Alphabetic Vs Non-Alphabetic Writing - Baroni

Alphabetic vs.
non-alphabetic writing:
Linguistic fit and natural tendencies
Antonio Baroni
This article has two main purposes. The first one is to prove that the
alleged superiority of the alphabet to other writing systems (syllabic and
logosyllabic ones) is an ethnocentric prejudice and that the optimality of
a writing system has to be measured following a series of criteria which
cannot be reduced to the faithful mapping of sounds. The second one is to
incorporate into the graphemic theory external data and new approaches
to develop new methods of investigation and to emancipate graphemics
from phonology. The structure of the article is composed of seven parts.
First of all, we discuss some definition problems; then, in the introduction,
the main points of view about the alphabetic principle are exposed and in
chapter 2 the relationships between writing systems and language perception are investigated. In chapter 3 we attempt to define some criteria to
judge the degree of optimality of the different writing systems. In chapter
4 we try to find some patterns of predictability of the degree of opacity and
transparency of some of the main European writing systems (the opaque
English, French and Danish orthographies and the shallow Finnish and
Italian orthographies). In chapter 5 we shortly examine the natural evolution of writing in recent times: Internet, SMS and new writing systems.
Finally, in chapter 6 we try to draw some temporary conclusions.*
Definitions
Before starting our investigation about the degrees of optimality
of the different writing systems, it would be better to deal with definition problems.
*
Symbols and abbreviations
[a]
/a/
<a>
{a}
|a|
sing.
pl.
m.
f.
phone
phoneme
grapheme, graphoneme or allograph
morpheme
morphoneme
singular
plural
masculine
feminine
Rivista di Linguistica 23.2 (2011), pp. 127-159
(received September 2011)
Antonio Baroni
By writing, we mean a series of graphic symbols arranged on a

surface (be it physical, like a sheet of paper, or virtual, like a screen),
in a certain order and in a certain sequence, so that this series is likely to be interpreted (read) by an interpreter (reader) who knows how
to decipher the meaning of these signs.
A writing system can then be cenemic, if its elements stand for
units of phonic expression, or pleremic, if its elements stand for units
of content (Coulmas 1989: 38-39). Nevertheless, one must not forget
that there is no pure writing system: in every tradition cenemic and
pleremic components coexist, to different extents.
Among pleremic writing systems, some display pictograms, whose
shape recalls iconically the object represented, e.g. <w> Egyptian
pictogram meaning sun, whereas ideograms, also iconic, represent
abstract concepts, e.g. Chinese <> sn means three and is composed of three strokes. In logographic systems, the graphic elements
stand for a word or for a morpheme.1
A cenemic writing system can be a:
syllabary, where every graphic unit stands for one syllable,
normally a CV-type of syllable; Japanese kana, Linear B and
Cherokee are syllabaries;2
abjad (Daniels & Bright 1996), where consonants are represented but vowels are not, even if there is the possibility to add
diacritic vocalic signs to disambiguate. This system is typically
adopted by Semitic languages, which share the characteristic to
b
possess triconsonantic lexical roots, e.g. from Arabic
abaah he sacrified,
abata you-m. sacrified,
abbaa he slaughtered, etc.
abugida. It is a syllabary where the graphic elements standing
for the consonants and the ones standing for the vowels are
recognisable, e.g. Indian devangar, Ethiopian Geez script;
alphabet, where ideally all the phonemes of a language are
noted by separate elements;
featural writing (e.g. Korean hangl), an alphabet where the
shapes of the graphic signs correlate with phonemic distinctive
features.
A script selects the modality of graphic expression making it fit
to the structure of the language it represents, e.g. Latin and Greek
had the same writing system (an alphabet) but employed different
scripts.
An orthography selects the possibility of a script through uniform
and standardised procedures of correspondence between graphic elements and linguistic units. For example, Italian and French share the
128
Alphabetic vs. non-alphabetic writing
same writing system (alphabet) and the same script (Latin) but follow
different orthographic rules. Interpunction, spacing and capitalisation
are part of the orthography.
An orthography can be transparent or shallow if, given a set of
basic rules, it is always possible to read and write a word, even an
invented one (e.g. Finnish, Italian, Spanish, Turkish, Georgian, etc.).
In opaque orthographies, on the contrary, the correspondence between
spoken words and written words cannot be reduced to a set of rules
and, depending on the degree of opacity, one must learn by heart
a certain number of graphic words (e.g. French, English, Danish,
Khmer, Chinese, etc.).
We define grapheme the minimal meaningful graphic unit, in every tradition; its meaning can be cenemic (e.g. <a> /a/), both cenemic
and pleremic (e.g. Chinese <> bn, down, under) or only pleremic
(e.g. <h> in Italian ha has whose function is just to indicate that this
word belongs the paradigm of the verb avere, to have, although it
does not correspond to any sound). If we consider merely the soundletter correspondence, then we are not talking about graphemes
but graphonemes (Hoej 1971: 186), e.g. in French <c> and <h> are
graphemes, but <ch> is a graphoneme,3 inasmuch as it stands for one
phoneme (mostly // but sometimes /k/).
1. Introduction
Currently, the most widespread point of view in Western linguistics about graphemics is still biased by several prejudices, all of
which can be reduced to the teleological position formally expressed
by Ignace Gelb that there would have been a constant improvement
in the historical evolution of writing, in which the alphabet would be
the pinnacle of perfection, regarded both as the cause and the effect
of a high degree of civilisation. It goes without saying that any deviation from this principle is considered as an aberration or an imperfection; non-alphabetic systems are therefore deemed to be inferior to
alphabetic ones and opaque orthographies inferior to shallow ones.
According to Gelb, writing followed a linear evolution, passing from
an early stage of logography and subsequently switching to the syllabic principle and culminating with the alphabet. This path is seen
as unavoidable (Gelb 1963: 240).
Other scholars agreed with this view: Diringer (1948) calls the
alphabet a key to the history of mankind, while Ong (1986) goes so
far as to claim that Latin alphabet will replace Chinese characters
129
Antonio Baroni
as soon as everyone in the Peoples Republic of China share the same

language. The idea of the superiority of the alphabet (and of Western
civilisation), as opposed to other writing systems (and cultures), is
promoted also by McLuhan (1964), McLuhan & Logan (1977), Innis
(1991) and Logan (2004).
Among the opponents to this position, Venezky (1970: 120)
claims that [the fact] that homo sapiens is somehow more at ease
with a one-letter one-sound system has often been assumed, but no
evidence has ever been produced to substantiate this limitation on
mans mental capacities. Also Gleason (1961: 419) disagrees with
Gelbs theory, since he doubts that an alphabet which did accurately
record speech would be practical. Coulmas (2009a: 105) points out
that:
[f]rom a Near- and Far-Eastern point of view [the] validity [of
alphabetocentrism] is not so evident. Japanese kana, for example,
is much simpler and more elegant than almost all scripts using an
alphabetic notation. The system is so simple that children can be
expected to have mastered it before they enter elementary school.
There is no need to teach it there.
Daniels (1992: 83) brings also some scientific data, arguing that
the phoneme, the unit on which alphabetic writing is based, is not a
natural unit,4 given that
[i]nvestigations of language use suggest that many speakers do not
divide words into phonological segments unless they have received
explicit instructions in such segmentation comparable to that
involved in teaching an alphabetic writing system.
2. Writing and language perception

Not surprisingly, perfectly phonemic orthographies do not exist.
Moreover, mixed systems such as English and French orthographies,
defective writing systems such as the Arabic abjad and logographies
such as the Chinese one are the most widespread writing systems
in the world. Some scholars argue that it was the invention of the
alphabet to affect language perception; modern linguistics could not
have been theorised without the constant influence of the alphabet.
According to Faber (1992: 127),
130
segmentation ability as a human skill may have been a direct result

(rather than an impetus to) the Greek development of alphabetic
writing. Thus, the existence of alphabetic writing can not be taken
eo ipso as an evidence for the cognitive naturalness of the segmentation that it reflects. () [W]e as linguists feel that, because we can
describe linguistic system in terms of phonetic segments, we must
do so [but] () every technical linguistic tradition that refers to segments arose in an alphabetic milieu (). In contrast, the indigenous
Chinese linguistic tradition () has as phonological primitives syllables initials and finals,5 that is, onsets and rhymes. This analytical
division is not supported by the logographic Chinese orthography, a
lack which strengthens the force of the analysis.
Several studies (Foss & Swinney 1973, Liberman et al. 1974,

