Ma’alahi: Use of a Simplified Language to Test a
Linguistic Hypothesis
Auteur: Jeffrey R. Brun
Date MS: 12-09-2014
Date FL: 01-01-2015
Numéro FL: FL-000028-00
Citation: Brun, Jeffrey R. 2014. “Ma’alahi: Use of a
Simplified Language to Test a Linguistic
Hypothesis.” FL-000028-00, Fiat Lingua,
droits d'auteur: © 2014 Jeffrey R. Brun. Ce travail est sous licence
sous une attribution Creative Commons-
Non Commercial-Pas de Dérivés 3.0 Licence non transférée.
http://creativecommons.org/licenses/by-nc-nd/3.0/
Fiat Lingua est produit et maintenu par la Language Creation Society (LCS). Pour plus d'informations
à propos du LCS, visitez http://www.conlang.org/
Maʻalahi:
Use of a simplified language to test a linguistic hypothesis
by Jeffrey R. Brun
Abstrait
Maʻalahi is a constructed language derived from a single source language, Hawaiian, with a
ruthlessly simplified Polynesian grammar. This makes it an appropriate candidate for
investigating hypotheses about the ease of L2 language acquisition. An exploratory study was
performed to determine whether grammatical features or external factors (social or personal) sont
more significantly correlated with perceived ease of learning and correct performance on
translation tasks. Only external factors were shown to be significantly correlated.
(Please refer to glossary for unfamiliar terms.)
Hypothesis
The hypothesis of the study was that the ease of learning a language, whether a natural or a
constructed one, is not significantly related to the simplicity or regularity of its grammar; plutôt,
it is related to three other factors:
Cognacy: The similarity of the lexicon, and to a lesser extent, the grammar, to a language
already spoken by the learner;
Polyglottism: The number of different languages, especially from diverse language
families, spoken by the learner; et
Metalinguistic Awareness: The degree of linguistic knowledge of the learner, et le
ability to conceptualize languages abstractly.
To test this hypothesis, a study was done in which the participants were asked to learn a
grammatically simple conlang, Maʻalahi, which was based on Hawaiian – a language whose
grammar and lexicon differ markedly from any European language, and which would likely be
unfamiliar to the participants. After studying the language, the participants were asked to
perform translation exercises from Maʻalahi to English and from English to Maʻalahi. Ease of
learning was evaluated by scoring the correctness of the translation exercises, and by asking the
participants to report on how easy they felt the language was.
Maʻalahi
1
Other questions in the questionnaire were used to assess cognacy (familiarity with any
Polynesian language); polyglottism (number of languages spoken and reported competency in
eux, ainsi que: [a.] number of language families, et [b.] number of auxlangs spoken); et
metalinguistic awareness (as based on: [a.] formal education in linguistics, [b.] the number of
auxlangs constructed by the participant, et [c.] the number of artlangs constructed by the
participant).
The Background of the Study
For a long time, I have felt that Hawaiian was the most beautiful natural language in the world. je
have regretted occasionally that Mozart hadn’t written any operas in Hawaiian, as lovely as
Italian opera is. Admittedly this is a subjective, aesthetic opinion.
In January 2014, David Peterson, Sylvia Sotomayor, and I were invited to be on a panel of
conlangers for the student linguistics club at University of California – San Diego. One of the
questions posed was this: “If you had to choose one single natural language for the entire world,
what would it be?” Both David and I answered “Hawaiian” immediately. (Sylvia, I believe, était
somewhat dumbfounded by the alacrity of our replies.)
I decided the next day to create a conlang with the same sound (phonology and phonotactics) comme
Hawaiian.1 I agonized over what specific features of Hawaiian phonology gave it its particular
phonoaesthetics. Alors, I had one of those rare “aha!” moments of obviousness: the language that
would sound the most like Hawaiian, was Hawaiian.
The problem with learning Hawaiian, mais, is that its grammar is quite complex: deux
categories of possessives (volitional and non-volitional), close to three dozen pronouns (if one
counts all the possessive forms), complex syntax for relative clauses, odd word order shifts for
topicalization, and inscrutable rules (still debated by linguists) governing the ellipsis of markings
for nouns and verbs. It’s a hard language to learn.
