Decipherment and the Relation between
Mayan Languages and Maya Writing
NICHOLAS A. HOPKINS
FLORIDA STATE UNIVERSITY
Hopkins begins with a discussion of methods of decipherment and concludes with arguments for
an alternate interpretation of a sign read hi by Stuart (1987). Ten years later, the majority of epigraphers
accept Stuart's reading. Nevertheless, Hopkins raises important points relating to the evaluation of
proposed decipherments. And, ten years later, consensus among epigraphers remains greatest for those
signs appearing in a variety of environments, especially those used to spell noun and verb roots, and
‘weakest for those signs indicating grammatical morphemes, particularly verbal affixes.
Disagreement arises from uncertainty about the nature of verb morphology in lowland Mayan
languages during the Classic Period. Noun and verb roots tend to remain stable though time and across
related languages and dialects; the uses of morphological affixes are more varied. Likewise, the
principle of synharmony of phonetic signs, which operates in virtually all syllabic writing systems,
occurs most often in the spelling of noun and verb roots, and appears to be violated most often in words
which are morphologically complex. Continued reconstruction of proto-Ch’olan and proto-Yucatecan
morphological systems, coupled with detailed examination of patterns of syllabic spellings in the
glyphic texts, can provide valuable information on Classic Period verb morphology.
‘A continuing problem is uncertainty about how Maya scribes used syllabic signs, particularly
various combinations of vowels, glottal stops, and y. When a glyphic spelling appears to violate
synharmony—when a CVC word is spelled with two CV syllabic signs, and the second sign has a
vocalic value different from the first—epigraphers differ in their interpretations. Hopkins cautions that
uncritically ignoring the value of the second vowel in these variant spellings is to risk loosing valuable
information.
Decipherment is the interpretation of
hieroglyphic writing in terms of a spoken
Janguage, the conversion of ciphers into plain
text’ One can of course interpret an inscription
without deciphering it, but if an inscription is
deciphered, it is converted into a text ina spoken
language. There are various degrees to which
this can be done. One level of decipherment is
represented by the translation, which expresses
the content of the inscription in some language
other than the original. Another level
‘encompasses glosses and paraphrases, which
translate the content of an inscription to another
language, but attempt to capture some of the
grammatical and syntactic features of the
original. A third level is that of readings, which
render the text in the language in which the
inscription was originally written.
Legitimate differences of opinion can be held
as to the content of an inscription, the nature of
the graphic system used to record it, the
grammatical and syntactic properties
exemplified in a text, and other questions of fact
Martha J. Macri
and analysis. Likewise different opinions may
be held as to what language was being recorded,
and what the precise features of that language
were at the relevant place and time.
In this paper I review some of the issues
raised by the current state of Mayan epigraphy,
especialiy attempts to assign readings to words,
phrases, or whole texts. This is an era of intense
activity in the production of proposals
concerning the language of the Maya
hieroglyphic inscriptions. Workshops and
conferences abound, and hardly an occasion
goes by that does not witness the announcement
of a newly deciphered syllable or word. In the
frenzied excitement of such a time, it is easy to
lose track of general principles of decipherment,
even deliberately to ignore caution for the sake
of squeezing out yet another decipherment.
The scientific method
Since activity is so intense, it is even more
important to keep in mind that the scientific
method and principles of decipherment do not
7exist to retard progress; they exist to make sure
that progress is real rather than illusory. Bad
methodology can be expected to produce a high
percentage of bad decipherments, decipherments
which appear to be valid only in the short run,
but which in the long run will have to be
discarded. This is not to say that even the best of
methodologies applied in the most rigorous way
will not occasionally produce incorrect
decipherments, but the ratio of good to bad
decipherments is maximized by the application
of principles which tend to screen out proposals
less likely to be true.
The most destructive aspect of bad
decipherments, that is, decipherments which are
untrue, is that they contaminate good
decipherments. They are in a sense like
computer viruses; once they get into a system
they infect everything with which they come into
contact. New decipherments depend on the
interpretation of a series of relations of
substitution and contrast. In the course of the
‘examination of these relations, hypotheses about
the possible values of a glyph are tested by
seeing if the proposed values make sense in the
contexts in which the glyph to be deciphered
occurs. If the context is badly defined because of
a previous mistake in decipherment, judgments
about the acceptability of a value being tested
will also be incorrect. An incorrect
decipherment is not only wrong in and of itself,
but it will cause other decipherments to be
wrong as well, by misleading the investigator as
to the nature of the contexts used to test
hypotheses. Thus, in the early stages of
decipherment, caution is in order not for
ideological reasons, but for the extremely
pragmatic reason that mistakes made early on
will multiply themselves throughout the entire
process of decipherment and result in a false
picture of what the writing system is like, as well
as obscuring the content of the inscriptions.
antific objectivity
In the search for correct decipherments, it
must be kept in mind that the scientific method
is concerned, not with proving hypotheses, but
with testing hypotheses. The very essence of
scientific objectivity is that just as vigorous a
search must be made for evidence that would
disprove the hypothesis as is made for evidence
that would support it. The investigator should
not be primarily concerned with demonstrating
that his or her ideas are correct; rather he or she
must first be concerned with discovering
8
whether or not that idea fits the facts, objectively
evaluated.