Morais et al. 1979, Cossu et al. 1988, Daniels 1992, Goswami 2005)
have shown that the perception of the phoneme, though having a psychological reality, always follows syllable-awareness. The phoneme cannot be separated from the syllable, anyone can verify this by recording
his/her own voice in a computer and then processing and analysing it
using a speech spectrum software (e.g. Praat); one will find out that it
is impossible to listen separately to the consonantic and to the vocalic
part of the syllable in isolation, the result being something similar to
an electronic reverberation. These data would explain the ubiquity, in
different cultures, of syllabaries and the rarity of alphabets.
Another experiment (Cho & McBride-Chang 2005) has shown
that Korean children apply their syllable-awareness to learn hangl,
the native Korean script, but when learning English orthography,
they use their phoneme-awareness; depending on the script to learn,
they use different skills.
However, one must not forget that, even if writing affects the
metalinguistic view of language users, any writing system invented
through history could not have seen the light without a prior linguistic consideration. After all, if Greeks had not thought about a concept
similar to that of phoneme, they could not have invented a segmental
writing system.
3. The optimality of writing systems: some possible criteria
Shall we therefore conclude that the prestige acquired by the alphabet since its first appearance is solely due to ethnocentric prejudices?
Well, we think that the alphabet has its own merit beyond its prestige.
131
Antonio Baroni
The syllable is a more salient perceptive unit so an ideal script,

if it exists, should be based on it. But syllabaries have always caused
several problems, since they should represent all the syllables of a
language. When a language has a syllabic structure that is mainly
CV, the syllabogram inventory is still manageable, but when it comes
to Indo-European languages like German or Russian, the list of syllabograms would be endless. The devangar script uses graphemes
for single vowels and graphemes for consonants with an inherent
vowel // (transliterated <a>). If the syllable to represent graphically comprehends another vowel, diacritics are used. Conjuncts are
employed for complex syllables like CCV or CCCV but it is often hard
to reconstruct the way in which they have been composed. Moreover,
the number of symbols to memorise, if one considers the conjuncts as
well, rises considerably (Masica 1993: 162). The devangar script is
an abugida which, albeit being a sophisticated system, is less flexible
than the alphabet in transcribing foreign words and consonant clusters.
Logographic scripts are very cumbersome systems to learn; it
takes several years for Chinese children to master the basic graphic
system and the knowledge of a larger number of characters is a lifelong learning process. Japanese children need considerably less time
even if they have to learn not only one, but three systems: the kanji
series (Chinese characters with an ideo/logographic function), hiragana (syllabograms) and katakana (syllabograms used mainly for foreign words), not to mention the Latin alphabet. Nevertheless, Chinese
logographic system has one great advantage: it is interdialectal; the
Peoples Republic of China is linguistically fragmented and regional
languages and dialects are not always mutually intelligible, but everybody manages to communicate thanks to writing and that is one of
the reasons why Chinese script is still in use today (Coulmas 1983:
246).
The script that seems closest to the ideal is the Korean hangl,
inasmuch as it is iconic in relation to both speech articulation and
syllable organisation and is able to graphically render very complex
syllables such as sang < >and balp < >, not being excessively tied
to its own linguistic fit. Besides these particularities, hangl is an
alphabet since every phoneme is represented; so does that mean that
the alphabet is indeed the best writing system? It is difficult to judge
the optimality of a given writing system because different criteria
have to be taken into account and they are often in opposition with
each other. We will consider five main criteria:
132
3.1. Maximum distinctiveness

According to the first criterion, the Latin alphabet would be
one of the least functional, since the differences between the various
graphemes are not really striking and it is easy to confuse them with
each other. On the contrary, Chinese characters, Egyptian hieroglyphs
or Maya glyphs differ greatly one from the other. Compare sequences
as <aeiou> vs. <>, <dhoplm> and <
>.
Sampson reports an anecdote (Chiang 1973: 3-4), according to
which literate Chinese have claimed that European script gives an
impression of monotony and lack of distinctiveness, something similar to what we might experience when encountering a page printed
in Morse code (Sampson 1985:164). Apparently, Hebrew abjad is even
less visually distinct than Latin alphabet; the latter (in its lower case
version) presents at least ascending and descending lines such as in
<b, f, k, t> and <g, j, p, y> which make for a greater recognisability
(Sampson 1985: 94). Thanks to these lines, pre-literate children would
already be able to distinguish some written words basing their decision on the bouma (from the name of Dutch psychologist Herman
Bouma, meaning the contour of written words), normally when they
are around 4 years old (Pontecorvo 1994: 278). On the contrary, most
Hebrew graphemes consist of a horizontal line on the top and a vertical line on the right: <
> and some experiments (Gray 1956:59)
suggest that readers of Hebrew and Arabic make longer eye-fixations
than European readers.
3.2. Size of the graphemic inventory
According to the second criterion, the Latin alphabet, with a
number of graphemes comprised between 20 and 40, depending on the
orthography considered, would be relatively fast to memorise, whereas it takes several years to learn the thousands of Chinese characters.
The different mnemonic weight has an evident neurological counterpart: it appears that Chinese children who suffer from dyslexia,
characteristically present an under-activation of the Exner area (in
the left medial frontal region of the brain), whereas the problem for
European dyslexic children seems to be caused by an anomaly in the
left temporal lobe; this is because the main difficulty of learning written Chinese is not at the phonological level but at the mnemonic level,
since one needs to remember at least 3000 characters. To cope with
that, Chinese kids use the visuo-motory memory necessary to draw
the characters (Dehaene 2009: 283-285).
133
Antonio Baroni
3.3. Cognitive salience

By the criterion of cognitive salience we mean the cognitive effort
required to transfer words from speech to writing and vice versa. The
higher the level of abstraction, the greater the required cognitive
effort. Gleitman & Rotzin (1977 cited in Martlew 1983: 261), two psycholinguists, affirm that [y]oung children () are aware of language
as meaning units, only later aware of the phonological and syntactic
substrata of language. From this point of view, the alphabet would be
the most complicated system and pictography would be the simplest.
3.4. Maximum naturalness
According to the criterion of maximum naturalness, the simplicity of the graphemes is regarded as one of the most important factors.
Also in this field, alphabets are among the easiest to learn, but so are
Japanese kana and Korean hangl.
Perhaps it is not obvious what we mean by simplicity: our intuition consists in the fact that recurring features which are found in
most existing writing systems are somewhat natural for human
beings. An interesting study (Changizi & Shimojo 2005: 267) compared more than 100 writing systems, attempting to find similarities
between them and it turned out that
[b]ecause writing systems are under selective pressure to have
characters that are easy for the visual system to recognise and for
the motor system to write, these fundamental commonalities may be
a fingerprint of mechanisms underlying the visuo-motor system.
The results indicate that a character is preferably composed of

three strokes; this is because it is still possible, for the brain, to process, in a parallel way, three elements but beyond this number, more
time is required.6 It goes without saying that writing systems containing several characters composed of more than 3 strokes are considered harder to reproduce manually.
Redundancy seems to be another important factor, since most
graphemes are redundant, namely, not all the strokes composing
them are necessary for their recognisability (Changizi & Shimojo
2005: 273).
According to the neurophysiologist Stanislas Dehaene, all writing systems share traces of an underlying structure. Although different writing systems were historically invented independently one
from the other, his Neuronal Recycling Theory (NRT) predicts that
134
human inventiveness must be limited by the organisation of the circuits of our brain (Dehaene 2009: 201-204). The NRT tries to explain
how human beings are so at ease with reading even if writing was
invented only 5400 years ago and the alphabet is just 3800 years
old, since our genome would not have had the time to develop brain
circuits specific for reading. The brain would not have developed new
circuits, but neurons which once were employed for one task, switched
their function and they specialised in discriminating graphemes from
other visual stimuli, such as faces, objects, numbers, etc.
The Japanese neurophysiologist Keiji Tanaka has discovered
that chimps possess neural sensors which react to elementary shapes
and their function is to recognise objects (Tanaka 2003: 90-99). These
simple shapes are a sort of alphabet because, combining them, every
object can be described and, moreover, they look surprisingly like
some elements of our writing systems (Dehaene 2009: 153). That is
why Dehaene calls them protoletters, among which the most widespread in all writing systems are similar to <T, F, Y, L, 8>. What do
these protoletters have in common? When they impress our retina,
they appear as structured objects, and the cerebral cortex judges them
as non-accidental (Dehaene 2009: 169).
All writing systems, be they alphabetic, syllabic or logographic,
draw on a small set of stroke configurations whose spread follows a
universal tendency; the most frequent configurations are more likely
to be found also in nature and are therefore encoded by neurons in the
inferior temporal cortex even before learning how to read (Changizi et
al. 2006: 117-139).
Since neurological studies (Dehaene 2009: 66) also showed that,
regardless of the writing system considered, human beings use the
same part of the brain to read (the left occipitotemporal region, with
minimal differences), some configurations underlying our graphemes
must be universally easier (more natural) than others.
3.5. Inner consistency
When we talk about the criterion of inner consistency, we refer
to the degree of iconicity7 in relation to language mapping. To put it
down more clearly, we will make some examples. Among cenemic writing systems, hangl displays a high degree of consistency, since similar characters stand for similar sounds, whereas the Latin alphabet
is only partially consistent: some graphemes which stand for similar
sounds look similar too, e.g. <m> and <n> (both nasal), <s> and <z>
(both sibilant, dental or alveolar, depending on the language), and
135
Antonio Baroni
 (both labial), and <v> (both labial), but they are the minority of the cases, since the similarities in shape of other characters
are completely unrelated to the sounds they stand for: <N> and <Z>,
 and <d>, and <q>, etc. The hek employed in some Slavic
languages and the German umlaut are good examples of consistency
of a graphic system, since: /s/ : // = <s> : <>, for example, in Czech,
so that the hek stands for [+palatalised] and /u/ : /y/ = : <> in
German, so that the umlaut stands for [+front].
Pictographies should be, ideally, highly consistent, since they
should represent in a univocal way concepts but here one needs a
common background or the risk is to incur serious misunderstandings. Ideographies and logographies are not consistent at all in relation with phonology and may display different degrees of iconicity
with the meanings they convey, depending on the given system.
Of these five criteria that we tentatively outlined, one can not
choose the most important one: the degree of greater or lesser simplicity will also depend on the cognitive strategies of individuals.
3.6. Other criteria
One should also consider:
the point of view of the reader vs. the point of view of the writer;
the point of view of the native speaker vs. the point of view of
the non-native speaker.
In the act of reading, words function as units of meaning. The
design of the word may be composed of strokes (as in ideograms) or of
letters of the alphabet (also composed of strokes), it does not matter to
the reader as long as his/her mental orthographic lexicon is activated
by the word recognition process, in which written words are perceived
as visual Gestalts by expert readers. The same does not apply to the
writing process, in which the production of the graphic sequence is
analytic rather than synthetic.
For the reader, the criteria of maximum distinctiveness and
maximum naturalness are extremely useful, whereas the writer is
probably more comfortable with a reduced number of symbols easy
to reproduce. Similarly, for native speakers a phonologically detailed
information is not necessary, they just need a phonological cue and
are then able to identify the word thanks to the context. This is not
true for non-native speakers. So, to whom should we pay more attention? Readers or writers? Native or non-native speakers? According to
Sampson (1985: 212),
136
[a]ny literate adult, even a professional author, reads far more than
he writes; so if () the ideal script for a reader is a somewhat unphonemic script, () the balance of advantage has been tending to move
towards the reader and away from the writer: extra trouble in writing a single text can now be massively repaid by increased efficiency
of very many acts of reading that text. () [I]t is worth spending
more time nowadays to learn an orthography, if the extra time is the
cost of acquiring a system that is relatively efficient once mastered,
because the period during which the average individual will enjoy
mastery of an orthography is now longer than it used to be.
So far, it seems unavoidable to consider other elements to define

