DOCSLIB.ORG

A Computational Approach to Deciphering Unknown Scripts

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
A Computational Approach to Deciphering Unknown Scripts A Computational Approach to Deciphering Unknown Scripts Kevin Knight Kenji Yamada USC/Information Sciences Institute USC/Information Sciences Institute 4676 Admiralty Way 4676 Admiralty Way Marina del Rey, CA 90292 Marina del Rey, CA 90292 [email protected] [email protected] Abstract • minimal input (monolingual inscriptions We propose and evaluate computational techniques only) for deciphering unknown scripts. We focus on the case in which an unfamiliar script encodes a known This situation has arisen in many famous language. The decipherment of a brief document cases of decipherment--for example, in the Lin- or inscription is driven by data about the spoken ear B documents from Crete (which turned language. We consider which scripts are easy or hard out to be a "non-Greek" script for writing an- to decipher, how much data is required, and whether cient Greek) and in the Mayan documents from the techniques are robust against language change Mesoamerica. Both of these cases lay unsolved over time. until the latter half of the 20th century (Chad- 1 Introduction wick, 1958; Coe, 1993). In computational linguistic terms, this de- With surprising frequency, archaeologists dig up cipherment task is not really translation, but documents that no modern person can read. rather text-to-speech conversion. The goal of Sometimes the written characters are familiar the decipherment is to "make the text speak," (say, the Phoenician alphabet), but the lan- after which it can be interpreted, translated, guage is unknown. Other times, it is the reverse: etc. Of course, even after an ancient docu- the written script is unfamiliar but the language ment is phonetically rendered, it will still con- is known. Or, both script and language may be tain many unknown words and strange con- unknown. structions. Making the text speak is therefore Cryptanalysts also encounter unreadable doc- only the beginning of the story, but it is a cru- uments, but they try to read them anyway. cial step. With patience, insight, and computer power, Unfortunately, current text-to-speech sys- they often succeed. Archaeologists and lin- tems cannot be applied directly, because guists known as epigraphers apply analogous they require up front a clearly specified techniques to ancient documents. Their deci- sound/writing connection. For example, a sys- pherment work can have many resources as in- tem designer may create a large pronunciation put, not all of which will be present in a given dictionary (for English or Chinese) or a set of case: (1) monolingual inscriptions, (2) accom- manually constructed character-based pronun- panying pictures or diagrams, (3) bilingual in- ciation rules (for Spanish or Italian). But in scriptions, (4) the historical record, (5) physical decipherment, this connection is unknown! It is artifacts, (6) bilingual dictionaries, (7) informal exactly what we must discover through analysis. grammars, etc. There are no rule books, and literate informants In this paper, we investigate computational are long-since dead. approaches to deciphering unknown scripts, and report experimental results. We concentrate on 2 Writing Systems the following case: To decipher unknown scripts, is useful to under- stand the nature of known scripts, both ancient • unfamiliar script and modern. Scholars often classify scripts into • known language three categories: (1) alphabetic, (2) syllabic, 37 purely logographic; it is sometimes called mor- phophonemic. Ancient writing is also mixed, and archaeologists frequently observe radical writing changes in a single language over time. 3 Experimental Framework In this paper, we do not decipher any ancient scripts. Rather, we develop algorithms and ap- ply them to the "decipherment" of known, mod- ern scripts. We pretend to be ignorant of the connection between sound and writing. Once our algorithms have come up with a proposed phonetic decipherment of a given document, we route the sound sequence to a speech synthe- sizer. If a native speaker can understand the speech and make sense of it, then we consider the decipherment a success. (Note that the na- Figure 1: The Phaistos Disk (c. 1700BC). The tive speaker need not even be literate, theoreti- disk is six inches wide, double-sided, and is the cally). We experiment with modern writing sys- earliest known document printed with a form of tems that span the categories described above. movable type. We are interested in the following questions: .