Lecture: Deciphering scripts: Narrator: Listen to part of a lecture in an anthropology class. Professor: Archaeologists sometimes discover an unknown, ancient written script, and they need to decipher or figure out the meaning of this script in order to read and understand it. And now a script is not the same as a language, right? A script is a written symbolic communication system that's associated with a spoken language. So the written symbols of a script can be used to represent the words or ideas of a language or languages. Now deciphing an ancient script is interesting to anthropologists, because the script can both help us to reconstruct the language of a civilization and give us insight into that civilizations' customs and daily life. But traditionally deciphing a script takes a long time. However, a new computer program may help speed the process. To understand just how much, let's first look at the traditional approach to deciphing one script, a script that was deciphered only recently in the 20th century – Ugaritic. Tablets with Ugaritic writing on them were found in Syria in the late 1920s. Now, when faced with an unknown script, you'd ask questions like, first, what type of script is this? There are two main types, either, well, each symbol represents the whole idea or word, or each symbol represents a sound or a combination of sounds, the way an alphabet does. In the case of Ugaritic, only thirty symbols were used on all the tablets, suggesting the symbols were alphabetic. You'll also ask whether the language itself is related to any other known language. And because of where these tablets were found and their age, we thought Ugaritic might be related to ancient Hebrew. This is important because we can use the kown script as a clue to the unknown script. So if we know a script alphabetic, we try to map its symbols onto the alphabet of the kown script. But this is no simple task. You have to figure out pieces one by one. You may start out by mapping the unknown alphabet onto the known alphabet, only to later learn it was an incorrect mapping, or you may just get stuck. And the study of Ugaritic was heading in that direction until the discovery of five axes with writing on them. Each one had the same combination of six symbols. And the fifth axe also had an additional four symbols. Researchers guess the written symbols common to all the access would denote a name. But the fifth axe, they believed the four symbols that preceded the main would spell out the word AXE in Ugaritic, what followed, took a lot of work. But it turns out the ancient Hebrew word for AXE has the same number of symbols and is quite similar. These four symbols led to further discoveries and now we're mostly able to read Ugaritic. So this traditional approach can work, but it's time consuming. Now recently, computer scientists devised a program meant to decipher unknown scripts. They tested the computer program with Ugaritic to see if the programs decipherment matched the human one. Amazingly, the program deciphered Ugaritic in just a few hours, compared that two years it took humans. Ok, the program itself, it works by quickly comparing the new script to a kown script. So what exactly does it compare? Well, its program to make certain assumptions about the features of the two scripts and languages. First, it assumes related letters or symbols in the scripts will appear about the same number of times with a similar frequency. And the second assumption concerns word structure, um, the way words are built. Corresponding words in the related language might also have a similar structure. So the system uses these assumptions together to attend to decipher the script. Student: It's kind of limited, though, right? Because the program can't decipher a script unless there's a kown script to compare it to. Professor: It is. But remember, that's a limitation, no matter the approach. If there's no related language to compare it to, a script simply can't be deciphered. Take a Etruscan, Etruscan dates to around 700 B.C.E. in what's present day Italy. It isn't related to any known language as far as we know. So we can see the traditional approach won't be successful. On the other hand, the computer program, well, it should soon be able to scan multiple scripts at a time to look for similarities. And so it's entirely possible that the program itself might someday discover a previously unknown relationship between two languages.