the optimality of a writing system (cf. Coulmas 2009b). The most likely conclusion we can reach is that there is not such thing as a system
that would be optimal for every language in every society, since every
language has got its own linguistic fit.
4. The linguistic fit of the Latin alphabet: Opaque and shallow orthographies
4.1. English
English orthography is undoubtedly one of the most debated topics in graphemic studies and this is due to the prestige of English as
the global lingua franca. One striking aspect of English orthography
is its sobriety, since it does not display any diacritics or special letters,
it is therefore optimal according to the criterion of the ease of reproduction because one never needs to use special keyboards or to worry
about accents, diaeresis, and so on.
Nevertheless, English orthography is considered to be very hard
to master, both for native and non-native speakers. It used to be
closer to spoken English in the past, but around 1400, when a common orthographic standard was established, a peculiar linguistic phenomenon, the Great Vowel Shift, started to change the pronunciation
of almost all vowels and it came to an end only around 1600. That is
why English vocalic graphemes are pronounced so differently from
all the other European languages; they normally have two possible
phonic meanings, corresponding to the free pronunciation or to the
checked pronunciation. The former is usually indicated by the diacritic grapheme <e> on the following graphic syllable, compare mat /
mt/ vs. mate /met/, bit /bt/ vs. bite /bat/, con /kn/ vs. tone /tn/,
etc. The diacritic <e> has provoked a lack of isomorphism between the
137
Antonio Baroni
graphic syllable and the phonic syllable, inasmuch as words composed

of two graphic syllables often correspond to one phonic syllable.
Moreover, during the Renaissance until the 18th century, many
etymological letters were introduced, so that cors, langage, doute,
samon became corpse, language, doubt, salmon. Aesthetic and typographic questions modified the length of some words: <e> was added
or removed to adjust the margins of the page: doe, goe, heere vs. do, go,
here (Mitton 1996: 19).
Throughout the last century, many have tried to understand if
English orthography possessed a sort of underlying principle (see
Venezky 1970, Albrow 1972) and some have pointed out that the socalled chaos of English orthography is, unexpectedly, often useful. The
most widespread example is the distinction of homophones, e.g. rite vs.
write vs. right vs. Wright; in this way, like visual morphemes (Bolinger
1946), these words speak directly to the eyes avoiding the possibility
of confusion in the process of understanding the text.
According to Chomsky and Halle, English orthography is a
good representation of the underlying form of lexemes (Chomsky
& Halle 1968: 47-49). An optimal orthography, they argue, should
be a deep one, having a single representation for each lexical entry.
Spanish spelling would thus be somewhat inferior to that of English:
the conjugation of the verb pedir to ask undergoes a morphonemic
alternation: the sound /e/ of the root {ped-} becomes /i/ in a stressed
position, so that the paradigm is pido, pides, pide, pedimos, peds,
piden. Since this transformation is quite regular and predictable
for a native speaker, a deep orthography would always write <ped->
(Sampson 1985: 200). One might agree with Chomsky and Halle, if
English orthography were truly a deep one, but this assertion can not
be accepted uncritically analysing a number of examples chosen ad
hoc. Firstly, it seems implausible that native English speakers have
such a fine and detailed awareness of the deep phonological processes
of their own language, and secondly, there are many cases where the
alleged depth of English orthography does not show: one writes speak
but speech, collide but collision, sight but see, etc.
A solution is to consider written English as a system that initially was strongly based on phonology and then went slowly acquiring a somewhat logographic component. This trend could be seen
as an adaptation of writing to the change of its functions: a phonological notation was absolutely necessary in a period when the only
raison dtre of a written text was to be read aloud, but with the diffusion of silent reading, this is no longer needed. As Berry (1977: 10)
suggests, [t]he reader reads fluent English or French or German
138
efficiently only insofar as he treats the written language as if it were

ideographic.
In the particular case of English, the evolution of its spelling
shows also a striking parallelism with the evolution of spoken English:
the spelling was more shallow at the time when English was a flexive
language, closer to other European ones, and evolved into a graphic
system with a high degree of logography and morphography until
today, while English is considered typologically closer to an isolating
language like Chinese rather than to its Indo-European cousins.8
4.2. French
The development of written French is somehow tied to that of
English orthography, since French is responsible for many changes
that occurred in written English, either directly through the Norman
invasion, or indirectly, thanks to the prestige of French culture during most of the Modern Age. The two graphic traditions share many
similarities: French also, at the beginning, was written in a way which
reflected the phonetic reality simply and precisely. It was the spelling employed by jesters, singers and poets who, because of their job,
needed a rapid transcription of the texts they used, characterised by a
typically oral structure. When French became important in public life
replacing Latin in official documents, the scribes began to add many
unnecessary letters. Flaunting the presence of etymological letters
that were no longer pronounced was a way to confer French the same
prestige of its ancestor (Fournier 1940: 261).
Among the main difficulties of current French orthography, we
can name the following:
the alternation between diacritical accents and silent letters to
indicate the timbre of a vowel, e.g. fidle /fidl/ vs. nette /nt/;
/E/ in an open syllable can be realised as [e], [] or [], whereas
in a closed syllable it is always []; however, there is a medium
[E] that is halfway between [e] and [] and usually appears
before a syllable containing []; the instability of the schwa
makes the syllable itself unstable and it can be realised as an
open syllable or as a closed syllable, so that the timbre can be
itself open or closed. The French Academy normally indicates
this medium [E] by <>, that usually stands for [e], although
the pronunciation tends towards [], e.g. mdecin [meds] can
switch to [mds];
among the sixteen vocalic phonemes of French, four are nasal
phonemes, whose notation is still problematic;
139
Antonio Baroni
many graphemes are employed to distinguish between homophones (cinq, 9 sein, sain, seing, saint stand all for [s ] ; cf.
English buy, bye, by for [ba], Danish vr, hver, vrd, vejr for
[v]); moreover, many letters are conserved for matters of
prestige, sometimes etymologically motivated, (philosophie,
xnophobie) sometimes not (lys from Latin lilium, nnuphar
from Arabic ninufar).
One of the most striking aspects of French orthography is the
discrepancy between oral and written morphology. Written French
requires much more grammatical knowledge than spoken French.
In speech, there is often no difference between the singular and the
plural of nouns and only the article, context or agreement can disambiguate the grammatical number of the noun. Some nouns are not
even differentiated even between the masculine and the feminine
form. In verbal conjugation, the first three persons and the sixth have
often the same desinence. In writing, on the contrary, masculine and
feminine are normally differentiated, the plural is always indicated
by <-s>, the verbal conjugation preserves different desinences for
almost every grammatical person, and so on, e.g. <lami> the friend
(m.) [lami] vs. <lamie> the friend (f.) [lami]; in the series je parle,
tu parles, il parle, nous parlons, vous parlez, ils parlent (I speak, yousing. speak, he speaks, we speak, you-pl. speak, they speak) parle,
parles, parlent stand all for [pal()].10 These data could lead us to the
conclusion that French native speakers morphological competence is
often expressed only graphically.
It is interesting here to point out that written French morphology is much more natural according to the Natural Morphology
framework (see Dressler 1987a, 1987b, 1990, Dressler et al. 1987,
Mayerthaler 1981, Kilani-Schoch 1988, Wurzel 1994, Dressler,
Mayerthaler et al. 1987) than its spoken counterpart. According to
the principle of diagrammaticity or constructional iconicity, forming
the plural adding a suffix, such as the morpheme {-s}, corresponds
to the first degree of the scale of diagrammaticity, namely, the most
natural one, and so it happens in written French: <chien> (sing.),
<chiens> (pl.). But in spoken French, the plural is normally formed
without any alteration of the base, through the morphological technique of metaphoricity, eg. /j/ (sing.), /j/ (pl.). Metaphoricity corresponds to the third degree in the scale of diagrammaticity. If we
consider the formation of the feminine, written French still employs
the agglutinative affixation (1st degree) but spoken French employs
subtraction, which is considered anti-iconic (since feminine is more
marked than masculine).
140
Concerning morphological naturalness, writing has the primacy