~ and (3) log6graphic (Sampson, 1985). • Can automatic techniques decipher an un- known script? If so, how accurately? Alphabetic systems attempt to repre- • What quantity of written text is needed sent single sounds with single characters, for successful decipherment? (this may be though no system is "perfect." For exam- • quite limited by circumstances) ple, Semitic alphabets have no characters for vowel sounds. And even highly regular • What knowledge of the spoken language is writing systems like Spanish have plenty of needed? Can it to be extracted automati- spelling variation, as we shall see later. cally from available resources? What quan- tity of resources? Syllabic systems have characters for entire syllables, such as "ba" and "shu." Both • Are some writing systems easier to decipher Linear B and Mayan are primarily syllabic, than others? Are there systematic differ- as is Japanese kana. The Phaistos Disk ences among alphabetic, syllabic, and lo- from Crete (see Figure 1) is thought to be gographic systems? syllabic, because of the number of distinct • Are word separators necessary or helpful? characters present. • Can automatic techniques be robust Finally, logographic systems have charac- against language evolution (e.g., modern ters for entire words. Chinese is often cited versus ancient forms of a language)? as a typical example. • Can automatic techniques identify the lan- Unfortunately, actual scripts do not fall guage behind a script as a precursor to de- neatly into one category or another (DeFrancis, ciphering it? 1989; Sproat, forthcoming). Written Japanese will contain syllabic kana, alphabetic roomaji, 4 Alphabetic Writing (Spanish) and logographic kanji characters all in the same Five hundred years ago, Spaniards invaded document. Chinese characters actually have a Mayan lands, burning documents and effec- phonetic component, and words are often com- tively eliminating everyone who could read and posed of more than one character. Irregular write. (Modern Spaniards will be quick to point English writing is neither purely alphabetic nor out that most of the work along those lines 38 Sounds: B-+borv B, D, G, J (ny as in canyon), L (y as D~d in yarn), T (th as in thin), a, b, d, e, G~g f, g, i, k, l, m, n, o, p, r, rr (trilled), s, J--+fi t, tS (ch as in chin), u, x (h as in hat) L--+ llory a .--~ a or £ Characters: b --.~. b or v d--~d fi, £, 6, i, o, u, a, b, c, d, e, f, g, h, i, j, e ---~ e or 6 k, l, m, n, o, p, q, r, s, t, u, v, w, x, y, f~f Z gag i~ior~ Figure 2: Inventories of Spanish sounds (with l~l rough English equivalents in parentheses) and m-+m characters. n--+n o-moor6 p--~p had already been carried out by the Aztecs). r--Jr Mayan hieroglyphs remained uninterpreted for many centuries. We imagine that if the Mayans t--rt tS~ch have invaded Spain, then 20th century Mayan scholars might be deciphering ancient Spanish u--~ u or fi x-+j documents instead. We begin with an analysis of Spanish writing. nothing --+ h T (followed by a, o, or u) ~ z The task of decipherment will be to re-invent these rules and apply them to written docu- T (followed by e or i) --+ c or z T (otherwise) ~ c ments in reverse. First, is necessary to settle on the basic inventory of sounds and characters. k (followed by e or i) ~ q u k (followed by s) ---+ x Characters are easy; we simply tabulate the dis- tinct ones observed in text. For sounds, we need k (otherwise) ~ c something that will serve as reasonable input to rr (at beginning of word) ~ r rr (otherwise) ---,, rr a speech synthesizer. We use a Spanish-relevant subset of the International Phonetic Alphabet s (preceded by k) ~ nothing s (otherwise) --+ s (IPA), which seeks to capture all sounds in all languages. Actually, we use an ASCII version of the IPA called SAMPA (Speech Assessment Figure 3: Spanish sound-to-character spelling Methods Phonetic Alphabet), originally devel- rules. The left-hand side of each rule contains a oped under ESPRIT project 1541. There is also Spanish sound (and possible conditions), while a public-domain Castillian speech synthesizer theright-hand side contains zero or more Span- (called Mbrola) for the Spanish SAMPA sound ish characters. set. Figure 2 shows the sound and character inventories. • silence can produce a character (h) Now to spelling rules. Spanish writing is clearly not a one-for-one proposition: Moreover, there are ambiguities. The sound L • a single sound can produce a single charac- (English y-sound) may be written as either ll or ter (a-+ a) y. The sound i may also produce the character y, so the pronunciation of y varies according to * a sound can produce two characters (tS context. The sound rr (trilled r) is written rr in ch) the middle of a word and r at the beginning of • two sounds can produce a single character a word. (k s ---+ x) Figure 3 shows a sample set of Spanish 39 spelling rules. We formalized these rules com- Armed with these basic probabilities, we can putationally in a finite-state transducer (Pereira compute two things. First, the total probabil- and Riley, 1997). The transducer is bidirec- ity of observing a particular written sequence of tional. Given a specific sound sequence, we characters cl ..
