Twelve Coding Challenges for Christmas (Part 2)

By Dr Patricia O’Connor, Irish Postdoctoral Researcher

Happy New Year from the OG(H)AM team! Last year, we brought 2024 to a close with the gift of twelve coding challenges for the twelve days of Christmas. Our last blog post focused on six of the twelve challenges for Christmas in descending order and demonstrated how to avail of the EpiDoc Guidelines for encoding decorative features associated with the inscription, the find spot of the inscribed stone, as well as the language and direction in which the inscription is written. Additionally, the previous challenges explained how to utilise ontologies or controlled vocabularies such as the FAIR Epigraphy Vocabularies for encoding the purpose of a text and the Europeana EAGLE project’s vocabularies for encoding the carving technique of the inscription. For those wishing for more of an introduction to encoding, I’ve also written an overview on mark-up and the importance of encoding standards such as the EpiDoc Guidelines in an earlier blog post on the History of Encoding. Resuming from where our previous blog post concluded, this companion post is chiefly concerned with the challenges that arise from encoding the inscription itself. The OG(H)AM team’s final six coding challenges for the twelve days of Christmas address structuring biscriptal inscriptions, encoding the form and appearance of letters such as ligatured or reversed letters, handling both errors as well as corrections in an inscription, representing ambiguous, damaged or illegible characters, adding alternative readings of inscriptions, and accurately portraying standard and non-standard ogham letterforms. 

6. Biscriptal, bi-version and bilingual inscriptions 

In our last blog post we shared some examples of stones with both ogham and Latin inscriptions, and now we finally get to show you how to encode such biscriptal inscriptions. The term biscriptal designates inscriptions that are written in two different scripts. In the context of the OG(H)AM project, biscriptal inscriptions usually consist of texts that are written in both the ogham and Latin script. The biscriptal inscriptions that the OG(H)AM team have encoded thus far can be divided into two categories: bi-version and bilingual.

In our blog post Ogham in the British Museum, the OG(H)AM team reported on a granite pillar stone which was originally found at a farm in Fardel in Devon (E-DEV-001). The stone bears a bi-version inscription, independent texts in two different languages (see Fig. 1). There are two Latin inscriptions: one on the front and the other on the back of the stone, both facing left and reading downwards, and there is also an ogham inscription reading vertically upwards on the two angles of the front face of the stone (see Fig. 2). 

The Multi-Lingual Texts section of the EpiDoc Guidelines offers guidance on how to structure the edition correctly for these three inscriptions. The OG(H)AM team complies with the second method, using the <div type=“textpart”> approach for marking-up the three texts. In Fig. 3, <div type=“textpart”> occurs thrice: once for the ogham text and twice for the Latin texts. The attribute @n is used to number each text part.

Representing bi-version inscriptions also requires some restructuring of the edition section of the EpiDoc file. To facilitate findability and ensure a legible display of multi-lingual inscriptions on the OG(H)AM project database, a corresponding edition division is provided for each text part (see Fig. 4). The ogham inscription and transliteration “SVAQQUCI MAQI QICI” constitute the first two editions and the first text part. The second text part and third edition division is for the first of the two Latin inscriptions, “FANONI MAQVI RINI”, which is written in two lines on the front of the stone, while the final text part and edition division is for the single word “SAGRANVI” inscribed on the back of the stone.

On the possibility that the ogham and Latin inscriptions were perhaps linked, Patrick Sims-Williams (2003) proposed that the letters ‘SV’ in the ogham inscription were a substitution for the ‘F’ in the Latin inscription and further suggested that the carver had confused the orientation of the H- and B-aicme letters. While this suspected mistake makes sense in relation to the first name with ‘QQ’ for ‘NN’, it is more difficult to explain “QICI” as an error for “RINI”. Conversely, Macalister maintained that the three inscriptions represented three distinct phases of use and that there was no connection between them (1945, 469). There are, however, other examples of biscriptal inscriptions which are connected.