also when it comes to morphotactic and morphosemantic transparency
(a form is morphotactically and morphosemantically transparent if
it possible to find, in the structure of the signifier, morphs which correspond to components of meaning) and transparency of encoding
(according to which, synonymy, homonymy, suppletion and allomorphy
are not natural) (Thornton 2007: 164-165, Crocco Galas 1998: 25-41).
Phonological naturalness is often in contrast with morphological
naturalness; the natural phonological processes tend to diminish the
articulatory effort, whereas morphological processes tend to increase
it to improve the perception for the listener.
Part of the opacity of French orthography may be due to the need
to preserve a certain degree of distinctiveness in a language where
many important grammatical distinctions were lost because of phonological processes.
4.3. Danish
It is interesting to also consider Danish orthography here because
it is somewhat similar to English orthography, but unlike the English
one its use is limited to a relatively small community.
According to a small scale study comparing 13 European orthographies (Seymour, Aro & Erskine 2003), Danish and English children lag
far behind other children in reading and writing performances by the
end of the first year of school. The origin of Danishs orthographic depth
may be found in the choice of the first scribes, during 13th and 14th centuries, to select diasystemic spelling forms that did not reflect any specific
dialect (a sort of interdialectal writing system, like Chinese script). As a
result, the orthography they created reflected the archaic pronunciation,
so for example, lov [la] was spelled <logh>, although the last phoneme
was already []; similarly, final [], originally [], was still spelled <th>.
Since the 1200s, Danish has undergone several phonological
changes that did not affect related languages such as Swedish, which
has a much more shallow orthography (Elbro 2003: 33). Danish,
Swedish and Norwegian maintain a certain degree of mutual comprehensibility, but spoken Danish is the most difficult to understand
(Basbll 2005: 7). What makes the Danish language particularly
unsuitable for a phonologically transparent notation through the
Latin alphabet are the following features:
an extremely rich vowel system, with far more distinctions in
vowel quality, both in rounded and unrounded vowels, than
other related languages;
141
Antonio Baroni
radical reduction processes, which make the language very

hard to understand for foreigners; these phenomena affect
vowels, obstruents and approximants/glides as well;
schwa-assimilation, which means that the schwa, as a final
neutral vowel, can either manifest itself as such or be assimilated by the preceding consonant, turning it into a syllabic segment, e.g. hedde to be called, [he] vs. [he];
a unique feature of Danish is the std (glottal stop), whose
presence is unpredictable in synchrony and affects word formation and other morphological processes (Basbll 2005:8).
Even though std is not generally indicated consistently in the
orthography, there is a certain correlation between std and
unpronounced <d>, e.g. spild waste [sbil], hund dog [hun],
mand man [man] vs. spil play [sbel], hun she [hun], man
one, people [man];
many morphonemes can have different phonological and
phonetic counterparts, e.g. the morphonemes |v| and |g|
can be both realised phonologically as /v/ and phonetically
as []; in these cases the orthography tends to operate at a
morphonemic level, rather than at a phonological or phonetic
one, cf. <filolog> philologist, |filolog|, /filolov/, [filolo]
and <sav> saw (noun), |sav|, /sav/, [s] (Basbll 2005:
74-77);
given a certain number of cases of homophony between words
which differ greatly with regard to their grammatical function,
Danish employs the orthography as a means of disambiguation, e.g. at bore (to drill, infinitive form) and borer (drills)
sound the same but are spelled differently; similarly, the
homophonous endings -ene (plural definite nouns) and -ende
(present participle of verb) are distinguished in writing for
morphological reasons (Elbro 2005: 40).
another striking feature of Danish compared to other
European languages is the distinction between prosodic
commas and grammatical commas; during the 20 th century, people tried to follow the grammatical use of commas,
namely, commas around sentences regardless of utterance
prosody, but later, in the 90s, the existence of two competing systems began to be felt as a problem by many. Today
the placement of a comma before a subordinated sentence is
optional (Basbll 2005: 89-90), cf. English He says (that) it
tastes good, Italian Dice che buono, Danish Han siger, det
smager godt.
142
4.4. Finnish
In the scientific literature, the most common example of a shallow orthography is undoubtedly Finnish; this Finno-Ugric language
appears, in its written form, to be very close to the phonemic ideal,
inasmuch as each grapheme corresponds to one phoneme and there
are no ambiguous or contextual graphemes (the only exception is the
digraph <ng> that stands for //). Furthermore, Finnish orthography
represents both vowel and consonant length, which are distinctive
in speech, by doubling the grapheme, e.g. <mutta> /mut a/ but vs.
<muuttaa> /muta/ to change vs. <muta> /muta/ mud.
How is such a nearly perfect system possible? Firstly, the
Finnish language is spoken by more or less 6 million people, the
dialectal fragmentation is minimal and almost the entirety of its
speakers live between Finland and Sweden; secondly, before the 16th
century there was no Finnish literature, as the Bible and the academic publications were written in Latin or in Swedish and a true
orthographical standard was reached only in 1880, after Finnish
had already undergone several important phonological transformations (// > /d/, // > /ts/, // > /v~/); thirdly, the Latin alphabet fits
Finnish phonology. Finnish possesses 13 consonants and 8 vowels,
with a ratio of 96 consonants each 100 vowels in speech. Vowels conserve their full value in unstressed syllables and the dominant principle in word formation is to avoid any phoneme that requires a difficult articulation; a syllable never begins with a consonantic cluster
and this greatly limits the phonological resources and the number of
monosyllabic roots.
The voiced consonants /b, d, / do not belong, originally, to the
Finnish language, but lately people pronounce them under the influence of foreign loans and of orthography, which employs the graphemes <b, d, g>, triggering a fortition process (but, according to Brown
& Koskinen 2011, only the phonological status of /d/ is undisputed,
whereas /b/ and // are still pronounced [p] and [k] in everyday life).
The trend of the most educated speakers to pronounce <b, d, g> as
voiced consonants has provoked a more tense pronunciation of voiceless stops /p, t, k/ to keep the phonemes distinct. Given that many
morphological oppositions are indicated by both vocalic and consonantic length, Finnish pronunciation requires a greater accuracy than
other European languages (Hakulinen 1961: 5-17).
If we compare the Finnish structure to that of French and
English it is clear that:
Finnish has 21 phonemes, vs. 36 of English and 37 of French;
143
Antonio Baroni
Finnish is spoken by 6 million people in Finland and Sweden,

whereas 205 million people speak French, in France, Belgium,
Switzerland, Canada, Haiti, in many African countries, etc.
Even more people speak English (around one billion and 351
million, including those who speak it as second language) and
it is the official language in several countries (UK, Ireland,
USA, Canada, Australia, New Zealand, etc.);
the first attempts to write in English and French date back to
Middle Age, Finnish literature began in the 19th century;
since they have an orthographic standard, French and English
have undergone several phonological processes (especially with
regard to vowels in English and consonants in French), whereas Finnish phonology was relatively stable when a graphic
standard was established;
Finnish has few monosyllabic words, whereas French and
English have many.
4.5. Italian
Italian, just like Finnish and German, owes its standardisation
to the fixation of a written language, created by mixing features of
different dialects. Written Italian is a sort of amended Florentine,
namely an orthography based upon Florentine speech without the
local features that were too marked (like Tuscan gorgia, the fricativisation of voiceless stops in post-vocalic position) and with some
Lombard influences.
The stability of Italian orthography is a relatively recent phenomenon and until the 19th century there were many possible spellings, especially to indicate the palatal phonemes that did not exist
in Latin (such as // and //), to distinguish between // and /e/ and
between // and /o/, but few attempts were made to indicate /i/ and /u/
differently from /j/ and /w/ (Migliorini 1994: 146-147, 206-207).
Even if Italian is thought to have a very transparent orthography, it is less shallow than others. It maintains some etymological
letters, such as <q>11 in words of Latin origin, silent in words
like scienza, efficiente, deficiente, cielo and <h> in ho, hai, ha, hanno
I have, you-sing. have, he/she/it has, they have.12 The main stress,
albeit variable, is indicated by a graphic accent only in polysyllabic
oxytones and in some monosyllabic words with a morpholexical distinctive value. The timbre of <e> and <o> is not generally indicated
and there are some other inconsistencies. However, Italian spelling
is based upon a straightforward relation between writing and speech
144
and a native speaker generally knows how to write or how to read a