Load more

Recommended publications
  • Talking Stone: Cherokee Syllabary Inscriptions in Dark Zone Caves
    University of Tennessee, Knoxville TRACE: Tennessee Research and Creative Exchange Masters Theses Graduate School 12-2017 Talking Stone: Cherokee Syllabary Inscriptions in Dark Zone Caves Beau Duke Carroll University of Tennessee, [email protected] Follow this and additional works at: https://trace.tennessee.edu/utk_gradthes Recommended Citation Carroll, Beau Duke, "Talking Stone: Cherokee Syllabary Inscriptions in Dark Zone Caves. " Master's Thesis, University of Tennessee, 2017. https://trace.tennessee.edu/utk_gradthes/4985 This Thesis is brought to you for free and open access by the Graduate School at TRACE: Tennessee Research and Creative Exchange. It has been accepted for inclusion in Masters Theses by an authorized administrator of TRACE: Tennessee Research and Creative Exchange. For more information, please contact [email protected]. To the Graduate Council: I am submitting herewith a thesis written by Beau Duke Carroll entitled "Talking Stone: Cherokee Syllabary Inscriptions in Dark Zone Caves." I have examined the final electronic copy of this thesis for form and content and recommend that it be accepted in partial fulfillment of the requirements for the degree of Master of Arts, with a major in Anthropology. Jan Simek, Major Professor We have read this thesis and recommend its acceptance: David G. Anderson, Julie L. Reed Accepted for the Council: Dixie L. Thompson Vice Provost and Dean of the Graduate School (Original signatures are on file with official studentecor r ds.) Talking Stone: Cherokee Syllabary Inscriptions in Dark Zone Caves A Thesis Presented for the Master of Arts Degree The University of Tennessee, Knoxville Beau Duke Carroll December 2017 Copyright © 2017 by Beau Duke Carroll All rights reserved ii ACKNOWLEDGMENTS This thesis would not be possible without the following people who contributed their time and expertise.
    [Show full text]
  • Writing Systems Reading and Spelling
    Writing systems Reading and spelling Writing systems LING 200: Introduction to the Study of Language Hadas Kotek February 2016 Hadas Kotek Writing systems Writing systems Reading and spelling Outline 1 Writing systems 2 Reading and spelling Spelling How we read Slides credit: David Pesetsky, Richard Sproat, Janice Fon Hadas Kotek Writing systems Writing systems Reading and spelling Writing systems What is writing? Writing is not language, but merely a way of recording language by visible marks. –Leonard Bloomfield, Language (1933) Hadas Kotek Writing systems Writing systems Reading and spelling Writing systems Writing and speech Until the 1800s, writing, not spoken language, was what linguists studied. Speech was often ignored. However, writing is secondary to spoken language in at least 3 ways: Children naturally acquire language without being taught, independently of intelligence or education levels. µ Many people struggle to learn to read. All human groups ever encountered possess spoken language. All are equal; no language is more “sophisticated” or “expressive” than others. µ Many languages have no written form. Humans have probably been speaking for as long as there have been anatomically modern Homo Sapiens in the world. µ Writing is a much younger phenomenon. Hadas Kotek Writing systems Writing systems Reading and spelling Writing systems (Possibly) Independent Inventions of Writing Sumeria: ca. 3,200 BC Egypt: ca. 3,200 BC Indus Valley: ca. 2,500 BC China: ca. 1,500 BC Central America: ca. 250 BC (Olmecs, Mayans, Zapotecs) Hadas Kotek Writing systems Writing systems Reading and spelling Writing systems Writing and pictures Let’s define the distinction between pictures and true writing.
    [Show full text]
  • UC Santa Barbara Previously Published Works
    UC Santa Barbara UC Santa Barbara Previously Published Works Title Scriptworlds Permalink https://escholarship.org/uc/item/41d2f9kq Author Park, Sowon Publication Date 2021-06-28 Peer reviewed eScholarship.org Powered by the California Digital Library University of California Scriptworlds Sowon Park Script and world Thinking about literature in relation to ‘world’ tends to invite the broad sweep. But the widened scope needn’t take in only the majestic, as conjured up in lofty concepts such as planetarity or world-historical totality. Sometimes the scale of world can produce a multifocal optic that helps us pay renewed attention to the literary commonplace. Taking script, the most basic component of literature, as one of the lenses through which to view the world literary landscape, this chapter will examine if, and how, ideas about writing can influence both our close and distant reading. Script is something that usually escapes notice under traditional classifications of literature, which organize subjects along linguistic borders, whether as English, French, or Spanish, or as Anglophone, Francophone, or Hispanophone studies. The construction of literatures by discrete language categories supports a wide-spread view that differences between languages are not just a matter of different phonemes, lexemes and grammars but of distinct ways of conceiving, apprehending and relating to the world. This was a view that was developed into a political stance in Europe in the nineteenth century.1 It posited a language as an embodiment of cultural or national distinctiveness, and a literature written in a national language as the sovereign expression of a particular worldview. One of the facets this stance obscured was the script in which the texts were written.