I decided to create a conlang called “Maʻalahi,” which means “simple” in Hawaiian. I retained
the Hawaiian phonology (with one change: letting /k/ and /t/ function as separate phonemes,
rather than allophones in complementary distribution), as well as the entire Hawaiian lexicon,
thus preserving the exact sound of the language. I then drastically simplified the grammar,
removing all but the most common forms, coalescing types wherever possible, and regularizing
the syntax and the rules concerning markings. The result was a grammar no more complex than
Esperanto’s, et qui, I hoped, would be as easy to learn.
1 I should point out that David has already done this in his conlang, Kamakawi.
Maʻalahi
2
Around this same time, there was one of the recurrent arguments taking place on the Brown
University Auxlang list: What factors led to the adoption of an auxlang? One contingent pointed
to sociological factors, whether political, economic or socio-prestigious. The other contingent
emphasized the ease of learning the auxlang. This branched off (as usual) into a secondary
argument as to which inherent attributes of an auxlang contributed primarily to its ease of
apprentissage: Was it the simplicity and regularity of its grammar? Or the recognizability of the words
of the lexicon, because they were cognate to words in a language already spoken by the learner?
That question had never been studied quantitatively. After a thorough search through the
littérature, and wading through unsubstantiated claims made by devotees promoting one auxlang
or another, I found only these few published articles [see Bibliography for complete citations]
peripherally related to this question:
Folia et al, Artificial Language Learning in Adults and Children
Lindstedt, Native Esperanto as a Test Case for Natural Language
Montúfar, The Contemporary Esperanto Community
Roehr-Bracklin & Tellier, Enhancing Children’s Language Learning: Esperanto as a
Tool
None of them addressed directly the issue of cognacy (the degree to which the words in the
auxlang’s lexicon resemble words in a natlang spoken by the learner)2. Toutefois, one intriguing
discovery was that Esperanto could function as a tool to increase children’s metalinguistic
awareness, thereby accelerating their acquisition of foreign languages.3
I felt that it would be a boon to auxlangers to know which inherent linguistic factors led to the
ease of learning, and hence likelihood of adoption, of an auxlang (regardless of the extent to
which external factors are critical to that adoption). Most of the auxlangs in circulation today,
comme l'espéranto, Ido, Interlingua, Sambahsa, etc., are based on European source languages, et
their speech communities are drawn heavily from people whose natal and/or second languages
are also European. This makes it difficult to disentangle which of the two factors, simplicity or
cognacy, is primarily contributive to the ease of learning.
My first thought about testing this was to consider the feasibility of evaluating the ease of
learning of Esperanto among the Quechua who live in the highlands of Peru and Bolivia (where I
2 Or, aussi, grammatical cognacy: The degree to which grammatical forms resemble those of a natlang spoken by the
learner.
3 As an example, British middle-school children who studied one semester of Esperanto followed by three semesters
of French, learned more French than those students who studied four semesters of French without exposure to
Esperanto. This may indicate that the purposeful construction of simplified languages can be a useful pedagogical
tool to enhance second-language acquisition.
Maʻalahi
3
had visited), as there is no commonality between Quechuan and the source languages of
Esperanto. The logistics of, and resources for, such a study, cependant, would be prohibitive.
En plus, most of the Quechua speak Spanish as a second language, which might invalidate
any test results.
My second idea was to test the ease of learning of an auxlang that was derived exclusively from
one or more non-European source languages, among speakers of European languages. C'était
then that I realized that I had already constructed such a test vehicle. It was Maʻalahi, a language
with a simplified and regularized grammar whose lexicon (and grammar) is derived from a single
source language: Hawaiian (which is definitely not Indo-European). Only those familiar with
Hawaiian, or some other Polynesian language, would be likely to recognize any words or
grammatical forms.
This was origin of the study which is presented in this paper.
