Multi-Script Handwriting: Latin, Devanagari, Arabic and CJK

Writing systems are classified by the relationship between their graphic units and the linguistic units they represent. The distinction matters for forensic examination because the script structure determines the inventory of class characteristics (shared features produced by the teaching method) and influences the range of individual characteristics that develop over a writer's history with that script.

An alphabet represents individual phonemes with individual characters. Latin script (Roman), Greek, Cyrillic, Armenian, and Georgian are alphabets. In Latin script, the letters 'b' and 'p' differ by the vertical position of the bowl relative to the baseline. These differences are class characteristics taught explicitly in schools; the individual variations from that class template are the forensic examiner's raw material. The forensic literature on Latin-script examination, from Osborn (1910) through Bradford (2007) through the SWGDOC 2013 guidelines, is based entirely on this structure.

An abugida (also called alphasyllabary) represents consonants with base characters, with inherent vowels that are modified rather than replaced by diacritical marks attached above, below, or around the consonant body. Devanagari (Hindi, Sanskrit, Marathi, Nepali), Bengali, Gujarati, Tamil, Telugu, Kannada, Malayalam, Thai, Tibetan, and Ethiopic (Ge'ez) are abugidas. In Devanagari, the distinctive feature is the headline (shirorekha): a horizontal stroke at the top of most characters that connects adjacent letters into words. The shirorekha is a class characteristic of Devanagari; its individual variation, continuity, pen pressure, and the exact height at which it is placed relative to the consonant body, are among the most individualising features in Devanagari examination.

An abjad represents only consonants; readers supply vowels from context. Arabic, Hebrew, Urdu, Persian (Farsi), and Pashto are abjads. Arabic script runs right to left, and each letter has up to four context-dependent forms (initial, medial, final, isolated), depending on its position within a word. The ligatures connecting letter forms within a word are executed as continuous strokes in fluent writing; the number of pen lifts within a word, the rhythm of ligature strokes, and the proportional balance between letter bodies and connecting strokes are among the most individualising features.

A logograph (or logosyllabary) represents morphemes or syllables with complex characters rather than phonemes. Chinese (Hanzi), Japanese (Kanji, shared with Chinese plus the syllabic Hiragana and Katakana), and Korean (Hangul is technically an alphabet arranged in syllable blocks) are the major logographic or logosyllabic systems. Chinese Hanzi characters are built from a small inventory of approximately 32 canonical strokes (heng, shu, pie, na, etc.), each with a defined stroke order specified by the National Standard of China (GB/T 15834). The stroke order is a class characteristic taught rigorously in primary school; deviation from standard stroke order is an individualising feature that experienced Chinese forensic document examiners use for attribution and chronological assessment of a writer's proficiency level.

Devanagari examination in India is conducted primarily by the CFSL (Central Forensic Science Laboratory) network under the Union Home Ministry and by state FSLs (Forensic Science Laboratories) under the respective state directorates of forensic science. The BFS (Bureau of Forensic Science, Gujarat) and the RFSL (Regional Forensic Science Laboratory) Hyderabad both have dedicated questioned-document divisions examining Devanagari, Gujarati, and Telugu scripts respectively. The methodological framework applied in Indian national laboratories follows the general principles of the Albert Osborn tradition, adapted for Devanagari class and individual characteristics through decades of casework-derived understanding, supplemented by research publications from the Sagar University Department of Forensic Science and the Dr. Harisingh Gour University.

The shirorekha (headline) is the most distinctive class feature of Devanagari. It is present in most base consonants and is the horizontal joining element of words. Individual variation in the shirorekha includes: the height at which it is placed relative to the body of the consonant (high, standard, or low attachment), the pen pressure applied (lighter than body strokes in some writers, equal or heavier in others), its continuity (a single unlifted stroke across a word, or broken between individual characters), and its terminal treatment (blunt cut, tapering, hooked, or retroflexed). These individual variations are consistent across a given writer's naturalistic Devanagari writing and are detectable across disguise attempts.

The half-consonant forms (halant + reduced consonant: for instance, the half 'ta' in 'sthan') and conjunct consonants (consonant clusters written as ligatures) are another rich source of individual variation. The degree to which a writer forms conjuncts as discrete fused elements (the trained pedagogical form) versus as rapidly flowing merged strokes (the naturalised adult form) reflects years of practice and correlates with education level and regional schooling conventions (Maharashtra school curricula prescribe slightly different conjunct forms than UP or Bihar curricula).

Numeric character forms in Devanagari (the Brahmi-derived numeral set used alongside Western Arabic numerals in many Indian documents) carry strong individual variation in the forms of digits 2, 3, 6, 7, and 9, which differ structurally from their Western Arabic equivalents. Documents mixing Devanagari script with Western Arabic numerals require the examiner to assess both character sets independently against exemplars in both forms.