On the matter of objectivity and investigator
bias, it is disappointing that in vanguard works
in Maya hieroglyphics one is often left with the
impression that a kind of selective ignorance is
being practiced. The search for evidence stops
when data supporting the hypothesis are found,
and no exhaustive search is made for possible
negative evidence. One is reminded of the old
saw: “My mind is already made up; don’t
confuse me with the facts.” This sort of
approach may be acceptable in fields such as
literary criticism, where one could argue that
there is no truth to be discovered, only positions
to defend. It is specifically alien to scientific
research, in which it is assumed that there is a
truth to be discovered, and that it can only be
discovered through dispassionate, objective,
honest examination of all the evidence.
Use of sources
From the linguist’s point of view, it is
disturbing to see instances in which hypotheses
are supported by the presentation of only part of
the relevant linguistic evidence, particularly
when the interpretation of the evidence
presented seems forced in a direction favorable
to the hypotheses. The available sources of data
on Mayan languages, past and present, are by no
means easy to use. But data from these difficult
sources are routinely cited in support of
epigraphic hypotheses by investigators who
have no particular training in any of the fields
which might prepare them to make a proper
interpretation of the evidence: linguistics,
Mayan languages, ethnohistory and the use of
documentary sources, sixteenth century Spanish,
modern regional dialect of New World Spanish,
etc,
In fact, it could be argued that the popularity
of the Cordemex dictionary (Barrera Vasquez et
al. 1980) as a source for supporting data in
Mayan epigraphy is due only in part to its
extensive compilation of Mayan lexicography.
Lexical entries in the Cordemex are not analyzed
morphologically; roots are not identified and
given grammatical classification, and the glosses
supplied are phrased in a Spanish that is often
difficult to penetrate. The dictionary makes no
attempt to clarify the meaning of glosses given in
Spanish drawn from several centuries and many
regional dialects. The entries thus provide a
wonderful source of authentic Mayan data that
can easily be misinterpreted on behalf of afavorite hypothesis.
Whether this is done deliberately or
unconsciously, it is clear from instance after
instance of bent and twisted interpretations of
Mayan language data that there is a crying need
for higher standards of scholarship if we are
serious about discovering the truth of the matter
and not just playing with Maya inscriptions. We
must make a greater effort to go beyond the
limited data we have and try to understand what
the realities of the linguistic situations were
during the Classic period. It is imperative to
remember that while the limited data may make
it possible for us to sustain an unlikely argument
simply for lack of contrary data, this has little
bearing on the truth, that is, the real system used.
by Classic Maya writers.
Methodology
‘The methodology generally practiced in the
decipherment of Maya hieroglyphic writing is a
brand of structuralism, and is derived from the
standard scientific method: observation of
phenomena, formulation of hypotheses, testing
of hypotheses, and evaluation of results. From
observation, intuition, or uneasy dreams, an idea
is formed about the possible value of a glyph (for
instance, a semantic or phonetic value). In effect,
a hypothesis is formulated. The glyph, along
with whatever signs are considered its glyphic
variants, is then tracked through its occurrences
in the corpus to see if this value fits in all cases—
testing and evaluation of results. While usually
practiced in an informal way, the method is
standard inductive empirical science.
On the other hand, a strong element of
deductive reasoning is characteristic of some
research, most notable that of Fox and Justeson
(1984, and Justeson and Campbell, this volume
supra) who propose values for Maya glyphs on
the basis of the structural and statistical
characteristics of other writing systems
elsewhere, moving from the characteristics of
writing systems in general to those of Maya
writing in particular. It is certainly true that
Maya writing should obey whatever universals
pertain to writing systems in general. But one
should bear in mind that New World writing
systems were developed in total isolation from
the systems that have contributed to the
formulation of these universals. Thus, Maya
writing is potentially different. Maya writing is
essentially the test case against which these
universals can be judged, and as such it should
be analyzed independently, without biasing its
analysis towards supposed universal patterns or
trends. Otherwise we run the risk of glossing
over differences between Maya writing and
other writing systems which could lead to the
formulation of newer, more broadly applicable
generalizations.
Evaluation by consensus
In the current exploratory phase of Maya
decipherment, no fixed standards of proof have
been developed. Readings and other
interpretations are not judged right or wrong by
applying some set of standard criteria of
evaluation; rather, the acceptance of a
decipherment is more a matter of consensus. In
the best of cases, this consensus is drawn from a
broad range of widely differing viewpoints,
rather than from a small and like-minded work
group. Thus, although the procedures of
evaluation are informal, a reading or other
interpretation, in order to be adopted in general
practice, has to meet various criteria of
evaluation, since many different interests are
represented in the informal groups which confer
acceptability.
In general one is inclined to accept a reading
if and only if (1) one agrees with the facts
presented; (2) one accepts the methodology of
decipherment and the principles of application;
and most importantly, (3) the new reading is
productive, that is, it leads to further insights,
pethaps to further readings and decipherments.