Status of the Hawaiian language
The readers of this paper may find it helpful to know a little about the Hawaiian language and its
status today. Hawaiian is a language of the “Eastern Polynesian” family. Selon
Wikipedia:
The Polynesian languages are a language family spoken in geographical Polynesia as well as
on a patchwork of “Outliers” from south central Micronesia, to small islands off the northeast
of the larger islands of the Southeast Solomon Islands and sprinkled through Vanuatu. Ils
are classified as part of the Austronesian family, belonging to the Oceanic branch of that
famille. Polynesians share many unique cultural traits which resulted from about 1,000 années
of common development, including common linguistic development, in the Tonga and
Sāmoa area through most of the first millennium BC.
There are approximately forty Polynesian languages. The most prominent of these are
Tahitian, Sāmoan, Tongan, Māori and Hawaiʻian. Because the Polynesian islands were
settled relatively recently and because internal linguistic diversification only began around
2,000 il y a des années, their languages retain strong commonalities.
One of the difficulties of creating a simplified Hawaiian, such as Maʻalahi, is that the Hawaiian
langue (ʻŌlelo Hawaiʻi) does not present a single target. Synchronically (c'est, the current
jour) there are three different forms of spoken Hawaiian:
1. Those who have learned Hawaiian as children and use it as their primary language (L1).
The only speech community where Hawaiian is used as the primary language is on the
Maʻalahi
4
island of Niʻihau and a few areas on Kauaʻi. There are between 100 and 200 speakers.
The pronunciation and accent of the Niʻihau speakers differ significantly from speakers
on the other islands.
2. Those who have learned Hawaiian as children but use English as their primary language.
This community consists of mostly of elderly Hawaiians, many of whom live in the rural
parts of the islands. They speak Hawaiian mostly with their elderly friends or with their
relatives. The size of this community is rapidly shrinking, and today numbers no more
than 1,000 speakers.
3. Those who have learned Hawaiian as adults as a second language (L2), but use English as
their primary language.
Starting in the 1970s, as part of the activism movement in Hawaiian heritage and culture,
schools were started to teach Hawaiian as a second language. A large number of students
have taken language courses, often at the college level (par exemple, at the University of
Hawaiʻi – Hilo). There may be as many as 10,000 individuals who have studied Hawaiian
as a second language, but their level of competency and subsequent frequency of usage
varies greatly. This language is sometimes referred to as “Neo-Hawaiian” to distinguish it
from the traditional language. Compared to traditional Hawaiian, Neo-Hawaiian is
grammatically simpler and more regular, and the intonation is more similar to American
English than to a Polynesian language.
In addition to those three categories, there are also four distinct diachronic (historique) périodes
during which the language was studied or examples of its usage were recorded in writing:
1. Before 1825, some early explorers documented the language. Even in this earliest period,
there is evidence of loan words from English, Portuguese and Chinese.
2. From 1825 through the end of the 19th century was a period during which Christian
missionaries documented the language, created the alphabet that is still in use today,
promoted Hawaiian literacy, encouraged the printing of newspapers in Hawaiian, et
transcribed many samples of the language.
3. After the American annexation of Hawaii in 1893, usage of Hawaiian dropped off, partly
because it was made illegal to teach it in any school (although it was still used privately at
maison). English was taught to all native Hawaiian children and became the primary language
of the territory.
4. Starting in the late 1950s, the prohibition against teaching Hawaiian was dropped, et le
first schools were opened to teach the language. Progressivement, more immersion schools have
Maʻalahi
5
opened, college courses have been offered, and more students have learned the language,
some of whom had never been exposed to it until entering university.
There are changes in the documented language during these periods. The “modern period” (#4)
has seen the rise of Neo-Hawaiian (Hawaiian as an L2) whose grammar is simpler and more
regular than the prior time periods. [NeSmith] The documentation from the earliest period (#1) est
scant. Some of those early transcriptions appear to be inconsistent with the known rules of
Hawaiian grammar, as documented in the latter part of the 19th century. Could it have been a
transcription error? Could the storyteller who was being recorded have made an ungrammatical
énonciation? Or was traditional Hawaiian more irregular than how it was spoken at the end of the
19th century? It is impossible for anyone alive to answer those questions, because there is no
standard to use as a guide.