word that he or she has never heard before.
Among the Romance languages, Italian is the one which
remained closer to Latin phonology (if we do not consider Sardinian),
so the alphabet still fits the language. From a strictly formal point of
view, Italian orthography is defective, because it does not distinguish
between // and /e/, // and /o/, /i/ and /j/, /u/ and /w/, /s/ and /z/, // and
//, but these distinctions are not relevant for most native speakers.
A more phonetic orthography would risk to impose an artificial pronunciation and to discriminate those who do not follow the Tuscan
model. Eventually, the choice not to indicate these differences turned
out to be functional.
Nevertheless, if in Finland a highly transparent orthography
correlates with a high degree of literacy of the population (around
100%), in Italy teachers have pointed out that the average level of
orthographic competence of Italian students is decreasing more and
more dramatically and there are many cases of adult illiteracy and
functional illiteracy (Giscel 2007).
4.6. Opacity vs. shallowness
In summary, we may assume that there is a high probability that
an alphabetic orthography is opaque if:
there is a very old literary tradition, so that the need to conserve the graphic image of words has come to light and is felt
by language users;
the language is widespread and there are many local varieties
(one exception is Danish, which has an opaque orthography
but is spoken only by 6 million people13); in these cases, opaque
orthography permits people to communicate even if their spoken
varieties differ greatly, just as a logographic system would do;
the available graphemes are not sufficient to transcribe all the
phonemes of the given language;
the language has undergone many phonological changes during or after the standardisation of the spelling;
the language undergoes significant phenomena of vowel reduction. It is often the case of stress-timed languages (such as
English, Swedish, Russian),14 which are less likely to have an
univoque written representation of vowels than syllable-timed
languages (such as Finnish, Italian, Spanish). This is not
always true, though; for example, French is a syllable-timed
language but has an opaque orthography, whereas German is
145
Antonio Baroni
a stress-timed language but it is quite regular in the graphic

representation of vowels;
the given language has many homophonous words.
By contrast, alphabetic orthographies tend to be transparent if:
the standard orthography was introduced (or revised) in relatively recent times;
the language has a relatively stable phonological structure in
which most vowels are pure and vowel reduction phenomena
only happen at a phonetic level but not at a phonemic one;
the number of phonemes and the number of graphemes do not
differ too much;
the language has not many varieties or it is not very widespread outside its country (some important exceptions are
Turkish, Spanish, Italian);
the language has few cases of homophony;
the language has accepted an underspecified notation, in
which some phonemes are neutralised in writing (e.g. Italian
orthography).
5. The naturalness of artificiality
According to Walter Ong (1986: 124), writing is a kind of technology, since it has been invented by human beings and it is not natural
and spontaneous like speech, but even if writing is artificial, yet artificiality is natural for human beings.15
Our assumption here is that, since writing conveys language
through graphemes, as much as speech conveys language through
sounds and signing conveys language through gestures, then writing,
which is itself artificial, starts to be subjected to the same constraints
of language and to work as language. Its conservative nature led
many scholars to think that orthography can be reformed at will and
that it should perfectly reflect speech. But this is not how language
works. We as linguists know very well that the different planes of
language are very seldom in a 1:1 relationship and that isomorphism
is a concept artificially constructed in grammar books, but that does
not show in the language used by people every day. So, most of the socalled discrepancies of written language might be merely a result of
the adaptation of the written medium to the specific logic of language.
Primus (2005: 240) points out that [o]rthographies have been criticised for mapping spoken language imperfectly. But functional imperfection is a natural trait of language.
146
The spontaneous deviations from the norm operated by language

users are not really disturbing for the system, it has been the overzealousness of pedantic purists to affect orthographies in unnatural
ways throughout history.
As we have seen, the history of English and French orthographies is normally depicted as a corruption from a shallow to an
opaque orthography. Similarly, Gelb and other scholars thought
that all the writing systems had to eventually evolve into alphabets,
because of the alleged inherent perfection of the alphabetic principle.
But linguists should not think like that; when considering writing,
they should be able to find what is natural in artificiality, that is,
how natural linguistic processes find their way and show themselves
through the artificial written medium. If we do that, it is clear that
there is no such thing as a linear evolution from picto/logographic to
phonemic writing. Each writing system evolved and keeps on evolving
but they hardly become purely phonemic.
One of the most ingenious writing system today is the Japanese
one, which employs all possible solutions in writing: ideography,
logography, syllabography and phonography. Besides or thanks to
that, very high rates of literacy are achieved in Japanese society.
English orthography works in a very similar way to Japanese one, it
is just less apparent since English uses only the Latin alphabet and
Japanese employs kanji, hiragana, katakana, furigana (small kana
printed next to a kanji to indicate its pronunciation), Arabic numerals
and Latin letters.
5.1. Natural processes in private written communication
If the hypothesis that human beings are more at ease with a
perfectly phonemic system were true, we would expect spontaneous,
privately written communication to work in this way, but if we analyse the very recent evolution of writing habits developed with Short
Message Texting and the Internet, we still find the same old tricks
employed by Ancient Egyptians (e.g. rebus, such as m8 for mate) or
by the scribes of the Middle Age, e.g. abbreviations such as <q> for
qu in Spanish or <w , , > for with, that, thou in Middle English
(McLaughlin 1963: 44-45).
Some other examples:
 or ; the name of the grapheme /ju/ is
homophonous with the pronoun you (rebus); in the second
case, the verb love is replaced by the symbol of a heart (ideography); the use of <> has become so common lately that, to
147
Antonio Baroni
cope with its absence from the keyboard, internet users write
<3. Given that all the words in the sequence I love you are
monomorphemic and monosyllabic, writing would, at
the same time, satisfy the economy principle and create an isomorphic relation between graphic signs and syllables;
In Italian, <x> is both a letter and the mathematical symbol for the multiplication called per; because of that, <x> can
replace both the preposition per for (logography) and the
phonic sequence /per/ (rebus), e.g. Lo faccio x te I do it for you,
xfetto, xdono, xso instead of perfetto, perdono, perso perfect, forgiveness, lost. The homophony between the names of numbers
and other words allows to write for as <4> in English and to
write sei you-sing. are as <6> in Italian (cf. Mioni 2009: 36-38);
Very common abbreviations reduce words to a three-grapheme
root, a process which reminds closely of Semitic writings such
as Hebrew and Arabic scripts: e.g. English mmt for moment,
Spanish tmb for tambin; Italian cmq for comunque, etc. It
is interesting to note that very often these triconsonantic
abbreviations ignore the postvocalic nasal (moment, tambin,
comunque), a phenomenon very well documented in many
Mediterranean writing systems (Miller 1994: 19, Justeson
1976: 76), as well as in childrens early spelling in English,
Dutch and Spanish (Read 1986: 80-86);
Another bit of evidence that expert readers and writers do not
produce, spontaneously, phonologically detailed utterances,
is the general phenomenon of the abandonment of diacritics.
Young French and Czech speakers, whose orthographies make
a great use of diacritics, when writing texts with their mobile
phone or chatting on the internet, do not use them. A study
about Portuguese native speakers communicating on the internet with a non-native keyboard underlined that they tried to
mark words (e.g. using an apostrophe instead of an unavailable accent) only when there was the possibility of semantic
ambiguity but they had no interest to do so when the use of
diacritics was related only to phonology (Jensen 1995). Context
itself helps decoding meaning. About this, Nina Catach (1992:
24) defines French diacritics gnants, lourds grer, peu
utiles dans bien des cas and points out that in private their
use is already decreasing. This seems a general phenomenon,
applying to very distant languages: Coulmas (1989: 237) talks
about Vietnamese, where diacritics [i]n handwritten script (...)
are easily and often omitted; and in print they make for a clut148
tered appearance. This tendency does not appear surprising if

one thinks of Hebrew: the only written materials where vowels
are normally noted by matres lectionis are, besides the Bible,
books for children and poetry, but in everyday life, adult readers do not need them (Sampson 1985: 89).
The situation is even more complex: alongside with a return
to logography, consonantic writing, etc. we also encounter phonetic
spellings, e.g. French koi and jam instead of regular quoi /kwa/ and
jamais /ame/ or American English dont cha or whatcha say for dont
you and what (do) you say.
As we have already pointed out, these strategies are not necessarily related to the new media but are as old as writing. Let us think
about cursive handwriting: in private use one does not feel the need
to draw the shape of each grapheme in a clear, univocal way; on the
contrary, the more expert one is, the faster he/she writes and probably the less clear his/her writing will appear. We can then assist to
phenomena of graphemic distinctiveness reduction, in which many
graphemes look very similar, if not identical, e.g. , in the handwriting of many people, could stand for <n> or for ; <l > normally
stands for < l > but in fast handwriting could merge with <t> or <f>;
the same applies to <e > and <c > and so on. Something similar happens in East Asia, where
to write neatly to an educated man could actually be seen as insulting, since it suggested that he was thought incapable of reading
cursive forms. Normal handwriting () is some way removed from
the neatness of print, fusing what are printed as separated dots and
strokes into continuous, smooth motions of the brush. (Sampson
1985: 192).
After all, cursive handwriting stand to print (in writing) as allegro forms stand to lento forms (in speech; cf. Dressler 1975); when one
speaks fast, in a familiar context, with no or little social pressure, he/
she applies a series of phonological reductions but yet, his/her speech
is still comprehensible to a native speaker; similarly, graphic reductions, if they are not excessive and are put in the right context, do not
affect the intelligibility for the reader.
5.2. New writing systems
In the last centuries, many languages that were only spoken
acquired a written form. Normally when a society needs its own
script, it adapts or adopts an existing one but from to time to time
149
Antonio Baroni
there have been some individuals who, once aware of the existence
of literacy, invented a writing system from scratch. The creation of a
script is called grammatogeny, an operation that can be sophisticated if
it implies a certain degree of phonetic knowledge, or unsophisticated,
if the maker of the script cannot read any language and does not
know anything about phonetics (Daniels 1996: 579).
Among sophisticated systems it is worth to consider the Pollard
script, invented by Samuel Pollard in Southern China around 1887
to transcribe the Western Hmong language. The script is composed of
32 letters corresponding to single consonants or consonantic groups
and 37 diacritical letters corresponding to vowels and nasal finals,
placed differently (at the top, upper right, middle or bottom) depending on the tone. The shape of the symbols resembles the Latin letters,
besides being more geometrical, and the system works almost like
hangl, i.e. an alphabet where the graphemes are grouped in syllabic
glyphs (Daniels 1996: 580).
It is said that the Pollard script was influenced by the Cree syllabary, invented in 1840 for Cree and Ojibwe languages in Canada and
then adapted to Athabaskan and Inuit. Like Pollard script, Cree can be
defined as a featural-cum-abugida system (Daniels 2001); single vowels are indicated by a triangle shape and the rotation changes according
to vowel quality; consonant-initial syllables indicate the consonant by
the shape and the vowel by the orientation (e.g. Inuit stand
for /i, u, a/ and stand for /pi, pu, pa/) (Nichols 1996: 608).
Pollard and Cree scripts seem to work quite well for the languages they convey, but the study of unsophisticated writing systems
is more effective in identifying natural tendencies. It appears that
writing systems devised independently from each other end up more
often being syllabaries rather than alphabets and some of them pass
through a logosyllabic stage. Edgerton, in his criticism of Gelbs theory (Edgerton 1952: 287), points out that, among new scripts, Cherokee
(created by Sequoyah around 1810), Vai (designed by Dualu Bukele of
Jondu in the 1820s in Liberia), Alaska (devised by Uyaqoq between
1901 and 1905) and Bamum (invented by King Njoya of the Bamum
tribe after he had a dream about it) started out as mainly logographic
systems but quickly became syllabaries. The first part of Gelbs theory
is then confirmed, but the inevitability of the evolution of these systems into alphabets does not show, since they are still employed as
syllabaries, although Alaska and Bamum scripts show some tendencies toward alphabetisation (Gelb 1963: 209).
Autochthonous African alphabets, such as Bassa and Nko, were
introduced respectively by Flo Darvin Lewis, a Bassa native speaker
150
who went to America to study medicine (he was therefore deeply