When we say ‘bilingual inscriptions’, this implies that the same text has been inscribed in two different languages. The Llanfaelog stone from Anglesey in Wales preserves a commemorative Latin inscription and an accompanying ogham inscription (W-AGY-001). On the largest smooth face of the stone is the Latin inscription “MAILISI” reading vertical downwards while the right-hand-edge of the same face has a one-line ogham inscription reading vertically upwards. The ogham inscription is fragmentary, however, the space available between the surviving characters supports the assumption that the ogham corresponds to the roman letter inscription. Furthermore, Charles Edwards reasoned that the closeness of the ogham inscription to the roman-letter inscription was presumably to indicate that one was the translation of the other and that the ogham “should be read, as expected in Irish, as MA[ILI]SU (nominative)” (2013, 164-5).

In accordance with the EpiDoc Guidelines for encoding multi-lingual texts, the bilingual inscriptions on the Llanfaelog stone are divided into two text parts (see Fig. 6). The OG(H)AM team supplies specific edition sections for the ogham inscription, the ogham transliteration, and the Latin inscription respectively, using the <div type=“edition”> element and attribute combination (see Fig. 7). The attribute @subtype is used to distinguish between the ogham, the transliteration, and the roman inscription. 

5. Form and appearance of the letterforms 

Authentically conveying the form and appearance of the letterforms preserved in inscriptions is a chief consideration for the OG(H)AM team. While encoding biscriptal inscriptions, the team has encountered examples of ligatured letters as well as reversed or inverted letters in the Latin inscriptions. To encode these letterforms, the OG(H)AM team consulted the specific sections of the EpiDoc Guidelines which are dedicated to offering guidance and examples for encoding Ligatured Letters and Reversed or inverted letters. The EpiDoc Guidelines recommend using the highlighted element (<hi>) to mark-up or encode letters, a word or a phrase as graphically distinct from the surrounding text. The attribute @rend is used to encode how the inscribed text was rendered or presented. 

In addition to preserving the commemorative bilingual inscription “MAILISI”, the Latin inscription of the Llanfaelog stone also contains evidence of inverted as well as ligatured letters (W-AGY-001). The first two letters “MA” are ligatured and the final letter of the inscription “I” is inverted. Encoding the form of these letters first involves enclosing the ligatured letters “MA” and inverted “I” with the highlighted element’s opening (<hi>) and closing (</hi>) tags. As you can see outlined in Fig. 8, the value of the attribute @rend is crucial for differentiating between inverted and ligatured letters. To encode ligatured letters, the attribute @rend is added to the opening highlighted element tag <hi> and the value of the attribute @rend is always “ligature”. Similarly, an attribute @rend value of “inverted” denotes that the letter(s) enclosed by the <hi> element are inverted. 

Encoding reversed letters is almost identical to the process of encoding inverted and ligatured letters. In the two-line Latin inscription “DVNOCATI HIC IACIT FILI MESCAGNI” from Lancarffe (E-CON-X02), there is a reversed capital as well as two ligatured letters. In the first line of the inscription, the third letter of DVNOCATI, the capital “N”, is reversed; while in the second line, the first two letters “FI” are ligatured (see Fig. 9). The OG(H)AM team followed the EpiDoc recommendations and used the highlighted element <hi> to encode the reversed capital “N” as well as the two ligatured letters “FI” in the inscription. In this instance, the main difference is that the value of the attribute @rend in the opening <hi> element tag for the letter “N” is “reversed”. For the ligatured letters “FI” at the start of the inscription, the attribute @rend remains as “ligature” as in our previous example (see Fig. 10). 

If you read last month’s blog post, then the rest of the elements in the encoding of the Lancarffe inscription are likely familiar. We covered encoding personal names using the <persName> and <name> elements. Additionally, we discussed basic structural elements such as the <div> element for marking subdivisions of text; the <ab> element for indicating anonymous blocks or units of text that are analogous to, but without the same constraints as, a paragraph; and the <lb> element for marking the beginning of a line of text. There are more advanced structural elements too such as the word element <w> which is used to encode a word.  