    [Show full text]
  • Writing Systems: Their Properties and Implications for Reading
    Writing Systems: Their Properties and Implications for Reading Brett Kessler and Rebecca Treiman doi:10.1093/oxfordhb/9780199324576.013.1 Draft of a chapter to appear in: The Oxford Handbook of Reading, ed. by Alexander Pollatsek and Rebecca Treiman. ISBN 9780199324576. Abstract An understanding of the nature of writing is an important foundation for studies of how people read and how they learn to read. This chapter discusses the characteristics of modern writing systems with a view toward providing that foundation. We consider both the appearance of writing systems and how they function. All writing represents the words of a language according to a set of rules. However, important properties of a language often go unrepresented in writing. Change and variation in the spoken language result in complex links to speech. Redundancies in language and writing mean that readers can often get by without taking in all of the visual information. These redundancies also mean that readers must often supplement the visual information that they do take in with knowledge about the language and about the world. Keywords: writing systems, script, alphabet, syllabary, logography, semasiography, glottography, underrepresentation, conservatism, graphotactics The goal of this chapter is to examine the characteristics of writing systems that are in use today and to consider the implications of these characteristics for how people read. As we will see, a broad understanding of writing systems and how they work can place some important constraints on our conceptualization of the nature of the reading process. It can also constrain our theories about how children learn to read and about how they should be taught to do so.
    [Show full text]
  • Proposal for Ethiopic Script Root Zone LGR
    Proposal for Ethiopic Script Root Zone LGR LGR Version 2 Date: 2017-05-17 Document version:5.2 Authors: Ethiopic Script Generation Panel Contents 1 General Information/ Overview/ Abstract ........................................................................................ 3 2 Script for which the LGR is proposed ................................................................................................ 3 3 Background on Script and Principal Languages Using It .................................................................... 4 3.1 Local Languages Using the Script .............................................................................................. 4 3.2 Geographic Territories of the Language or the Language Map of Ethiopia ................................ 7 4 Overall Development Process and Methodology .............................................................................. 8 4.1 Sources Consulted to Determine the Repertoire....................................................................... 8 4.2 Team Composition and Diversity .............................................................................................. 9 4.3 Analysis of Code Point Repertoire .......................................................................................... 10 4.4 Analysis of Code Point Variants .............................................................................................. 11 5 Repertoire ....................................................................................................................................
    [Show full text]
  • THIS Winitkbagk) SYLLABARY
    Amelia Louise Susman Schultz, Sam Blowsnake, and Ho-Chunk Auto-Literacy1 In June 103 years ago, Amelia was born in New York City into a German Jewish family, sharing the Brooklyn home of tante, a great aunt who was a seamstress. At age six, she started grade school at PS 44, skipped 2 grades, and graduated as valedictorian of James Marshall High School at 14. Joseph Greenberg shared her subway ride, sometimes conversing in Latin with a school mate. Since she was just under five feet tall, she was ineligible to work at Macys, the usual graduation option for women until marriage. Instead, she enrolled in Brooklyn College, 1931-1935, gaining an AB degree in Psychology with Solomon Asch while earning $20/month through National Youth Administration (NYA) of the CCC. Seeking a graduate degree, she was warned by classmate Irving Goldman that only Columbia anthropology under Franz Boas willingly accepted women into its graduate program. Most other disciplines were strictly men only. Accepted at Columbia University, 1935-1939, she wrote two dissertations. The first was a study of acculturation at multitribal Round Valley in northern California, relying on documents in the San Francisco archive more than interviews with these intertribals whose traumas included genocide, girl slavery, and land loss to the ranching family of their BIA agent. Her work was part of a comparative team study in eight tribes, eventually published as Acculturation in Seven American Indian Tribes (Marian Smith = Puyallup; Jack Harris = White Knife Shoshoni; Marvin Opler = Southern Ute; Henry Elkin = Northern Arapaho; Natalie Joffe = Fox; Irving Goldman = Carrier; and William Whitman = San Ildefonso).