Different opinions about facts may involve
either epigraphic or linguistic matters. For
instance, one may not agree that all the
epigraphic variants treated as a set are in fact
allographs of the same sign. Or, one may
disagree with the details of a grammatical or
syntactic analysis, or about the identification of
the target language. In current practice,
methodology and procedures are probably less
important than results. If one agrees with the
facts presented, and the reading is productive,
then there is a strong tendency to accept the
reading even if it was arrived at by flawed
methodology or by procedures that would
otherwise be considered unacceptable.
However, there is often disagreement and
differences of opinion regarding the principles of
writing systems in general and Maya writing in
particular. The question of what kinds of
features we should and should not expect to find
in Maya writing weighs heavily on standards of
evaluation, including the evaluation of the
validity and productivity of individual readings.
”The match between glyphs and language
From the linguist’s point of view, a major
evaluation criterion has to be the fit between
epigraphic and linguistic behavior. That is, do
the glyphs involved in a proposed decipherment
behave the way corresponding units in the
languages behave? Is the reading productive in
the sense of resulting in other decipherments
which also agree with known or suspected
linguistic patterns and elements?
The linguistic patterns a reading or other
decipherment should fit are not those of any
modern Mayan language, but are those of the
Mayan languages which were spoken by the
creators of the inscriptions in particular regions
at particular times. Most linguists accept proto-
Ch’olan and proto-Yucatecan as the major
possibilities, depending on place and time. Our
knowledge of these languages is imperfect, and
is limited to hypothetical reconstructions
(Kaufman and Norman 1984; Bricker 1986),
which do not match the written evidence exactly.
In addition to hypothetical reconstructions of
these ancestral languages, we may use other
sources. In descending order of strength as
evidence, these include: data from Colonial
attestation of Ch’olan and Yucatecan languages,
data from modern varieties of Ch’olan and
Yucatecan, patterns of Mayan languages in
general (especially those of the Western branch,
the close relatives of Ch’olan), data from
individual Mayan languages, and finally,
supposed universals and /or general principles.
Some of the features of Mayan languages
which could be expected to show up in Maya
inscriptions, or which have already been
identified, are sketched below.
Phonology
Maya writing involves a combination of
logographic signs, which represent words
without particular reference to their phonology,
and phonetic signs, which represent syllables,
most of the shape CV (consonant-vowel). The
most common syllables may be represented by
several distinct signs (Stuart 1987). Signs have
not been identified for all the possible CV
syllables, and it appears likely that either some
syllables were not written, or that the writing
system had some degree of under-
differentiation, so that two or more similar
syllables could be written with the same sign. On
the other hand, at least some of the syllables for
which signs have yet to be identified are rare in
most Mayan languages (for instance, those with
80
¥ and p’), so it is at least possible that further
research will result in a complete syllabary.
Glottalized versus non-glottalized consonants
All Mayan languages distinguish between
sets of glottalized and non-glottalized (or
“plain”) stops and affricates. Modern Ch’ol, for
instance (Josserand and Hopkins 1988), has a
phoneme inventory which includes p, , 2, ch,
and k (non-glottalized) as well as p’, ”, #2’, ch’
and k’ (glottalized). There is little if any
alternation between glottalized and non-
glottalized counterparts in morphological
processes, and there is no attested pattern of
word-play which associates one set with the
other. That is, we have no reason to believe that
Maya, modern or Classic, could be said to make
a mental association between counterparts from
the different sets (between p and p’, for instance).
There occasionally appears to be evidence in
Colonial or modern data for a lack of distinction
between glottalized and non-glottalized
consonants; for instance, a word may appear to
be alternately written with and without
glottalization on one of its consonants. Such
instances can invariably be accounted for
without positing free substitution of glottalized
and non-glottalized consonants. Some are the
result of an inadequate orthography; variants of
the Spanish-introduced Colonial Yucatec
orthography do not consistently mark any of the
contrasts other than k versus k’, written and
, respectively. Some result from scribal or
typographical errors (leaving off the apostrophe
that marks glottalization). And some are the
result of a mistake in interpretation on the part
of the investigator (it is not really the same word
in both instances). Some confusion may also be
introduced by misunderstood morphological
processes. In Ch’ol, for instance, CVC adjective
roots ending in stops and affricates combine
with a derivational suffix -’an to produce verbs
with CVC’an forms, where the root-final
consonants are glottalized. The adjective root
ch’ok ‘budding, emergent’ yields the
semantically related verb ch’ok'an ‘to be born’.
But this does not imply free alternation of C and
C's it is merely a case of an atypical
morphological device which results in
glottalized and non-glottalized stops in related
words.
Given the high functional load assigned to
the plain/glottalized contrast and the lack of any
attested association between the sets on the part
of Mayan language speakers, we must greet witha strong degree of skepticism any reading which
involves a lack of distinction between glottalized
and non-glottalized consonants, to say nothing
of the insistence that as a matter of principle
there should be no distinction between plain and
glottalized consonants in the Classic written
language. Certainly such a reductionist proposal
is to be proved by evidence rather than simply
asserted.