In Fig. 11, the <w> element is employed three times to enclose the words “HIC” “IACIT” and “FILI” in the Lancarffe inscription. These three words are examples of formulaic words, words which occur frequently in inscriptions. This information is conveyed within the encoding through the attribute @type value “formula”, effectively demonstrating that attributes are an incredibly useful means for providing additional relevant information as well as for distinguishing between reversed, inverted and ligatured letters.

4. Errors and corrections 

The stone known as ‘Long Cross’ at St Endellion in Cornwall (E-CON-X04) featured in last month’s blog post as an example of a decorated inscribed stone, but it also contains evidence of a carving error in the inscription. The stone allegedly preserved a bilingual inscription. The presence of an ogham inscription reading “BROCAGNI” on the right-hand angle of the stone was suggested by Macalister but described as “doubtful” (1945, 457) and later dismissed by Okasha (1993, 235). The two-line Latin inscription reads “BROCAGNI IHC IACIT NADOTTI FILIVS” (see Fig. 12).

In the first line of the Latin inscription, the formula word ‘HIC’ is incorrectly carved as ‘IHC’. The Erroneous Substitution section of the EpiDoc Guidelines outlines an encoding strategy that allows both the error and the correction by the editor to be read in the edition. Encoding both the mistake and correction involves the consistent usage of three core elements: <choice>, <sic>, and <corr>. The <choice> element acts as a wrapper or container and groups alternative encodings at a specific point within a text. In the encoding example illustrated in Fig. 13, the <choice> element is employed to group the error in the original inscription together with the correction to the text. The <sic> element supplies the means to represent the text authentically, even its inaccuracies, and surrounds the incorrect word or characters. Lastly, the purpose of the correction element (<corr>), as its name suggests, is to contain the correct character or word which was erroneously written in the text. For the Latin inscription found at St Endellion, the <sic> element encloses the incorrectly spelled ‘IHC’, while the <corr> element surrounds the correct form of the formula word ‘HIC’ which is supplied by the editor. The <choice> element wraps around both the <sic> and <corr> elements, and, in doing so, enables the OG(H)AM team to present the reader with both the erroneous and correct reading of the inscription.

3. Characters: Ambiguous, Damaged, Supplied or Illegible 

Not all inscriptions are as clear as the inscription at St Endellion. Unfortunately, some inscriptions become so worn by weather or through damage that the interpretation of individual letters would be ambiguous outside of their context. Embedded on top of a barrow located at Island in Mayo (I-MAY-001) is an ogham-inscribed pillar stone that is now almost prostrate (see Fig. 14).

The ogham inscription “CUNALEGI AVI QUNACANOS” is carved on both angles and the top of the stone. As illustrated in Fig. 15, the inscription is legible but damaged in places. “AVI” (except the ‘A’ notch and distal ends of the ‘V’) and parts of some other scores are chipped away. Some letters in the first name are also damaged. Only one clear notch of the ‘U’ remains, one damaged and three full scores to the right follow before the ‘A’ notch. However, the space allows for a more likely ‘UN’ reading. The last name is remarkably clear and undamaged.

For encoding characters of which some traces survived, the OG(H)AM team consulted the Ambiguous Characters section of the EpiDoc Guidelines. The advice for encoding ambiguous letters is to use the unclear element (<unclear>). The EpiDoc Guidelines are a restricted subset of the TEI Guidelines, and, as a result, the use of certain elements is usually more restrictive. The unclear element is one such example: the use of <unclear> within the context of an EpiDoc file is concerned specifically with the physical preservation or execution of the characters. An optional addition is to include the attribute @reason when encoding ambiguous characters. The OG(H)AM team chose this option to encode additional relevant information when encoding the damaged characters ‘UN’ in the first line of the inscription (see Fig 16). The green underlined section of code in Fig. 16 demonstrates how the <choice> element can be employed to contain alternative encodings within the context of ambiguous letters. In this specific scenario, the number of vowel scores are apparent but whether the scores are for the letter ‘I’ or the letters ‘EA’ remains ambiguous. The <choice> element enables the OG(H)AM team to include each possibility in the encoding of the inscription. The <unclear> element is used to surround the alternative readings, ‘I’ and ‘EA’ respectively, and the <choice> element then encloses both readings.