In Indian legal proceedings, the Bharatiya Sakshya Adhiniyam 2023 (s 39, formerly Indian Evidence Act s 45) governs the admissibility of handwriting expert opinion. The Supreme Court's position in Prem Sagar Madan v. State of Delhi (2000) and subsequent cases is that expert handwriting opinion is admissible but not conclusive; the court weighs it alongside other evidence. A parallel position applies in Pakistan (under the Qanun-e-Shahadat Order 1984, Article 59) for Urdu/Nastaliq script examination.

Arabic script is used (in some form) for Arabic, Urdu, Persian, Pashto, Ottoman Turkish (pre-1928), and several other languages across North Africa, the Middle East, South Asia, and Central Asia. The script runs right to left, uses a baseline (but not a headline as in Devanagari), and most letters join to the preceding and following letter via a connecting ligature, producing a continuous cursive line within words. Short vowels (harakat: fathah, kasrah, dammah) are typically omitted in adult writing, making the same consonant skeleton readable in multiple ways depending on context: forensic examiners working on Arabic documents must be fluent readers of Arabic to interpret the text reliably.

The four positional forms of each Arabic letter (isolated, initial, medial, final) are class characteristics. In forensic examination, the focus shifts to individual variation within and around each positional form: the degree of openness in the loop of baa/ya (the sub-baseline descending loop), the specific angle and length of the connecting ligature between adjacent letters, the proportion of the letter body to the connecting stroke, the direction and curvature of terminal strokes at the end of words, and the positioning and form of diacritical dots (nuqat: the one, two, or three dots that distinguish b/t/th, n/y, and other letter pairs).

Nuqat examination is particularly diagnostically significant because the diacritical dots in fluent Arabic handwriting deviate substantially from the isolated-dot forms taught in school pedagogy. In rapid adult writing, groups of two or three dots frequently merge into a single stroke (a circumflex-like mark for two dots above, a caret for three dots). The specific form this merger takes is highly individual: some writers produce a horizontal bar, others a V-shape, others a wave. This collapsed-dot individualisation is not covered in most Latin-script forensic literature because Latin script has no equivalent feature.

Nastaliq (used for Urdu, Persian, and some Pashto writing) is a cursive variant of the Arabic abjad in which letters have a strong descending diagonal trajectory, creating a distinctive drooping calligraphic line. The Pakistani Forensic Science Agency and the CFSL Hyderabad examined-document division have both published casework series on Nastaliq attribution, noting that the individualising features in Nastaliq differ significantly from those in Naskh (the more upright standard Arabic form): in Nastaliq, the depth and curvature of the descending diagonal, the pen angle, and the degree of connection between the descending portion and the baseline ligature are the primary comparison features.

The ENFSI Document Examination Working Group's 2018 international reference guide to Arabic script examination provides standardised terminology for Arabic letter components and a tiered framework for reporting conclusion confidence levels in Arabic casework, using the same nine-point scale as Latin-script examination. The guide explicitly notes the absence of large-scale empirical validation studies for Arabic examination and calls for collaborative research programmes.

Chinese, Japanese, and Korean (CJK) handwriting forensic examination differs from Western examination in one fundamental respect: the character inventory is orders of magnitude larger. A literate native Chinese writer knows approximately 3,000 to 5,000 characters in regular use; the full GB 18030 standard for Chinese contains over 70,000 codepoints. This means that the class characteristics (standard form) are more numerous and that individual writers' deviations from those class characteristics are potentially more distinctive, because the deviation from a given stroke within a given character is observable across multiple instances of that character in the document.

The GB/T 15834 stroke order standard (China) specifies the canonical sequence for every stroke in every standard character. The pedagogical sequence for most characters follows consistent rules: left before right, top before bottom, horizontal before vertical, outside before inside. Writers who were educated in Mainland China write character components in this standard order; writers educated in Taiwan (where the Traditional characters and the Ministry of Education's stroke-order conventions differ slightly from the PRC standard) or in Japan (where Kanji is taught with slightly different stroke-order conventions under the Joyo Kanji standards) may produce certain characters with different stroke sequences. These stroke-order differences are class characteristics of the respective educational systems, not individual characteristics, but they allow the examiner to broadly categorise a writer's educational background.

Individual variation in CJK handwriting is found in: the spatial proportions of components within characters (in characters with two or more components arranged side by side, the proportion of space allocated to the left vs right component varies individually); the degree of character reduction in naturalistic fast writing (more experienced and more casual writers reduce characters more aggressively, merging or eliding strokes); the specific angle and curvature of individual strokes (the 横 heng horizontal stroke is individually variable in its ascent angle, its terminal hook, and its length relative to the character frame); and the degree to which a writer has adopted cursive (草書 caoshu) forms for specific common characters even when writing nominally in regular script (楷書 kaishu).