Syllable structure and synharmony
Mayan languages tend to have syllables of
the shapes CV (consonant-vowel) and CVC
(consonant-vowel-consonant), which result in
some but not many consonant clusters within the
word. Likewise, they are usually found to have
no vowel-initial or vowel-final forms, although
many would argue that this is an artifact of
analysis rather than a hard fact of language
structure.’ Furthermore, in these languages,
most notably those of the Western branch (which
includes Ch’olan), closed final syllables,
especially those ending in stops and affricates,
generally show the aspiration phenomenon
known as echo syllables. That is, final closed
syllable CVC is followed by an automatic, non-
contrastive, voiceless repeat of the last consonant
and vowel. Thus, a word like Ch’ol ja’ ‘water’ is
pronounced ja'(‘a), with the final [a] voiceless or
whispered (Koob 1979).
Knorozov (1952, 1965) was apparently
ignorant of this fact, but proposed on
independent evidence that Mayans wrote final
consonants with a syllable sign containing the
final consonant and a copy of the last vowel.
‘Thus, a word of the shape CVC would be written
CV-CV (Table 1). The final, “synharmonic,”
vowel would be deleted automatically, that is, it
would be silent, not pronounced. This pattern so
closely resembles the phonological pattern
observed in modern Mayan languages that
Knorozov's hypothesis is very attractive; it has
also been extremely productive in terms of
readings.
Table 1
Syllabic spelling of C, V; Cx: GV CV,
Example: k'uk’ ‘quetzal"
Epigraphic notation: tu)
Reading:
Kuk (kU)
Ch’ol:K’uk”‘trogon’
Paradoxically, the synharmonic hypothesis is.
also supported by the hieroglyphic use of signs
representing syllables with non-harmonic high
front vowel i after #, an alveolar consonant: pa-ti
for pat, mu-ti for mut (Table 2). This pattern was
viewed as an exception to the rule of
synharmony until it was recognized that the
Western Mayan languages, including Ch’ol,
have high front vowels as echo vowels after
alveolar and alveo-palatal consonants (t, t’, #2,
t2’, ch, ch’, n, 5, x). This co-variation between
linguistic and epigraphic elements is strong
evidence for the validity of the synharmonic
principle, and is one of the best examples of
match between the behavior of linguistic and
epigraphic elements.
Table 2
Syllabic spelling of CVt:
Example: mut bird, augury*
Epigraphic notation:
Reading:
Ch’ok: mut “bird”
Note that the synharmony principle predicts
that final written vowels which are the same as
the preceding vowel (and the final vowel i after
an alveolar or alveo-palatal consonant) are to be
ignored. On the other hand, there is no
expectation that any other written vowels are to
be ignored.
Readings which violate the principle of
synharmony contradict a well-established
principle, and are to be viewed with suspicion.
We have a basis in Mayan phonologies for
believing that the Maya used syllabic signs with
a principle of synharmony, including the use of
high front vowels after alveolars and alveo-
palatals. We have no basis in Mayan phonology
for expecting that other non-harmonic final
vowels can be ignored. The working hypothesis
has to be that a written non-harmonic final
vowel is meaningful.
This is not to say that the rule of
synharmony is necessarily absolute. There are a
number of readings which imply that there are
other exceptions to the rule of synharmony,
perhaps because the inventory of syllable signs
does not always allow perfect synharmonic
writing. But it is a mistake in research strategy
to assume that synharmony does not exist at all,
or that all differences in written final vowels can
be ignored, as doa number of recent papers. For
instance, Stuart (1987) proposes that the word
tz'ib ‘write’ can be written tz*i-bi ortz’i-ba,
without distinction, as both forms are found in
inscriptions, followed by apparent names in
phrases interpreted as authors’ credits. The form
atz’i-bi is written synharmonically, and could be
expected to be read as t2'ib’ Since the form tzi-
ba is not written synharmonically, another
analysis should be considered.
In modem Chol, the noun ‘writing’ is t2'jb,
but the verb ‘to write’ is tz’ijban (incompletive
aspect stem), with a completive aspect form
t'ijoa the wrote it’. By analogy with these
modern forms, the Classic forms may very well
be the noun form f2’ib ‘writing’, and the verbal
form fz’iba ‘wrote’: u-tz’i-b(i) ‘his writing’ and
w-tz’i-ba ‘he wrote it’ (Table 3). It would be very
difficult to distinguish between these two
meanings in an inscription if the final vowel is
ignored, since both would be followed by a
name phase, representing the possessor of the
noun or agent of the transitive verb.
Table 3
Syllabic spellings: tw’
wicba
Wwi-b(i) = wi-b(a) = t2"ib
Stuart's analysis:
‘write’
Alternative analysis: tz%+b(i) = f2ib ‘writing
wwi-ba = f2'iba ‘wrote’
rij ‘writing’
£2 jba(n) to write’
Chol:
The advantage of this analysis is that it
allows us to preserve the rule of synharmony,
which tells us when the final written vowel is to
be ignored and when it is not. By not assuming
that the final vowel can simply be ignored, we
are forced to account for the non-symharmonic
forms, and therefore learn (or come to suspect)
that both nominal and verbal forms of the
Mayan root were used. If we assume from the
first that it doesn’t make any difference which
vowel is written in the final syllable of a word,
we can never discover the potential difference
between t2*i-b(i) and tz’i-ba
Vowel harmony
Apart from the sub-phonemic vowel
harmony of the echo syllables, Mayan languages
have only limited vowel harmony rules. The
vowels of a very few nominal suffixes are
governed by the vowels of the root or stem to
which they are attached. But Mayan languages
do not have the extensive harmony rules of
languages like Turkish or Hungarian, in which
all the vowels following the root, throughout
several successive suffixes, are determined by
the root vowel. Likewise Mayan languages in
general do not have extensive vowel deletion
from affixes, although some vowel loss through
82
syncopation in sequences of affixes may occur.