The red underlined section of code in Fig. 16 highlights the use of the attribute @reason to specify that the encoded character is now ambiguous because of “damage”. Of course, there are instances where letters of an inscription have been damaged but are still legible. In such cases, the advice from the Damaged Characters section of the EpiDoc Guidelines is to use the <damage> element.

At Ballyknock in Co. Cork, fifteen ogham stones were found within a single souterrain inside a large ringfort at Castle Farm. One of these ogham stones (I-COR-035), which is on permanent display in the Stone Corridor at University College Cork (see Fig. 17), exhibits an inscription that has been damaged, but the letters remain in fair condition. The inscription “CLIUCOANAS MAQI MAQI-TRENI” is pocked in bold strokes on the two angles of the stone (see Fig. 18).

Fig. 19 demonstrates the OG(H)AM team’s approach for encoding the damaged letters of the Ballyknock ogham stone inscription (I-COR-035). In the second line of the inscription, the letter ‘I’ in ‘MAQI’ of the name ‘MAQI-TRENI’ is damaged. Representing this information in code is straightforward with the <damage> element which is underlined in red in Fig. 19. No doubt you have spied that the encoding samples in Fig. 16 and 19 also show another valuable element being utilised. The supplied element (<supplied>) is being used by the OG(H)AM team, in accordance with the advice outlined in the Editorial Restoration: Characters Lost and Restored Tentatively section of the EpiDoc Guidelines, to supply characters or text that can no longer be read because of textual loss or damage to the support. 

Regrettably, there are examples of inscriptions where characters have faded or been damaged to such an extent that they are no longer recognisable. The EpiDoc Guidelines have several sections on the subject of illegibility which detail transcription recommendations for encoding illegible characters, lines, and text to a known, unknown or approximate extent.

At Coomleagh East in Cork (I-COR-001) there is a pair of standing stones: one is now prostrate and the other which is still standing is inscribed with ogham. The inscription was badly damaged sometime after 1845 and only traces of the final four letters remain “TETA”. The rest of the inscription which is no longer legible is signified by the <gap> element as stipulated by the EpiDoc Guidelines’ Illegible Texts, Other Units section (see Fig. 21). The EpiDoc Guidelines also advise to include the attribute @reason with a value of “illegible” when using the <gap> element.

2. Alternative readings  

Although we have covered alternative encodings of letters or parts of a text, we have not yet addressed alternative readings of inscriptions in their entirety. The ogham-inscribed antler knife handle found at Weeting-with-Broomhill and now in the possession of Norwich Castle Museum (E-NFK-001) is a good example for demonstrating how to encode alternative readings. The ogham inscription consists of two lines and spans the length of both sides of the handle.

This inscription is intriguing because it is not certain which direction it ought to be read. One possibility is to read the inscription upwards beginning from the broad end of one side of the handle and to continue reading downwards to the broad end of the other side of the handle. Alternatively, the inscription could be read upwards commencing from the narrow end on one side of the handle and reading down towards the narrow end of the other side of the handle. Clarke (1952, 72) determined that the latter reading (beginning and ending with the narrow end) is “perhaps the more likely of these two readings”.

It is possible to represent both readings of the inscription within the EpiDoc file. When there are two or more possible readings of a text, EpiDoc provides the means for editors to represent the alternative readings using an apparatus criticus which can be encoded inline within the edition or externally in a different section of the file. The OG(H)AM team chose the external apparatus criticus option which appears in a specific division marked <div type=“apparatus”>. Fig. 24 shows the OG(H)AM team’s encoding of the Weeting-with-Broomhill antler knife handle where the more likely reading of the inscription is encoded as normal within the <div type=“edition”> section and the alternative reading immediately follows in the <div type=“apparatus”> section. The purpose of the apparatus element (<app>) is to contain one entry in a critical apparatus which may include readings and notes on the text while the list of apparatus entries element (<listApp>) element gathers all the apparatus entries together. The actual reading is then encoded using the reading element (<rdg>) and the OG(H)AM team included the attribute @source to specify the source for the alternative reading.

