In my last post I wrote about the apparent ‘problems’ in how the prehistoric Linear B script is used to write the Mycenaean Greek language, and how these are actually not ‘problems’ at all, but a compromise between accurate representation of the language and economy in the number of different signs in the writing signs – as demonstrated by the use of very similar orthographic strategies in how the modern Cherokee syllabary represents the Cherokee language. Today I want to look in more detail at how Mycenaean writers actually used the Linear B orthographic system, and what this can tell us about both their attitudes towards ‘correct’ spelling and the way(s) in which they were taught to spell in the first place.
It’s often remarked that many features of the Linear B writing system are remarkably consistent, despite the fact that it was used at different sites across Crete and southern/central mainland Greece, and over a period of at least 200 years (the texts that survive date from various points between c.1400 and c.1200 BCE). In general, this applies to spelling too: the system by which Linear B – a syllabary, all of whose 87 signs represented open syllables (ending in a vowel) – represented Mycenaean Greek (a language containing many sequences of two consonants, and word-final consonants, neither of which could be straightforwardly written with open syllabic signs) is extremely consistent despite its complexity. Broadly speaking, there are two ways to deal with the issue of needing to represent consonants that are not followed by a vowel with signs that all represent consonant-vowel sequences: 1) omit the consonant; 2) write the consonant, with an extra (‘dummy’) vowel.
We can see both of these in the Linear B spelling of the word tripodes ‘tripods’ as ti-ri-po-de: the initial tri- is written ti-ri-, with a ‘dummy’ vowel that’s the same as the following ‘real’ vowel, while the final –s is not written at all; similarly in the word pa-si = pansi ‘for all/everyone’, the n before the s is omitted. There are very consistent conventions about when the first type of spelling (‘plene’, or ‘full’ spelling) is used, and when the second (‘partial’) spelling is used – e.g. /tr/ (and similar clusters like /pr/ or /dr/) are always written in full, while /ns/ is always spelled partially). The full list is very long and complicated (and it’s difficult to explain exactly why the different types of spellings are used in each case), so I won’t go into that here, because what I’m interested in is not the system’s general consistency, but the points where it actually allows spelling variation, and what Mycenaean writers chose to do in those situations.
Most of the potential for spelling variation comes from the existence of two sets of signs within the syllabary: the basic or ‘core’ signs, standing for the 5 vowels and for series of single consonants plus those vowels, and the ‘extra’ signs, which have a range of functions. Some can be used in place of a single core sign to be more specific (e.g., there is no core series for the consonant /h/, which is therefore not usually written, so /ha/ can be written just with the core sign a – but the extra sign a2 represents specifically /ha/). Others can be used in place of a sequence of two core signs to represent a consonant cluster – e.g. /dwo/ could be spelt out in plene spelling as do-wo or du-wo (here there are two plene options: the presence of the w enables the use of u as a dummy vowel) or with the single extra sign dwo.
In principle, then, there’s the potential for variation between two or three different spellings of particular sequences – but how does this variation occur in practice? Do different writers prefer different spellings, or are particular spellings preferred for specific words? It’s previously been argued by the Mycenologist Yves Duhoux that the distribution of these various spellings in the texts from Pylos – the vast majority of which are certainly contemporary with each other (dating to the palace’s final destruction c.1200 BCE), so we know the variation is occuring within a single community of writers – is due to the existence of different training groups who learned different preferred spellings; however, my own recent analysis shows a rather different, and more complicated, picture.
In some cases, the possible variation rarely if ever actually occurs: the sign pte, for instance, is pretty consistently used for the sequence /pte/, with no secure evidence of the equivalent plene spelling pe-te ever being used instead (unlike at, e.g., Knossos, where the two spellings alternate): this looks like a group preference amongst the Pylos scribes.
In other cases, there still seems to be a preference, but it’s a less strong one: the use of the sign a2 for /ha/ is very frequent, but there are (at least) a few identifiable examples of the sign a used to represent this same sequence – and nearly all of those are written by scribes who’ve also used a2, while there are almost no examples of scribes with large numbers of examples of a2 who haven’t also used a.
So although there’s still a preference, some degree of variation seems to be normal even for individual writers – and there might well be more variation if it was easier to identify examples of a spelling /ha/. The ambiguity of the core signs clearly wasn’t a problem for Mycenaean scribes but it often is for Mycenologists! Finally, there are cases where variation between two or more orthographic options is entirely regular, with no preferences on display either overall or in the work of individual writers: for instance, this applies to all the sequences of consonant+w+vowel such as /dwo/.
As you can see from the examples above, not only is it very common for a single writer to use multiple spellings of the same sequence, this also often occurs in the same words. So this variation isn’t due to different (groups of) writers being taught or preferring different spellings, or to certain words having different correct/preferred spellings for the same sequence of sounds (as in the English orthographic system!). Rather, the Pylos writers’ training must have included learning that certain syllables could be spelled in multiple ways, and that such variation was entirely acceptable and normal: at different times, then, any individual writer might make different choices about which way they preferred to spell those syllables.
The process of decision-making behind any individual spelling choice is, of course, mostly invisible to us, but we can find traces of it where a legible erasure on a tablet shows that the writer first picked one spelling and then changed their mind and replaced it with another. For instance, in one text from Pylos the writer appears to have started writing the word diphtheraphoros ‘hide-carrier ‘as di-pe– and then replaced the pe with a pte, the spelling preferred over pe-te. In rare moments like these, we can see the kinds of decisions the writers must have made all the time – but usually left no trace of – as they created the Linear B texts.
For readers interested to know more, my article on this topic, entitled “Learning to spell in Linear B: orthography and training in Mycenaean Pylos”, is forthcoming with the Cambridge Classical Journal, and you can read an open-access preprint here. The article by Yves Duhoux which I referred to (‘The teaching of orthography in Mycenaean Pylos’, Kadmos 25.2, 1986) is available here. My recent article on “Scribes as editors: tracking changes in the Linear B documents” (American Journal of Archaeology 124.4, 2020), which includes a section on changes to spelling, can be read here.
The Linear B font used in the example images is ‘Alphabetum‘, created by Juan-José Marcos and used under licence.