Consequently, proposals which involve the
determination of the vowel of one affix by
reference to the vowel of another, or which
presuppose that some written vowels are to be
deleted or ignored in the interpretation of the
inscriptional evidence must be taken with a
grain of salt (see below, in the discussion of hi).
Distributional constraints
Consonants and vowels do not suffer severe
distributional constraints in Mayan languag
That is, there are no sets of consonants which
may occur only initially or only finally, nor are
there vowels which may not occur next to certain
consonants. A given syllabic sign, therefore,
should be expected to occur at the beginning, in
the middle, and at the end of a word, without
restriction. A sign which occurred in only one of
these positions would have a defective
distribution, and could be expected to group
with other signs with which it is in
complementary distribution.
@ @ Ga
Figure 1. The affix T17 (Thompson
1962:40-41, 445). Redrawn from Stuart
(1987.25),
Stuart (1987) advanced the hypothesis that
TI7 (fig. 1) represents the syllable yi, and
supported his hypothesis with readings of yila
‘his seeing’ (I would gloss this ‘he saw [it]"),
yitz'in ‘his younger brother’ and yich’ak ‘his
claw’. The data presented imply that the syllable
only occurs in initial position. On the basis of
the structure of Mayan languages, we should
suspect that yi would also occur outside initial
position, and that either T17 or other signs with
which it is in complementary distribution would
transcribe the syllable yi in final position. T17
itself is common in final position, and there is
every reason to believe that suffixed T17
transcribes -i, the completive aspect marker of
intransitive verbs, as in various period-ending
expressions, the wing-shell ‘death’ expression,
and so on (fig. 2). Note that the corresponding
suffix in modem Ch’ol, usually cited as -i, has
underlying -iy shapes, such as tza chim-iy-O the
died’, and za chitm-iy-on ‘I died’.
Once again, the principle that Maya writingwill reflect the patterns of Mayan languages
provides the motivation for attempting to fill out
an observed pattern, and results in greater
insight. As a matter of principle, the reading of a
syllable like T17, putative yi, should not be
proposed unless an attempt has been made to
account for occurrences of the sign in all
positions. While some limitations in the amount
and nature of the data presented to defend an
argument may be legitimate, it is obvious that no
proposal can be taken seriously if it only
attempts to read a sign in a few of its many
contexts, and ignores without comment
extensive occurrences of the sign in other
contexts.‘ In fact, the basic nature of proof in
decipherment is to show that a reading derived
from one set of contexts makes sense in a series
of other contexts.
Morphology
Mayan roots are basically CVC in shape,
with some CVCVC or other disyllabic forms;
affixes tend to be CV or VC in shape, and are
occasionally CVC. Roots belong to one
(sometimes more) of a number of root classes,
and these determine which derivational and
inflectional affixes will be taken by the root, and
in which sort of constructions the derived stems
will occur. For instance, it appears that the
language of the inscriptions, like Ch’olan and
other Western Mayan languages, has three major
verb root classes: transitive, intransitive, and
positional verb roots. Each root class takes a
distinct set of suffixes to mark completive and
incompletive aspect, as well as calling for
distinct sentence syntax associated with different
sets of arguments.
Figure 2. Examples of T17 as a suffix indicating
completive aspect (past tense) of intransitive verbs
Left, it] ended’. Right, ‘and then, he] died’.
Note that CVC shapes, if transcribed with
phonetic glyphs, would be written disyllabically,
CV-C(V), and CVCVC roots would be
correspondingly longer, CV-CV-C(V), Since
roots characteristically occur with some kind of
affixation, even the simplest word, written
phonetically, would be several signs long. In
fact, it is relatively rare for long strings of
phonetic signs to be combined to write whole
words. Instead, logographic signs are used to
represent even long and complex words with a
single sign, although it is common practice to
write with a combination of logographic and
phonetic signs, that is, logographic main signs
with phonetic affixes. It is fortunate for studies
of the morphology of the Classic Maya written
language that this tendency results in the
spelling out of many common affixes.
The written language of Classic Maya
inscriptions, like modern Mayan languages, has
relatively little nominal morphology, compared
to verbal morphology. Nouns may be possessed;
in the epigraphic record, possession is marked
with the ergative third person singular pronoun
u-. Sometimes nouns are also marked with -VI
suffixes, related to the possessive inflection of
certain nominal categories. Verbal inflection is
much more extensive. An attempt has been
made to collect all or most of the examples of
verbs in Classic inscription (Schele 1982), and
several critical elements of the verbal
morphology have been identified (Bricker 1986,
Macri 1988). On the other hand, there are any
number of known hieroglyphic affixes whose
function is unknown, and much expected Mayan
verbal morphology which is as yet unaccounted
for in the inscriptions. From what is currently
understood, it appears that (contrary to earlier
expectations) the morphology of the Classic
written language does not exactly match that of
proto-Ch’ olan as reconstructed by Kaufman and
Norman (1984)!