All of this feeds into the difficulty of creating a simplified Hawaiian such as Maʻalahi. Lequel
time period and which version of the language should be used as a starting point? The greatest
number of speakers today consist of L2 speakers. Should Neo-Hawaiian be taken as the starting
point for constructing a simplified language? Or would it be more appropriate to use the
traditional language as it was documented during the early 19th century? As is true for all
natlangs, there is no such thing as a “pure” Hawaiian language; even the earliest documented
transcriptions show the inevitable incursion of foreign words.
What I did for the creation of Maʻalahi was to lay out, in tabular form, all the alternants, and then
do my best to arrive at a consistent scheme for regularization.
Description of Maʻalahi
A complete description of the Maʻalahi language is available on the web. Rather than repeat that
information in appendices of this document, links are provided below.
A description of the grammar, including numerous sample sentences:
http://www.temenia.org/Ma_alahi/docs/Grammar.pdf
(Note that this description of the grammar is somewhat different than what was shown to the
participants in the study. The grammar document now includes a section on the differences
between Maʻalahi and Hawaiian, and also some of the explanations and examples have been
augmented for clarity.)
A short lexicon:
http://www.temenia.org/Ma_alahi/docs/Lexicon.pdf
(Noter que, except for the “grammar words,” Maʻalahi’s lexicon is identical to Hawaiian. Le
linked document is an abridged Hawaiian dictionary of about 500 words.)
Maʻalahi
6
A reference aid listing the “grammar words” (and the section of the grammar description
where they are explained) as an aid for translation:
http://www.temenia.org/Ma_alahi/docs/ReferenceAid.pdf
Simplicity of Grammar: Comparison to Esperanto
In order to test the hypothesis that it is not simplicity of the grammar, pour lui-même, that makes an
auxlang easy to learn, it is important to quantify, to some extent, the degree of difficulty of the
grammar of a language. This turns out to be an extremely thorny question to answer, perhaps an
intractable one.
Pour une chose, it is hard to assign numerical measures to grammatical attributes. Par exemple:
Are inflectional grammars always more difficult than agglutinating or isolating ones? To what
étendue? If the inflections are regular, but the equivalent ways to express those elements in an
isolating language are irregular with many exceptions, then how does one account for that?
One possibility is to fully specify the grammar with formal transformational rules. In my
expérience, language learners do not conceptualize the grammars of languages in that way, mais
rather use much less formal rules, or rules-of-thumb, which allow for exceptions.4 Formal
syntactic rules, de plus, do not take into account rules bound to the semantic meanings of
mots, such as gender (Is Susan a “he” or a “she”?), or animacy (Is the blowing of the wind
volitional or not?), all of which increase the difficult of attaining competency in a language.
Donc, I do not believe that it is possible to accurately quantify the difficulty of a grammar in
a way that is relevant to the question of how easy it is for a learner to internalize the rules.
Néanmoins, it is possible, à mon avis, to make crude comparisons. Virtually everyone
would agree that the grammar of modern standard Arabic is more complex than the grammar of
Esperanto. If nothing else, the fact that Haywood & Nahmad’s A New Arabic Grammar is over
450 pages in length (with another 150 pages for index, appendices, glossary, and supplements)
should serve as a demonstration of the greater complexity of Arabic grammar.
As part of this study, I wished to quantify, however roughly, the complexity of Maʻalahi’s
grammar relative to Esperanto’s. Many proponents of Esperanto point out that its grammar is so
simple that it can be described in only sixteen rules (and thereby “proving” to their satisfaction
its ease of learning) – see Don Harlow’s webpage (http://donh.best.vwh.net/Esperanto/rules.html)
for an example. Following that model, I created a short, preliminary table comparing the
grammars of the two languages:
4 As in the well-known English-spelling mnemonic: “I before E / except after C / or when sounded like A / as in
neighbor or weigh.”
