae are not restricted to this function. In the few religious formulae 13 to the other addressees, the mayor of his home town, his aunt and his brother-in-law, we can observe other functions. Example (9) has already demonstrated the use of religious formulae in a letter to the mayor in which he illustrated his sorrow at being brought back to the hospital, with a modified formula putting emphasis on his general argument that he should be released. He also includes a religious formula in a letter to his aunt: In this passage, Martin B. refers to a letter that he received from his aunt (not part of his file), in which she tried to calm him down by asking him to trust in God as an aide in any misery (cf. Psalm 124). He points to this shared knowledge but emphasises the importance of mercifulness on earth and thus the aunt's duty to help him. Although his text here is still appellative, the writer does not try to reach his goals by means of a threat, but by referring to a common knowledge and expressing a shared ethos with the recipient, a practice that has been observed in other contexts (Dossena 2013; see section 2.2). The threats only appear in the letters to his wife and his last letter to the mayor, when he may have realised that the politeness of his earlier letters would not lead to the goals envisioned.

. Text-Structural Functions
While our analysis has up to this point focussed on functions of religious formulae with regard to the content of the letters, we can also find formulae that primarily help to structure letters (cf. Stein 2003: 123-132). As we have seen above, Magdalena S. (example 1) starts her letter with a religious formula, while Georg W. (example 3) closes his letter with a religious formula plus Amen. To find out whether this practice of closing letters with Amen also appears in other patient letters, we searched the corpus for Amen. In several cases, Amen is conventionally used to close prayers that are recited in the patients' texts. Some are similar to Georg W.'s example, where a text ends with a religious formula that closes with observed 32 religious formulae. The number of tokens addressed to his wife is higher (about 5,100 words) than to the other addressees (together about 2,750 words), but the religious formulae are still clearly dominant in the letter to his wife. Religious lexis in general is significantly more frequent in the letters to his wife than to the other addressees.

Conclusion and Outlook
Our paper has illustrated the spectrum of occurrences and functions of religious formulae in the CoPaDocs corpus, comprising letters from patients of psychiatric hospitals born in the 19th century. Since most of the letters were written by lowerclass people with a low level of education, the letters permitted an insight into the use of formulaic language by ordinary people in the 19th century. Some of our findings can be regarded as specific to the letters investigated here. This holds in particular for passages where patients suffering from religious delusions claim that God or other religious instances have told them to do something using religious formulae known from other contexts. Another interesting observation was that lower-class writers in the 19th century already exploited the potential of formulaic language for occasional modifications, often in a remarkably complex and creative manner. This shows that the tendency for creative use of formulaic items has a long tradition and is not a development of recent decades. Interestingly, the modifications we found do not seem to aim at wordplay but are most obviously produced to achieve particular communicative goals.
The majority of our findings, however, rather confirm observations that have already been made for 19th century lower class letter writing in previous studies. Prominent features such as a high degree of variation in formulaic language use, and a tendency to refer to religious formulae to express shared values and to utilise the advantages of topical argumentation are not specific to the corpus investigated, but seem to be characteristic of 19th century lower-class letter writing in general. This also seems to apply to the secularised use of Amen in a text-structural function.
This volume is dedicated to Elisabeth Piirainen, who in a number of papers and books has researched on widespread idioms in Europe and beyond (e.g. Piirainen 2012Piirainen , 2015. Continuing this approach to formulaic language, contrastive studies -not only across the borders of single letter collections but also across different languages -might offer a variety of new insights, especially with regard to a "pan-European tradition of letter writing" (Rutten and Van der Wal 2013: 52) and the question of "whether there existed something like a central European stock of letter writing formulae and how they could have evolved or how they were transmitted into the different languages" (Elspaß 2012: 60). Our findings on the use of Amen could be related to this idea. However, the various writers' individual attempts at closing a letter by means of this formulae may as well have resulted from their shared European cultural and religious backgrounds.
In any case, religious formulae in historical lower-class writing have been shown to be a promising area for historical research on formulaic language. The corpus of historical patient documents provides new data that can shed light on the occurrence and functions of religious formulae in these letters, giving us valuable insights both into the writers' religious knowledge and into their competence in transforming this knowledge into a letter.