affected by English alphabetic orthography) and by Soulemayne
Kante in 1949, as an identitary action against those who considered
Africans inferior for not having a script of their own (Dalby 1969: 162,
cf. Pasch 2008).
The Neo-Tifinagh script ( <tifina>) is an alphabet and
is currently used in Morocco and Algeria to transcribe various Berber
languages; in its original form it was an abjad but vowels are now
noted as well. The passage from consonantic to alphabetic writing
might depend on three main factors: the refusal of the Arabic script
as an identitary action,16 the influence of the Latin alphabet and the
linguistic structure of Berber languages for which an abjad is not well
suited (OConnor 1996, Pasch 2008).
As suggested by Daniels (1996: 579), observable script inventions have much to teach about the possible scenarios of the three
ancient grammatogenies (Sumerian, Chinese, Maya). All these three
languages have a similar typology, i.e. morphemes are mostly monosyllabic, a feature which allows a good fit with a logosyllabic system
(Daniels 1992: 83).
As we have pointed out at 3.3, the most intuitive unit for a
human being is the word or the concept, so it comes as no surprise
that the first attempts of both Sequoyah and Uyaqoq consisted in
devising a symbol for every word, and it is even less surprising that
eventually they both chose to denote syllables and not phonemes.
Put in other words, logographic writing is very natural but not
economic at all (since it contrasts with the criterion of the size of
the inventory 3.2), so the most natural unit right after the word/
morpheme is the syllable (cf. Dressler & Dziubalska-Koaczyk 1994).
The existence of brand new alphabets cannot be taken as an evidence
for the naturalness of a segmental notation, since the prestige of the
Roman script is today too important to ignore and it sure has a weight
in modern grammatogenies.
6. Conclusion
The main aim of our article was to give a quick overview about
the linguistic fit of the different writing systems, in the attempt to
refute the Western axiom according to which the alphabet is the best
system whatsoever and shallow orthographies are superior to opaque
ones. While doing this, we were also hoping to contribute to the still
scarce amount of graphemic studies, namely, a linguistic approach to
151
Antonio Baroni
the problems of writing. To do so, we have tried to abandon the usual

path (a forced parallelism between graphemics and phonology) and
we ventured in two domains which, to our knowledge, had not met yet
graphemics in a consistent way: Natural Linguistics (see Dressler et
al. 1987) and neurolinguistics.
If linguists wish to treat graphemics as a somewhat independent
plane of language,17 they must find natural tendencies in the approach
of human beings to reading and writing. The current studies in neurology have demonstrated that specific areas of our brain respond to the
sounds of language and that other areas respond to graphemes. If it
seems plausible that spoken language has a biological basis and may
be largely innate, the same theory seems unlikely when it comes to
writing, due to its relatively recent invention. According to Dehaene,
it was writing that modelled itself to be adequate to brain circuits,
not vice versa (Dehaene 2009: 1-10). Once we realise that the shape of
our writing systems is not wholly accidental but that shared similarities between geographically and chronologically distant traditions are
not coincidences, but are due to neuronal constraints, we may dare to
identify natural tendencies in reading and writing.
If we consider the question without any relationship to speech,
the most natural writing system would display characters composed
of no more than three strokes (for the writer) but whose configuration
is perceptively redundant (for the reader).
When it comes to the naturalness of the encoding process of a
specific language, pictography, ideography and logography seem the
most natural methods, but they require a huge mnemonic effort (and
a great neuronal recycling). Moreover, it appears that, even if in the
reading process both the graphemic-phonemic and the graphemic-lexical ways are activated, the most important one remains the former,
since the universal phonological principle (UPP) () predicts automatic activation of phonological information in words in all languages, that is, prelexical phonological coding, even in deep orthographies
such as Chinese (Perfetti & Zhang 1991, 1995 cited in Chikamatsu
1996: 64). For the human being, the syllable would be a more natural unit than the phoneme, so syllabaries and abugidas would be
somewhat superior to alphabets. Nevertheless, alphabets require less
mnemonic effort and are more suitable for languages with a complex
syllabic structure. After all, each language has its own fit in relation
to writing, so any claim about the primacy of the alphabet has no linguistic value, but only an ideological one.
A linguistic analysis of opaque alphabetic orthographies reveals
that some alleged inconsistencies may play a positive role (e.g. main152
taining a certain degree of morphological naturalness in languages