The National Police Agency of Japan's forensic document examiners have published comparative casework studies on Japanese handwriting (Kanji and Kana) since the 1980s. The Public Security Intelligence Agency in China has an active forensic handwriting division. In Singapore, the Health Sciences Authority's forensic document team examines documents in Chinese, Malay, Tamil, and English, and its multi-script protocols have been published in the International Journal of Forensic Document Examiners. South Korean forensic document examination is conducted by the National Forensic Service under the Ministry of Justice.

The forensic examination of Korean Hangul is in some respects closer to Latin examination than to Chinese examination: Hangul is an alphabet arranged in syllable blocks, with a small set of consonant and vowel letters that combine into blocks. Individual variation is found in the spatial arrangement of components within blocks, the pen lifts between consonant and vowel elements, and the characteristic stroke terminals of individual consonants.

Cross-script disguise occurs when a writer attempts to imitate a different script system, typically to mislead investigators about the writer's linguistic or cultural background. A Latin-script writer producing text in a script that visually resembles Arabic, Hebrew, or Devanagari will produce characteristic errors that expose the disguise.

The most consistent diagnostic of cross-script disguise is wrong stroke order. In Arabic, the letters are formed from right to left; a Latin-script writer imitating Arabic will frequently produce strokes in left-to-right or top-to-bottom order (their habitual Latin-script construction direction), creating directionality discontinuities that are visible under oblique light or in high-resolution digital imaging. In Devanagari, the shirorekha is typically written after the body of the consonant in standard pedagogy; a forger unfamiliar with Devanagari will frequently draw the headline first (as it is visually prominent at the top), producing a construction sequence incompatible with any trained Devanagari writer.

Context-dependent letter forms are a second diagnostic. Arabic letters change form depending on position; a forger who has not studied Arabic will produce the same letter form (typically the isolated form, which is the most visually salient) regardless of position, creating medial and final positions with isolated-form characters that no fluent writer would produce.

The pen-lift pattern is a third diagnostic. Fluent Arabic writers connect most letters within a word without pen lift (except at letters that never join to the following letter: alef, dal, dhal, ra, zay, waw). A forger drawing Arabic characters will lift the pen between each character, producing a segmented trace that is visually detectable even without instrumentation.

These cross-script disguise diagnostics have been applied in several immigration-fraud contexts. The ENFSI Document Examination Working Group's 2018 language and script analysis guidance explicitly addresses the forensic assessment of documents claimed to originate from a specific country or language community. In the UK, the Home Office's forensic language analysis contractors conduct handwriting-based script-origin analysis as part of asylum-seeker nationality verification procedures, examining documents claimed to be from specific countries against reference collections of genuine documents from those countries. Similar procedures operate in Germany (via the Bundesamt fur Migration und Fluchtlinge, BAMF, forensic document services) and in Australia (via the Australian Federal Police's document examination capability).

The Scientific Working Group for Forensic Document Examination (SWGDOC), operated under FBI sponsorship until its transition to OSAC (Organization of Scientific Area Committees for Forensic Science) under NIST in 2014, produced a set of standards for forensic document examination that are widely applied in the US and referenced internationally. The 2013 SWGDOC standard for the examination of handwritten items specifies the process of examination (collection of exemplars, comparison methodology, conclusion reporting) in a script-neutral way, with guidance notes acknowledging that the specific class and individual characteristics differ by script.

SWGDOC's multi-script subcommittee produced advisory documents on Arabic script (2011), Chinese script (2013), and Devanagari script (2014), each providing terminology, classification frameworks, and guidance on examiner qualifications for non-Latin script work. These documents are available through the OSAC forensic document examination resource library and are referenced by the American Board of Forensic Document Examiners (ABFDE) in its certification requirements.

ENFSI's Document Examination Working Group (DEWG) covers European member laboratories across 22 countries and has produced best practice manuals for handwriting examination (2017 edition), ink analysis, and paper analysis. The multi-script coverage in the ENFSI BPM reflects the migration-related caseload of European document laboratories: Arabic, Pashto, Farsi, and Dari examination capabilities are explicitly addressed, given the volume of asylum and immigration casework involving documents from South Asia and the Middle East. The ENFSI external quality assurance exercises (proficiency tests) now include non-Latin script samples in the handwriting panel.

The International Association of Questioned Document Examiners (IAQDE) and the International Association of Document Examiners (IADE) have both called for the establishment of shared reference databases for non-Latin scripts, analogous to the CEDAR-FOX writer-identification database for Latin script. Progress has been slow because building a validated reference database requires systematic exemplar collection across demographic groups, countries, and educational traditions, a resource-intensive undertaking that has not yet attracted the funding that the Latin-script database work received through US Department of Justice grants to CEDAR in the early 2000s.