    [Show full text]
  • Writing Language
    Writing language Linguists generally agree with the following statement by one of the founders of the modern science of language. Writing is not language, but merely a way of recording language by visible marks. Leonard Bloomfield, Language (1933) Some version of this is clearly true, as we can see by looking at the history of the human species and of each human individual. In both regards, spoken language precedes written language. Speech Writing Present in every society Present only in some societies, and only rather recently Learned before writing Learned after speech is acquired Learned by all children in normal Learned only by instruction, and often not circumstances, without instruction learned at all Human evolution has made speaking Evolution has not specifically favored easier writing Another way to express Bloomfield's point is to say that writing is "parasitic" on speech, expressing some but not all of the things that speech expresses. Specifically, writing systems convey the sequence of known words or other elements of a language in a real or hypothetical utterance, and indicate (usually somewhat less well) the pronunciation of words not already known to the reader. Aspects of speech that writing leaves out can include emphasis, intonation, tone of voice, accent or dialect, and individual characteristics. Some caveats are in order. In the first place, writing is usually not used for "recording language" in the sense of transcribing speech. Writing may substitute for speech, as in a letter, or may deploy the expressive resources of spoken language in visual structures (such as tables) that can't easily be replicated in spoken form at all.
    [Show full text]
  • 2 Hangul Jamo Auxiliary Canonical Decomposition Mappings
    DRAFT Unicode technical note NN Auxiliary character decompositions for supporting Hangul Kent Karlsson 2006-09-24 1 Introduction The Hangul script is very elegantly designed. There are just a small number of letters (28, plus a small number of variant letters introduced later, but the latter have fallen out of use) and even a featural design philosophy for the shapes of the letters. However, the incarnation of Hangul as characters in ISO/IEC 10646 and Unicode is not so elegant. In particular, there are many Hangul characters that are not needed, for precomposed letter clusters as well as precomposed syllable characters. The precomposed syllables have arithmetically specified canonical decompositions into Hangul jamos (conjoining Hangul letters). But unfortunately the letter cluster Hangul jamos do not have canonical decompositions to their constituent letters, which they should have had. This leads to multiple representations for exactly the same sequence of letters. There is not even any compatibility-like distinction; i.e. no (intended) font difference, no (intended) width difference, no (intended) ligaturing difference of any kind. They have even lost the compatibility decompositions that they had in Unicode 2.0. There are also some problems with the Hangul compatibility letters, and their proper compatibility decompositions to Hangul jamo characters. Just following their compatibility decompositions in UnicodeData.txt does not give any useful results in any setting. In this paper and its two associated datafiles these problems are addressed. Note that no changes to the standard Unicode normal forms (NFD, NFC, NFKD, and NFKC) are proposed, since these normal forms are stable for already allocated characters.
    [Show full text]
  • A STUDY of WRITING Oi.Uchicago.Edu Oi.Uchicago.Edu /MAAM^MA
    oi.uchicago.edu A STUDY OF WRITING oi.uchicago.edu oi.uchicago.edu /MAAM^MA. A STUDY OF "*?• ,fii WRITING REVISED EDITION I. J. GELB Phoenix Books THE UNIVERSITY OF CHICAGO PRESS oi.uchicago.edu This book is also available in a clothbound edition from THE UNIVERSITY OF CHICAGO PRESS TO THE MOKSTADS THE UNIVERSITY OF CHICAGO PRESS, CHICAGO & LONDON The University of Toronto Press, Toronto 5, Canada Copyright 1952 in the International Copyright Union. All rights reserved. Published 1952. Second Edition 1963. First Phoenix Impression 1963. Printed in the United States of America oi.uchicago.edu PREFACE HE book contains twelve chapters, but it can be broken up structurally into five parts. First, the place of writing among the various systems of human inter­ communication is discussed. This is followed by four Tchapters devoted to the descriptive and comparative treatment of the various types of writing in the world. The sixth chapter deals with the evolution of writing from the earliest stages of picture writing to a full alphabet. The next four chapters deal with general problems, such as the future of writing and the relationship of writing to speech, art, and religion. Of the two final chapters, one contains the first attempt to establish a full terminology of writing, the other an extensive bibliography. The aim of this study is to lay a foundation for a new science of writing which might be called grammatology. While the general histories of writing treat individual writings mainly from a descriptive-historical point of view, the new science attempts to establish general principles governing the use and evolution of writing on a comparative-typological basis.