One characteristic of affixes in both the
inscriptional evidence and modern Mayan
languages is the occurrence in the inventory of
sets of two or more affixes which have the same
phonological shape, but which may occur in
different positions or with different stem classes
and have entirely different functions or meaning.
Thus, the epigraphic nominal prefix ah- marks
agentive nouns, while as a verbal suffix, -ah
marks the completive aspect of verbs. On the
other hand, as pointed out by Stuart (1987), the
same grammatical affix may be spelled out in a
variety of ways without changing its function or
meaning. It is clear that as far as affixes are
concemed, we are dealing not with
morphological signs, but with syllabic spellings
of morphemes.
Syntax and discourse structure
The discovery that Mayan inscriptions
8displayed the major patterns of sentence syntax
known from Mayan languages has been one of
the greatest breakthroughs in the history of
Maya epigraphy (Schele 1982:20, and Schele’s
Notebooks, 1978-1990). The simple realization
that what follows the initial temporal statement
(for examples, a Calendar Round) is a verb, and
that following the verb comes a noun phrase that
functions as the subject or object of that verb, has
been the key to deciphering the content of the
inscriptions, and has led to the possibility of
identifying the grammatical affixes which are
attached to the major sentence elements.
In fact, the epigraphic Mayan VOS (Verb-
Object-Subject) basic word order, like similar
elements of modern spoken languages, rarely
occurs by itself. Not only is it usually related to
a preceding temporal statement, but the sentence
often includes a complement or other oblique
phrase after the verb (such as a locative phrase),
and it is rare to find both subject and object
expressed in the same sentence (Josserand 1987).
Some more complex syntax has been noted,
particularly the conjoining of sentences in related
pairs. The development of knowledge about one
particular set of sentence conjunctions illustrates
the stages of discovery at this level of language
structure. The Anterior and Posterior Event (or
Date) Indicators were deciphered at the level of
interpretation—as relating one statement to
another—by an earlier generation of epigraphers
(cf. Thompson 1950). Capturing the sense of
‘Thompson's treatment, they were paraphrased
as sentence conjunctions by Schele (1978 and
later Notebooks). They have recently been
identified as markers of background and
foreground information in narrative text
structures, related to similar elements in modern
Ch’ol, and assigned phonological readings
(osserand 1987)
It is work on the interface of syntax and
discourse structure that will result in similar
advances in our understanding of other devices
of complex syntax. Questions of overall text
structure, text segmentation, participant
tracking, and other aspects of discourse structure
are beginning to yield to analysis (see Bricker,
Josserand, and Maxwell, this volume). ‘The basic
patterns, at least as far as the monumental
corpus is concemed, seem to be similar to those
of modem Mayan historical narrative (Hopkins
and Josserand 1990). The most fruitful aspect of
this kind of analysis is that we are beginning to
see the inscriptional texts as literature, not just
compilations of historical data. With these
84
remarks in mind, I return to the question of
criteria of evaluation by looking in detail at one
recently proposed reading.
Stuart’s syllable sign hi
Stuart (1987) has recently proposed ten new
syllabic readings with varying degrees of
support. He ranked the ten proposals in order of
their strength, and the last syllable sign, the least
well supported, is proposed as hi. I agree with
Stuart that this is the weakest of his proposals,
and as such it provides a good example of the
fit—or lack of fit—between epigraphic and
linguistic material.
Stuart’s proposal groups the variants of
three graphic units, T88, T136 and T758b. These
are graphically quite distinct (fig. 3, top): a paw-
like form (T88), an animal head (1758b), and a
line of small loops (T136). The basis for
grouping these three sets is the observation that
they replace one another as suffixes in a series of
identical contexts which, however, Stuart does
not decipher (fig. 3, bottom).
The phonetic value hi is derived from a
pattern of substitution which involves the
‘seating’ glyph (chum) and the suffix T126,
which Stuart reads ya (fig. 4). Stuart first
observes that the ‘seating’ glyph chum is usually
suffixed with -wa-n(i) or -la-h(a), forming the
words chum-wan and chum-lah, respectively, two
past-tense forms of the verb ‘to be seated’ (fig.
4a.c). He then notes that in clauses linked by
Distance Numbers with the ‘seating’ event as the
earlier of two events, the ‘seating’ verb is further
suffixed with T126, Thompson's Anterior Event
Indicator or (AEN), putative ya (fig. 4b,d).°
Stuart analyzes T126 in these constructions
as a phonetic complement rather than as a
syllable sign or morpheme; in his view, T126
occurs only after syllables which have as their
vowel, and serves only to “indicate that the i
vowel of the preceding syllable is to be
pronounced.” Thus, adding T126 to a
construction indicates that the preceding vowel
is i, and is to be pronounced. But Stuart
maintains that the T126 ya itself is not
pronounced, or rather it serves as the echo
syllable, dropping its vowel @ and merging the
remaining y with the i of the preceding syllable.