where phonological processes have heavily affected morphotactic
and morphosemantic transparency) and may be a natural evolution caused by the loss of the original linguistic fit of the script (e.g.
English, French, Danish, etc.).
Finally, empirical observations of spontaneous written utterances
in private communications show that, once a sufficient level of isomorphism with speech is established, all kinds of relation with all planes
of language (phonetics, phonology, morphology, etc.) can be fruitfully
exploited and that the level of accuracy can decrease, to some extents,
without interfering with the communication process.
Address of the Author
Universit degli Studi di Padova, Facolt di Lettere e Filosofia,
Dipartimento di Studi Linguistici e Letterari, via Beato Pellegrino 1,
35137, Padova, Italy <[email protected]>
Notes
Actually, it has been proven that logographic systems are indeed logosylor more precisely morphosyllabaries, since each character stands for a
morpheme, and the characters can be used for the sound of the morpheme as well
as for its meaning. (Daniels 2001: 43).
2
Syllables other than CV are rarely represented. In the Linear B, CCV would
be written as <CV>+<CV>, e.g. <ko-to-i-na> for [ktoyna], <de-ka-sa-to> for
[deksato], etc. (Miller 1994: 18-19). In Old Accadian, CVC sequences are often
rendered as <CV> + <VC>. Moreover, the few existing graphemes standing for
CVC sequences do not reflect the actual syllable boundaries, e.g.. <is-pur-am>
for [ispuram], rare variant of <is-pu-ra-am>. Signs for complex rhymes, such as
<VCC>, are not attested in any writing system (Dressler & Dziubalska-Koaczyk
1994: 68-69).
3
Graphonemes are the object of study of what Mioni (2009: 19) calls systematic
graphemics (as opposed to autonomous graphemics, which deals with the internal
structure of writing systems, regardless of their phonetic value).
4
Many authors other than Daniels made similar proposals, see Gleitman &
Rozin (1977), Bellamy (1989), Coulmas (1989), Aronoff (1992), Faber (1992).
According to Miller (1990, 1994) phonemes are part of the implicit linguistic
knowledge (but not of the explicit linguistic knowledge, see Chomsky 1986).
5
Probably inspired by Indian grammars made known in China by the
Buddhistic teaching.
6
This is evident if we consider Roman numerals: the numbers from one to three are
represented by, respectively, one stroke, two strokes and three strokes but to represent
four and the following numbers, other methods are used: <I, II, III, IV, V, etc.>.
7
From a diachronic point of view, the graphemes of Latin and Greek alphabets
derive, through Phoenician, from Proto-Sinaitic glyphs based on pictograms. If
1
labaries
153
Antonio Baroni
iconicity is still there in modern writing systems, it is likely to be residual or accidental, but even so, that does not imply that users cannot find a certain degree of
motivation in the shape of graphemes, be it related to sound, meaning or the relationship with other graphic elements.
8
In both languages, the majority of words are monosyllabic, even though
Chinese is slowly acquiring more and more polysyllabic words through the creation of polymorphemic compounds.
9
Cinq is pronounced [sk] before a vowel and can either be pronounced [s] or
[sk] in the other possible contexts.
10
Although in liaison an epenthetic [t] appears.
11
<q> in Italian stands for /k/ but can appear only before . In this position it
rivals with <c> and their distribution is based on etymological criteria, cf. cuore
vs. quale; both come from Latin, the former from cor, cordis, the latter from qualis,
qualis.
12
Spelling distinguishes <ho, hai, ha, hanno> from their homophones <o, ai, a,
anno>, which mean, respectively, or, to the-pl., to, year.
13
The opacity of Danish orthography can partially be due to the necessity to
conserve the mutual intelligibility between Nordic languages, considering also
the central role of Denmark in the Kalmar Union (1397-1523), during which
Denmark, Sweden, Norway, Iceland, Greenland and the Faroe Islands were
united. Norway and Denmark remained a single political entity until 1814 and
even today Bokml, one of the two official Norwegian written languages (the
other one being Nynorsk), differs minimally from written Danish. Denmark lost
its rule over Iceland only in 1943, whereas Greenland is still part of the Danish
Kingdom.
14
As Linell (1979: 56) suggests, careful pronunciations need not be, and are very
often not, the most frequent or normal pronunciations of the word forms involved.
On the contrary, they will be somewhat artificial and pedantic, particularly perhaps in languages that have heavy stresses and thus normally a great deal of
reduction (e.g. English, Danish, Russian). In such cases, it may be that speakers
may even construct full-vowel plans which are virtually never realized as such
(). Possibly such abstractness may be due to conventional orthography.
15
Artificiality pertains to the Peircean sign type symbol, which is more complex
than the icon. As a matter of fact, orthographies tend to be more symbolic and
less iconic. Orthographic rules are legisigns, namely, laws that are signs (cf. Peirce
1980).
16
The choice of a script is a very strong identitary act for a society. As Sebba
(2006: 100) points out, debates on orthography become symbolic battles over
aspects of national, regional or ethnic identity. Let us take the Tatar language
as an example. As a Turkic language, Tatar used to be written using the Arabic
script, but switched to the Latin alphabet in the 1920s and later to the Cyrillic.
In the late 1990s, the Tatarstan government decided to gradually restore the
Latin alphabet, but soon after, the Russian government accused Tatarstan to
threaten Russian unity by doing so, and in 2002 a bill was approved imposing all
national languages of the Russian Federation to use alphabets based on Cyrillic.
Consequently, in Tatarstan, the use of Latin alphabet to write Tatar has diminished in public situations, but individual users still employ it in their personal
websites or in private communication (Suleymanova 2010: 55-56).
17
Written language had already been analysed as partially independent from
speech by the members of the Prague School, with a functionalist approach (cf.
Artymovy 1932, Vachek 1939, 1973).
154
Bibliographical References
Albrow Kenneth Harold 1972. The English writing system: Notes toward a
description. London: Schools Council.
Aronoff Mark 1992. Segmentalism in linguistics: The alphabetic basis of phonological theory. In Downing, Lima & Noonan 1992. 71-82.
A rtymovy Agenor 1932. Fremdwort und Schrift. Charisteria Guilelmo
Mathesio quinquagenario a discipulis et Circuli Linguistici Pragensis
sodalibus oblata. Prague: Prask Linguistick Krouek. 114-118.
Basbll Hans 2005. The Phonology of Danish. Oxford & New York: Oxford
University Press.
Bellamy Rufus 1989. Bellerophons Tablet. Classical Journal 84. 289-307.
Berry Jack 1977. The Making of Alphabets Revisited. In Fishman Joshua
Aaron (ed.) 1977. Advances in the Creation and Revision of Writing
Systems. The Hague & Paris: Mouton. 3-16.
Bolinger Dwight Le Merton 1946. Visual Morphemes. Language 22.4. 333340.
Brown Jason & Pivi Koskinen 2011. On Voiced Stops in Finnish. Linguistica
Uralica 47.2. 94-102.
Catach Nina 1992. propos de laccent plat. Liaisons-HESO 19-20. 20-25.
C hangizi Mark A. & Shinsuke Shimojo 2005. Character Complexity and
Redundancy in Writing Systems over Human History. Proceedings:
Biological Sciences 272. 267-275.
Changizi Mark A., Qiong Zhang, Hao Ye & Shinsuke Shimojo 2006. The structure of letters and symbols throughout human history are selected to
match those found in objects in natural scenes. American Naturalist
167.5. 117-139.
C hiang Yee 1973. Chinese Calligraphy. Cambridge & London: Harvard
University Press.
Chikamatsu Nobuko 1996. The effects of L1 orthography on L2 word recognition: A study of American and Chinese learners of Japanese. Studies in
Second Language Acquisition 18. 403-432.
Cho Jeung-Ryeul & Catherine McBride-Chang 2005. Levels of Phonological
Awareness in Korean and English: A 1-Year Longitudinal Study. Journal
of Educational Psychology 97.4. 564-571.
Chomsky Noam 1986. Knowledge of Language: Its nature, origin, and use. New
York: Praeger.
Chomsky Noam & Morris Halle 1968. The sound pattern of English. New
York, Evaston & London: Harper & Row Publishers.
Consani Carlo, Cristiano Furassi, Francesca Guazzelli & Carmela Perta
(eds.). Atti del 9 Congresso dellAssociazione Italiana di Linguistica
Applicata Oralit e Scrittura. In memoria di Giorgio Raimondo
Cardona (February 19-20, 2009). Pescara: Guerra.
Cossu Giuseppe, Donald Shankweiler, Isabelle Y. Liberman, Leonhard E. Katz
& Giuseppe Tola 1988. Awareness of phonological segments and reading
ability in Italian children. Applied Psycholinguistics 9. 1-16.
Coulmas Florian & Konrad Ehlich (eds.) 1983. Trends in Linguistics, Studies
and Monographs Writing in Focus 24. The Hague: Mouton.
155
Antonio Baroni
Coulmas Florian 1983. Writing and literacy in China. In Coulmas & Ehlich
1983. 239-254.
Coulmas Florian 1989. The writing systems of the world. Oxford: Blackwell.
Coulmas Florian 2009a. Back to the Future: Literacy and the Art of Writing
in the Age of Cyberspace. In Consani, Furassi, Guazzelli & Perta 2009.
93-114.
Coulmas Florian 2009b. Evaluating Merit the evolution of writing reconsidered. Writing Systems Research 1.1. 5-17.
Crocco Galas Grazia 1998. The parameters of natural morphology. Padova:
Unipress.
Daniels Peter T. & William Bright (eds.) 1996. The worlds writing systems.
New York: Oxford University Press.
Daniels Peter T. 1992. The syllabic origin of writing and the segmental origin
of the alphabet. In Downing, Lima & Noonan 1992. 83-110.
Daniels Peter T. 1996. The Invention of Writing. In Daniels & Bright 1996.
Daniels Peter T. 2001. Writing Systems. In Aronoff Mark & Janie ReesMiller (eds.) 2001. The Handbook of Linguistics. Malden & Oxford:
Blackwell. 43-80.
Dehaene Stanislas 2009. I neuroni della lettura. Milano: Cortina. (Orig. vers.
Paris: Odile Jacob 2007).
Diringer David 1948. The Alphabet: A Key to the History of Mankind. New
York: Philosophical Library.
D owning Pamela, Susan D. L ima & Michael N oonan (eds.) 1992. The
Linguistics of Literacy. Amsterdam & Philadelphia: John Benjamins
Publishing Company.
Dressler Wolfgang U. 1975. Approaches to Fast Speech Rules. In Dressler
Wolfgang U. & Mare Frantiek V. 1975 Phonologica 1972. Vienna:
Wilhelm Fink. 219-234.
Dressler Wolfgang U. 1987. Word Formation as Part of Natural Morphology.
In Dressler, Mayerthaler, Panagl & Wurzel 1987. 99-126.
D ressler Wolfgang U. 1990. Sketching Submorphemes within Natural
Morphology. In Mndez Dosuna Julin & Carmen Pensado (eds.) 1990.
Naturalists at Krems. Salamanca: Ediciones Universidad de Salamanca.
33-41.
D ressler Wolfgang U., Willi M ayerthaler, Oswald Panagl & Wolfgang
U. Wurzel 1987. Leitmotifs in Natural Morphology. Amsterdam &
Philadelphia: John Benjamins Publishing Company.
Dressler Wolfgang U. & Katarzyna Dziubalska-Koaczyk 1994. Evidence of
syllabic writing against the phonological syllable. In Dressler Wolfgang
U., Martin Prinzhorn & John R. Rennison (eds.) 1994. Phonologica 1992.
Torino: Rosenberg & Sellier. 65-76.
Edgerton William F. 1952. On the Theory of Writing. Journal of Near Eastern
Studies 11.4. 287-290.
Elbro Carsten 2005. Literacy Acquisition in Danish: A Deep Orthography in
Cross-Linguistic Light. In Malatesha & Aaron 2006. 31-45.
Faber Alice 1992. Phonemic segmentation as epiphenomenon: Evidence from
the history of alphabetic writing. In Downing, Lima & Noonan 1992. 111134.
156