    [Show full text]
  • ONIX for Books Codelists Issue 40
    ONIX for Books Codelists Issue 40 23 January 2018 DOI: 10.4400/akjh All ONIX standards and documentation – including this document – are copyright materials, made available free of charge for general use. A full license agreement (DOI: 10.4400/nwgj) that governs their use is available on the EDItEUR website. All ONIX users should note that this is the fourth issue of the ONIX codelists that does not include support for codelists used only with ONIX version 2.1. Of course, ONIX 2.1 remains fully usable, using Issue 36 of the codelists or earlier. Issue 36 continues to be available via the archive section of the EDItEUR website (http://www.editeur.org/15/Archived-Previous-Releases). These codelists are also available within a multilingual online browser at https://ns.editeur.org/onix. Codelists are revised quarterly. Go to latest Issue Layout of codelists This document contains ONIX for Books codelists Issue 40, intended primarily for use with ONIX 3.0. The codelists are arranged in a single table for reference and printing. They may also be used as controlled vocabularies, independent of ONIX. This document does not differentiate explicitly between codelists for ONIX 3.0 and those that are used with earlier releases, but lists used only with earlier releases have been removed. For details of which code list to use with which data element in each version of ONIX, please consult the main Specification for the appropriate release. Occasionally, a handful of codes within a particular list are defined as either deprecated, or not valid for use in a particular version of ONIX or with a particular data element.
    [Show full text]
  • Chapter One Chapter
    M01_HULI7523_05_SE_C01.QXD 12/4/09 11:37 AM Page 1 CHAPTER ONE A Connection of Brains C HAPTER O BJECTIVES This first chapter is designed to pique your facilitate understanding of the following interest in speech and language as pro- topics: cesses within the broader process we call • Definitions of communication, language, communication. We define and describe and speech these processes, and we consider how • Speech and language as separate but speech and language interact to produce a related processes form of communication unique to humans. We also consider how a speaker’s thoughts • Design features of the human communi- are conveyed to a listener’s brain through a cation system series of communication transformations • The speech chain that connects the known collectively as the speech chain. speaker’s thoughts to the listener’s Specifically, this chapter is designed to understanding of those thoughts 1 M01_HULI7523_05_SE_C01.QXD 12/4/09 11:37 AM Page 2 2 Chapter 1 he idea for Born to Talk was cultivated long before the first word of the original Tmanuscript was written, and it was probably a good thing that there was a period of latency between the concept and the product. During that latency, I* observed lan- guage development firsthand in my two daughters, Yvonne and Carmen. I learned more about the power and wonder of language in observing them than I have in all the books and all the journal articles I have read over the course of my career because I witnessed their processes of discovery. I watched and listened as they made con- nections between the world in which they were growing up and the words and lan- guage forms that spilled out of them.
    [Show full text]
  • Cover Page the Handle
    Cover Page The handle http://hdl.handle.net/1887/42753 holds various files of this Leiden University dissertation Author: Moezel, K.V.J. van der Title: Of marks and meaning : a palaeographic, semiotic-cognitive, and comparative analysis of the identity marks from Deir el-Medina Issue Date: 2016-09-07 INTRODUCTION vii viii DEFINING VISUAL COMMUNICATION WE TEND TO think of communication in terms of speech and writing: speech as the oral expression of language and writing as its graphic expression.1 This thought can be traced back to ancient Greece, particularly to a statement written by Aristotle around 350 BC: ‘Spoken words are the symbols of mental experience and written words are the symbols of spoken words’2 This statement was understood in the sense that spoken signs are the key to language systems, while written signs are merely their representations, serving only to their needs. Even though Aristotle had not meant to say this,3 writing thus came to be considered written speech: spoken language recorded by marks the basic function of which is phonoptic.4 Western scholars throughout the Middle Ages and the Early Modern Period held the view of writing as a visible surrogate or substitute for speech. This view, called ‘the surrogational model’ by Harris,5 was prevailing especially in the 18th and 19th centuries. The French missionary De Brébeuf, for instance, in translating the poem Pharsalia by Lucan, spoke of ‘cet art ingénieux – de peindre la parole et de parler aux yeux’.6 In similar fashion Voltaire’s famous quote reads ‘l’écriture est la peinture de la voix : plus elle est ressemblante, meilleure elle est.’7 The Irish poet Trench in 1855 spoke of ‘representing sounds by written signs, of reproducing for the eye that which existed at first only for the ear’; and ‘The intention of the written word’ is ‘to represent to the eye with as much accuracy as possible the spoken word.’8 The surrogational view was still popular in much of the 20th century.
    [Show full text]