Thus, what is written chum-wa-ni-ya (fig. 4) is
interpreted as chum-wa-ni-y(a) and gives
chumwani whereas what is written chum-wa-ni
is interpreted as chum-wa-n(i) and gives
churewan.Figure 3. Stuart's syllable sign hi. Top, the affixes T88 (Thompson 1962:446);
1758 (Thompson 1962:335-359, 455), and T136 (Thompson 1962:53 447).
Redrawn from Stuart 1987:42. Bottom, substitution sets presented as
evidence by Stuart; ver
I columns show occurrence of a particular variant
in three different contexts; horizontal rows show substitution of the three
variants in a given context. Redrawn from Stuart (1987:43, Fig. 54)
Stuart believes that the final vowel i is the
aspect marker for ‘perfective’, and is “related to
the common perfective suffix -/ on intransitive
verbs in both Ch’olan and Yucatecan.”” Since
T126 can only follow i, it cannot be suffixed to
the past tense form chum-la-h(a), which has no
final vowel i (fig. 4c). Therefore, in order to
attach T126 ‘perfective’ to chum-la-h(a), the last
syllable must be changed to a sign with the
vowel i—and with the consonant k of the stem,
ergo hi. Instead of ‘chum-la-ha-ya, then, we get
chum-la-hi-ya, interpreted as chum-le-hi-yia),
and read chumlahi ‘he had sat’ (fig. 4d)
The same pattern is reported to occur on the
verb ts'ap ‘to plant a post (or stela)’, where the
simple past is written ts’a-pa-h(a) (read fs‘apah)
and the perfective is ts'apachi-y(a), read
ts'apacki. Likewise, in k’anasi, the Ch’olan name
for a month usually cited in its Colonial
Yucatecan form, Kayab (K’ayab’), the suffix T126
indicates the pronunciation of the final vowel;
the construction written k’an-a-si-y(a). Stuart
finds additional support for the reading hi in the
occasional appearance of putative hi at the end
of ch’ah ‘drop of liquid’, wah ‘food’, and Kuxah
‘bitten/hurt’: the written forms ch’a-hi giving
ch’ah, wa-hi giving wah, and k*u-xa-hi giving
Ruxah,
Problems with the hi reading
The paragraphs above summarize Stuart's
presentation of evidence for the reading hi. In
my opinion there are a number of
methodological problems raised by this reading
85CHUM-wa-n(i)
chumwan
he sot
CHUM-wa-ni-y(a)
chumwani
he hod sat
CHUM - la -hla)
chumlah
he sat
CHUM-la-hi -yl0)
chumlohi
he hod sot
Figure 4. Inflection of the positional verb chum “to be seated” according to Stuart
(1987:42-43). Redrawn from Stuart (1987.43, Fig. 5).
which argue for its rejection.
Tt is questionable whether the distinct sign
groups T88, T136 and T758b are phonetic
equivalents. It is clear that they substitute for
one another in at least some contexts. But
substitution does not imply identity, especially
in verbal affixation. One critical factor is
whether the substitution changes the meaning of
the construction. If substitution does not change
the meaning of the construction, then the glyphs
might be equivalent. But if substitution does
change the meanings, then the glyphs cannot be
equivalent. To take a simple example, the
various numbers substitute for one another
before the day names (for example, we can
observe the combinations 1 Ahau, 2 Ahau, 3
Ahau, etc). But the number glyphs are by no
means equivalent to one another, since the
meaning of the construction changes with each
substitution. Patterns of substitution may
establish paradigmatic substitution without
implying semantic or phonetic equivalence.
86
In the examples presented by Stuart, there is
no semantic control; it is unknown whether or
not the substitutions correlate with changes in
meaning. ‘The evidence is therefore inconclusive.
788, T136, and T758b may be members of the
same paradigmatic or syntagmatic class, but they
are not necessarily semantically or phonetically
identical.
‘A more serious problem with Stuart's
reading is that the grammatical constructions
said to be written by the glyph combinations are
not forms we would expect to occur in proto-
Ch’olan, presumably the language represented
(since the positional suffix -wan is Ch’olan and is
not found in Yucatecan). Stuart takes the suffix i
to be ‘perfective’, but this is problematic. First,
chum is a positional verb, and in proto-Ch’olan
this class of verb did not take the ‘perfective’ (or
‘completive aspect’) suffix -i (Kaufman and
Norman 1984).
Itis true that by Colonial times, the Chontal
language had generalized the use of this suffix toall verb classes, and it does appear in words like
chumwani and chumlahi in the Paxbolon papers
(Smailus 1975). But this is a Post-Classic Chontal
innovation, not a Classic pattern. In Classic
Ch’olan the suffixes -toan and -Iah themselves
had the function of marking the ‘perfective’ (or
‘completive’) of the positional verbs. The suffixi
marked the same aspect, but for intransitive
verbs only (Kaufman and Norman 1984:103).