Foss Donald J., David A. Swinney 1973. On the Psychological Reality of the
Phoneme: Perception, Identification and Consciousness. Journal of
Verbal Learning and Verbal Behavior 2. 246-257.
Fournier P. F. 1940. Sur lorigine des complications de lorthographe franaise.
Le franais moderne 8. 257-266.
Gelb Ignace Jay 1963. A study of writing. Chicago: University of Chicago
Press (1st edition 1952).
Giscel 2007. Educazione linguistica democratica a trentanni dalle 10 Tesi.
Franco Angeli: Milano.
Gleason Henry Allan 1961. An Introduction to Descriptive Linguistics. New
York: Holt, Rinehart & Winston (1st edition 1955).
Gleitman Lila & Paul Rozin 1977. The structure of acquisition in reading. In
Reber S. Arthur & Don L. Scarborough (eds.) 1977. Towards a psychology
of reading, Hillsdale, New Jersey: Lawrence Erlbaum Associates. 1-141.
Goswami Usha 2005. Orthography, Phonology and Reading Development:
A Cross-Linguistic Perspective. In Malatesha & Aaron 2006. 463-480.
Gray William S. 1956. The Teaching of Reading and Writing (Monographs on
Fundamental Education, X). Paris: UNESCO.
Hakulinen Lauri 1961. The structure and development of the Finnish language. The Hague: Mouton.
Hoej Vladimr 1971. Formes parles, formes crites et systmes orthographiques des langues. Folia linguistica 5. 185-193.
Innis Harold 1991. The Bias of Communication. Toronto: Toronto University
Press.
Jensen John B. 1995. Writing Portuguese Electronically: Spontaneous Spelling
Reform. Hispania 78.4. 837-845.
Justeson John S. 1976. Universals of Language and Universals of Writing.
In Juilland Alphonse, A. M. Devine & Laurence D. Stephens (eds.) 1976.
Linguistic Studies offered to Joseph Greenberg 1. Saratoga: Anma Libri.
57-94.
Kilani-Schoch Marianne 1988. Introduction la morphologie naturelle. Berne,
Francfort-s. Main, New York & Paris: Lang.
L iberman Isabelle Y., Donald S hankweiler, William F. Fischer & Bonnie
Carter 1974. Explicit syllable and phoneme segmentation in the young
child. Journal of Experimental Child Psychology 18. 201-212.
Linell Per 1979. Psychological Reality in Phonology. A Theoretical Study.
Cambridge, London, New York & Melbourne: Cambridge University Press.
Logan Robert K. 2004. The Alphabet Effect: A Media Ecology Understanding
of the Making of Western Civilization. New York: Hampton.
M alatesha Joshi R. & P. G. A aron 2006. Handbook of Orthography and
Literacy. Oxford & New York: Routledge.
Martlew Margaret 1983. The development of writing: Communication and
Cognition. In Coulmas & Ehlich 1983. 257-276.
Masica Colin P. 1993. The Indo-Aryan languages. Cambridge: Cambridge
University Press.
Mayerthaler Willi 1981. Morphological Naturalness. Ann Arbor: Karoma.
McLaughlin John 1963. A Graphemic Phonemic Study of a Middle-English
Manuscript. The Hague: Mouton.
157
Antonio Baroni
McLuhan Marshall & Robert K. Logan 1977. Alphabet, Mother of Invention.
Etcetera 34. 373-383.
McLuhan Marshall 1964. Understanding Media: The Extensions of Man.
London: Latimer.
Migliorini Bruno 1994. Storia della lingua italiana. Milano: Bompiani.
Miller Gary D. 1990. Homer and Writing: Use and misuse of epigraphic and
linguistic evidence. Classical Journal 85. 171-179.
Miller Gary D. 1994. Ancient Scripts and Phonological Knowledge. Current
Issues in Linguistic Theory 116. Amsterdam: John Benjamins Publishing
Company.
Mioni Alberto 2009. Conoscenze, memorie e riti della scrittura e della parola
Continuando il viaggio di Giorgio R. Cardona. In Consani, Furassi,
Guazzelli & Perta 2009. 13-46.
Mitton Roger 1996. English Spelling and the Computer. London & New York:
Longman.
Morais Jos, Luz Cary, Jsus Alegria & Paul Bertelson 1979. Does awareness
of speech as a sequence of phonemes arise spontaneously? Cognition 7.4.
323-331.
Nichols John D. 1996. The Cree Syllabary. In Daniels & Bright 1996. 599-611.
OC onnor Michael 1996. The Berber Scripts. In D aniels & B right 1996.
112-119.
Ong Walter J. 1986. Oralit e scrittura. Le tecnologie della parola. Bologna: Il
Mulino (Orig. vers. London & New York: Methuen 1982).
Pasch Helma 2008. Competing scripts: The introduction of the Roman alphabet in Africa. International Journal of the Sociology of Language 191.
65-109.
Peirce Charles Sanders 1980. Semiotica. Torino: Einaudi (Orig. vers.
Cambridge: Harvard University Press 1931-1935).
Perfetti Charles A. & Sulan Zhang 1991. Phonological processes in reading Chinese characters. Journal of Experimental Psychology: Learning,
Memory and Cognition 17.4. 633-643.
Perfetti Charles A. & Sulan Zhang 1995. Very early phonological activation in
Chinese reading. Journal of Experimental Psychology: Learning, Memory
and Cognition 21.1. 24-33.
Pontecorvo Clotilde 1994. Iconicity in Childrens First Written Texts. In
Simone Raffaele (ed.) 1994. Current Issues in Linguistic Theory 110.
Amsterdam & Philadelphia: John Benjamins Publishing Company. 277307.
Primus Beatrice 2004. A Featural Analysis of the Modern Roman Alphabet.
Written Language & Literacy 7.2. 235-274.
Read Charles 1986. Childrens Creative Spelling. London & Boston: Routledge
& Kegan Paul.
Sampson Geoffrey 1985. Writing systems. London: Hutchinson.
Sebba Mark 2006. Ideology and alphabets in the former USSR. Language
Problems and Language Planning 30.2. 99-125.
Seymour Philip H. K., Mikko Aro & Jane M. Erskine 2003. Foundation literacy acquisition in European languages. British Journal of Psychology
94. 143-174.
158

S uleymanova Dilyara 2010. International Language Rights Norms in
the Dispute over Latinization Reform in the Republic of Tatarstan.
Caucasian Review of International Affairs 4.1. 43-56.
Tanaka Kenji 2003. Columns for complex visual object features in the inferotemporal cortex: Clustering of cells with similar but slightly different
stimulus selectivities. Cerebral Cortex 13.1. 90-99.
Thornton Anna M. 2007. Morfologia. Roma: Carocci.
Vachek Josef 1939. Zum Problem der geschriebenen Sprache. Travaux du
Circle Linguistique de Prague 8. 94-104.
Vachek Josef 1973. Written language. General problems and problems of
English. The Hague: Mouton.
Venezky Richard L. 1970. The Structure of English Orthography. The Hague &
Paris: Mouton.
Wurzel Wolfgang U. 1994. Natural Morphology. In E. Asher Ronald & John M.
Y. Simpson (eds.) 1994. The Encyclopaedia of Language and Linguistics.
Oxford: Pergamon Press. 2590-98.
159

Alphabetic Vs Non-Alphabetic Writing - Baroni

Uploaded by

Copyright:

Available Formats

Alphabetic Vs Non-Alphabetic Writing - Baroni

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Alphabetic Vs Non-Alphabetic Writing - Baroni

Uploaded by

Copyright:

Available Formats

Alphabetic vs.

Symbols and abbreviations

Rivista di Linguistica 23.2 (2011), pp. 127-159

(received September 2011)

By writing, we mean a series of graphic symbols arranged on a

Alphabetic vs. non-alphabetic writing

as soon as everyone in the Peoples Republic of China share the same

2. Writing and language perception

Alphabetic vs. non-alphabetic writing

segmentation ability as a human skill may have been a direct result

Several studies (Foss & Swinney 1973, Liberman et al. 1974,

The syllable is a more salient perceptive unit so an ideal script,

Alphabetic vs. non-alphabetic writing

3.1. Maximum distinctiveness

3.3. Cognitive salience

The results indicate that a character is preferably composed of

Alphabetic vs. non-alphabetic writing

Alphabetic vs. non-alphabetic writing

So far, it seems unavoidable to consider other elements to define

graphic syllable and the phonic syllable, inasmuch as words composed

Alphabetic vs. non-alphabetic writing

efficiently only insofar as he treats the written language as if it were

Alphabetic vs. non-alphabetic writing

Concerning morphological naturalness, writing has the primacy

radical reduction processes, which make the language very

Alphabetic vs. non-alphabetic writing

Finnish is spoken by 6 million people in Finland and Sweden,

Alphabetic vs. non-alphabetic writing

and a native speaker generally knows how to write or how to read a

a stress-timed language but it is quite regular in the graphic

Alphabetic vs. non-alphabetic writing

The spontaneous deviations from the norm operated by language

Alphabetic vs. non-alphabetic writing

tered appearance. This tendency does not appear surprising if

Alphabetic vs. non-alphabetic writing

who went to America to study medicine (he was therefore deeply

the problems of writing. To do so, we have tried to abandon the usual

Alphabetic vs. non-alphabetic writing

taining a certain degree of morphological naturalness in languages

Alphabetic vs. non-alphabetic writing

Alphabetic vs. non-alphabetic writing

Alphabetic vs. non-alphabetic writing

You might also like