Editing Decipherment (section)

== Methods ==
There is no single recipe or linear method for decipherment, however: instead, philologists and linguists must rely on a set of [[Heuristic|heuristic devices]] that have been established. Broadly, it is important to be familiar with the relevant texts where the script or language occurs in, access to accurate drawings or photographs of these texts, information about their relative chronology, and background information on where the texts occur in (their geography, perhaps being found in the context of a funerary monument, etc).<ref name=":5" /> 

These methods can be divided into approaches utilizing external or internal information.<ref name=":1" />

=== External information ===
Many successful decipherments have proceeded from the discovery of external information, a common example being through the use of [[multilingual inscription]]s, such as the [[Rosetta Stone]] (with the same text in three scripts: [[Demotic (Egyptian)|Demotic]], [[Egyptian hieroglyphs|hieroglyphic]], and [[Greek alphabet|Greek]]) that enabled the decipherment of Egyptian hieroglyphic. In principle, multilingual text may be insufficient for a decipherment as translation is not a linear and reversible process, but instead represents an encoding of the message in a different symbolic system. Translating a text from one language into a second, and then from the second language back into the first, rarely reproduces exactly the original writing. Likewise, unless a significant number of words are contained in the multilingual text, limited information can be gleaned from it.<ref name=":1" />

=== Internal information ===
Internal approaches are multi-step: one must first ensure that the writing they are looking at represents real writing, as opposed to a grouping of pictorial representations or a modern-day forgery without further meaning. This is commonly approached with methods from the field of [[grammatology]]. Prior to decipherment of meaning, one can then determine the number of distinct [[grapheme]]s (which, in turn, allows one to tell if the writing system is alphabetic, syllabic, or logo-syllabic; this is because such writing systems typically do not overlap in the number of graphemes they use<ref name=":2" />), the sequence of writing (whether it be from left to right, right to left, top to bottom, etc.), and the determination of whether individual words are properly segmented when the alphabet is written (such as with the use of a space or a different special mark) or not. If a repetitive schematic arrangement can be identified, this can help in decipherment. For example, if the last line of a text has a small number, it can be reasonably guessed to be referring to the date, where one of the words means "year" and, sometimes, a royal name also appears. Another case is when the text contains many small numbers, followed by a word, followed by a larger number; here, the word likely means "total" or "sum". After one has exhausted the information that can be inferentially derived from probable content, they must transition to the systematic application of statistical tools. These include methods concerning the frequency of appearance of each symbol, the order in which these symbols typically appear, whether some symbols appear at the beginning or end of words, etc. There are situations where orthographic features of a language make it difficult if not impossible to decipher specific features (especially without certain outside information), such as when an alphabet does not express double consonants. Additional, and more complex methods, also exist. Eventually, the application of such statistical methods becomes exceedingly laborious, in which computers might be used to apply them automatically.<ref name=":1" />

=== Computational approaches ===
Computational approaches towards the decipherment of unknown languages began to appear in the late 1990s.<ref>{{Cite journal |last1=Knight |first1=Kevin |last2=Yamada |first2=Kenji |date=1999 |title=A Computational Approach to Deciphering Unknown Scripts |url=https://aclanthology.org/W99-0906.pdf |journal=Unsupervised Learning in Natural Language Processing}}</ref> Typically, there are two types of computational approaches used in language decipherment: approaches meant to produce translations in known languages, and approaches used to detect new information that might enable future efforts at translation. The second approach is more common, and includes things such as the detection of cognates or related words, discovery of the closest known language, word alignments, and more.<ref name=":2" />

=== Artificial intelligence ===
In recent years, there has been a growing emphasis on methods utilizing [[artificial intelligence]] for the decipherment of lost languages, especially through [[natural language processing]] (NLP) methods. Proof-of-concept methods have independently re-deciphered [[Ugaritic alphabet|Ugaritic]] and [[Linear B]] using data from similar languages, in this case [[Hebrew alphabet|Hebrew]] and [[Ancient Greek]].<ref>{{Cite book |last1=Luo |first1=Jiaming |last2=Cao |first2=Yuan |last3=Barzilay |first3=Regina |title=Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics |date=2019 |chapter=Neural Decipherment via Minimum-Cost Flow: From Ugaritic to Linear B |language=en |publisher=Association for Computational Linguistics |pages=3146–3155 |doi=10.18653/v1/P19-1303|arxiv=1906.06718 }}</ref>