The forms chumwan and chumlah are already
marked for this aspect (with -wan and -lah), as
are ts'apah and K'uxah, transitive verbs which
take the aspect marker -ah. In Chontal and
Chontal alone, a later innovation generalized the
aspect suffix -i to the positional and transitive
verbs as well as the intransitive ones, creating
the forms like chumwani attested in the Colonial
Chontal source. Since the forms chumzani and
chumlahi did not occur in the Classic language,
they should not have occurred in Classic
inscriptions, and any reading based on these
forms is suspect.
The occasional addition of putative hi to the
nouns ch’ah and wah, to write the final
consonant, violates the rule of synharmony. This
isno problem for Stuart, who has abandoned the
tule of synharmony as a guiding principle, but it
continues to worry me.
Probably most troublesome, Stuart's
interpretation of T126 introduces a new and
entirely unproductive principle to Maya writing.
We are asked to believe that T126, putative ya, is
added only to syllables having the vowel i, and
that the sole function of T126 is to assure the
Pronunciation of this vowel. This is
unprecedented in Maya epigraphy. More
significantly, it is unproductive. T126 is an
extremely common suffix (and frequently occurs
as a prefix as well). Are alll its occurrences
simply to reinforce a preceding vowel i? How
are other final vowels reinforced? Stuart's
analysis raises more questions than it answers,
and has implications that are simply
unacceptable.
Finally, the syllable sign hi occurs is a very
limited distribution. It is proposed in no
contexts other than final or suffix position. This
suggests that it is only part of a set of related
glyphs as yet unidentified.
Conclusion
The hi reading proposed by Stuart is
contradicted by the evidence he presents in its
support. It produces ungrammatical forms. The
arguments presented in its favor violate
established principles. The reading introduces a
new and unproductive principle which is easily
refuted.
To illustrate the weakness of the proposal,
let us suppose that the sign set (188, T136 and
1758b) does in fact form a legitimate group
(although that has not been demonstrated), and
that it reads not hi but ha. This reading would
fit all the examples presented by Stuart, would
produce grammatical forms, would violate no
established principles, and would introduce no
untenable new ones (Table 4; T126 is not
interpreted in these examples so as not to
introduce extraneous factors):
Table 4
]EPIGRAPHICFORM __LINGUISTICFORM
Jchum-la-h(a) -T126chum-lah (1126) ‘he sat, he was
seated’
tsta-pa-h(a)-T126 _ts'ap-ah (1126) ‘it was erected"
[ch’a-h(a)
wa-hia)
eu-xa-h(a)
ch’ak ‘drop(s)"
wah ‘food!
Keux-ah ‘it was bitten’
To summarize, the proposed reading hi
produces unacceptable results and there is at
least one better reading to account for the
evidence presented (ha). The proposal that T88,
T136 and T758v are hi is unsupported by the
evidence, and should be rejected. On the other
hand, there appears to be considerable evidence
that the proper reading for this substitution set is
ha.
NOTES
1 The material presented in this paper is based in
part on research supported by the National
Endowment for the Humanities (RT-20643-86), and
bythe National Science Foundation (BNS-8520749,
administered by the Institute for Cultural Ecology
Of the Tropics).
2 Roots in Mayan languages can generally be
divided into those that begin with a consonant
other than a glottal stop, and those that do not.
‘The two sets take different inflectional prefixes, for
instance Ch’ol tat ‘father’, a-tat ‘your-father’
versus ‘ab ‘hammock’, aw-2b * your-hammock’.
‘The consistent CVC shape of Mayan roots argues
for a structural analysis of words like ‘ab
‘hammock’ as roots which begin with a glottal
stop, but which lose the glottal stop when prefixed.
Alternatively, these roots can be treated as VC
vowel-initial Toots which acquire a glottal stop
when not prefixed. The choice between these
alternatives is determined by the analyst's
theoretical and methodological preferences, not by
the inherent structure of the languages. A’ similar
situation often pertains to initial k and more
frequently, to final h
3. In epigraphic notation, such a form is usually
87represented as tz'i-b(i), to indicate that the final
vowel is to be ignored.
At the very least, the existence of the sign's
‘occurrence in other contexts should be mentioned,
if only because few scholars have access to, much
less total recall of, the entire corpus.
The situation is not unprecedented, since
reconstructions of proto-Romance from modern
Janguages do not exactly match the forms attested
in Classical Latin. In both cases, the modern
languages are derived from vernacular dialects,
while the Classic written language records the
literary forms of the elite.
Note ‘that this analysis is distinct from that
sed by Josserand (1987), who treats T126 as
Broperecnve nope cutc i
‘There is a confusion in terminology resulting from
a Jack of standardization in linguistic, and
graphic studies. What is commonly cal
{hae by epigraphers is usually referred to as
completive aspect by linguists, while the
igraphers’ pluperfect is the linguists’ perfective
TERE Toleeand and Hopkin 1988. Sart
uses perfective in the sense of perfective aspect, but
the suffix -i he refers to is not perfective, but
completive aspect, that is simple past tense
(Kaufman and Norman 1984)—although Stuart
takes it to mean pluperfect. The suffix in Ch’olan
languages which corresponds in meaning to
Stuart's perfective is not the completive aspect
suffix -i but the perfective aspect suffix -ix (as,
Josserandl (1987) reads T126.