1. Ogham characters  

To conclude this month’s blog post, and the twelve coding challenges overall, we have left perhaps the most complicated challenge for last: encoding ogham characters. As you may recall from previous blog posts, the representation of the ogham script in a digital format has been made possible thanks to the introduction of ogham to Unicode. Adrian Doyle has contributed three excellent blog posts on this topic which cover the implementation and implications of digital ogham; how digital ogham came to be, and the challenges it faced along the way; as well as future potential updates to the Unicode ogham block. While the complexities of encoding the ogham script cannot be fully covered in this blog post, I will attempt to offer a brief introductory overview of the measures that the OG(H)AM team have taken to represent the idiosyncrasies of ogham characters as authentically as is presently possible.

For all of the preceding ogham encoding examples in this blog post, the OG(H)AM team has used the Ogham Script Unicode Entity Codes to represent the twenty ogham characters of the four aicmi as well as the five supplementary forfeda letters, two feathermarks and an ogham space character (see Fig. 25). However, the OG(H)AM team has also encountered more unusual ogham letters while encoding Pictish inscriptions such as the “Hammerhead A” and “Rabbit-eared D” characters which necessitated another means of representation. Fortunately, the BabelStone Fonts website supplies a set of twelve Unicode Ogham fonts to enable the graphic representation of the different styles of ogham characters. Two of these fonts, the BabelStone Ogham Pictish and BabelStone Ogham Special fonts, provide glyph forms which closely resemble the ogham characters found in Pictish ogham inscriptions. But how do we represent these characters in an EpiDoc file?

To represent these special ogham characters in EpiDoc, the editor must first provide information about these special characters before they can be used in the encoding of the inscription. Descriptive or declarative information or metadata is always found within the TEI Header of the EpiDoc file which is signified by the element <teiHeader>. Nested within the encoding description subsection (<encodingDesc>) of the TEI Header, is the character declaration element (<charDecl>) which contains the required information for non-standard characters or glyphs. The OG(H)AM team uses the glyph element <glyph> to provide descriptive information and adds the attribute @xml:id to specify the ogham character. The mapping element (<mapping>) contains the characters that are related to the ogham letterform specified by the glyph element. In Fig. 26, the attribute @type is used to indicate which glyph should be used for each edition of the text: ogham, diplomatic, or interpretation. The approach delineated here is not yet finalised as the OG(H)AM team are continually testing and updating it to achieve optimum results; but it is included to document our efforts to date and to serve as a useful case study for other researchers. 

Once these special characters have been declared within the TEI Header, the editor can encode these ogham characters within the <div type=“edition”> section of the EpiDoc file. The strategy that the OG(H)AM team currently employs to represent these non-standard ogham characters is to use the glyph element <g> with the attribute @ref to point to the description of the character in the TEI Header (see Fig. 27). Then, the <g> element is surrounded by the highlighted element <hi> and the attribute @rend with a value of “babel_special” indicates that the BabelStone Ogham Special font should be used to represent the rabbit-eared D ogham character. 

Hopefully, these twelve encoding challenges have piqued your interest sufficiently enough to encourage you to start encoding yourself and that these past two blog posts have supplied you with plentiful reference examples to help you encode ogham and biscriptal inscriptions to the EpiDoc community standards. Happy encoding! 

References 

Clarke, R. Rainbird. (1952). “An Ogham inscribed knife-handle from south-west Norfolk” The Antiquaries Journal, vol. 32 pp. 71-3. 

Macalister, Robert Alexander Stewart. (1945). Corpus inscriptionum insularum Celticarum.  

Okasha, Elisabeth. (1993). Corpus of Early Christian Inscribed Stones of South-west Britain. Leicester University Press.