Maʻalahi
7
Maʻalahi
Esperanto
Pages to describe grammar
Number of rules
14
25
Number of Letters/Phonemes 19
Remarques:
13
20
28
1. The page count for Maʻalahi excludes the sections on the differences from Hawaiian.
2. Differing text formats makes page count a crude comparison indeed.
3. The “Sixteen Rules of Esperanto Grammar” contain four implicit rules which are not
stated (tel que: word order is SVO) – perhaps because these seem self-evident to an
audience consisting primarily of Europhones. These four implicit rules are included in the
above total.
From this very short comparison we see that the descriptions of the languages are approximately
of the same length, with about the same number of grammatical rules and of phonemes. Pursuing
this tack, I made a more detailed table, counting the number of grammatical features of each
langue:
Maʻalahi Esperanto Notes
Verbal Types
Verbal Forms
Nominal Forms
Adjectival Forms
Adverbial Forms
3
6
1
0
0
Pronoms
14
Prepositional Cases
Types of Negation
Directionals
Demonstratives
interrogatifs
Comparatives
2
3
4
6
8
0
2
12
4
4
1
9
1
3
0
4
9
3
Esperanto: transitif & intransitif
Maʻalahi: transitif, intransitif & stative
12 forms in Esperanto (rich TAM structure
common to Euro-langs)
Esperanto: singular/plural &
nominative/accusative
No adjectives in Maʻalahi – Esperanto:
singular/plural, nominative/accusative
No adverbs in Maʻalahi
14 forms in Maʻalahi (because of dual number
and exclusive/inclusive 1st person, common to
Polynesian-langs)
No directionals in Esperanto (discounting use of
accusative to indicate destination)
Maʻalahi uses stative verbs. Esperanto:
standard/comparative/superlative
Maʻalahi
8
Maʻalahi Esperanto Notes
Ellipsis
2
0
TOTAL
49
52
Subject and direct object markers of common
nouns can be nulled in Maʻalahi
These totals are crude. All they indicate is that
the complexity of the two grammars are roughly
égal. A natlang such as English or Russian
would have a total of over 200.
Admittedly, the comparison shown in the above table is crude, and open to interpretation and
critique. My intent is not to prove that the grammar of Maʻalahi is simpler than Esperanto, mais
merely that Maʻalahi’s grammar, which was regularized and simplified based on auxlang design
principals, is approximately as simple an Esperanto’s or any other major auxlang, such as Ido,
Interlingua, or Sambahsa.
It is important to add, that although Maʻalahi’s grammar may be simple based on crude objective
measures as shown in the above table, it is very different from the SAE grammars that so many
of today’s auxlangs possess. It is this combination of properties: simple grammar, yet with little
similarity to SAE languages, that makes Maʻalahi a good test vehicle to investigate the question
as to whether it is the simplicity of the grammar, or the cognacy of the lexicon and grammatical
caractéristiques, that makes an auxlang easy to learn.
Description of the Study
A letter inviting participation in the study was emailed to the LCS Members listserver, and to the
Conlang and Auxlang listservers maintained by Brown University. A similar letter was posted on
the Facebook groups: Conlangs, Auxlangs, Polyglots, and Omniglot. The main part of that letter
is reproduced below:
Would you be willing to participate in a sociolinguistic study to test theories about the
ease of learning of auxiliary languages? (Please consider this, even if you have never
learned nor constructed an auxiliary language.)
If you volunteer, you will be sent:
(1) a short, multipart questionnaire;
(2) a concise 14-page description of an auxlang’s grammar (written for linguists
or conlangers, not as a primer);
(3) a 500-word lexicon;
(4) a translation exercise.
You will be asked to study the grammar and lexicon of the language, to complete the
translation exercise and the questionnaire, and to return your answers to me.
Maʻalahi
9
The questionnaire consisted of four parts:
1. Preliminary questions about the languages spoken by the participant, the level of formal
linguistics education, and any languages constructed by the participant;
2. Instructions on reading/studying the description of Maʻalahi (phonology, orthography,
grammaire, examples);
3. The translation exercises;
4. Final questions about the participant’s opinion concerning the difficulty of Maʻalahi and
the translation exercises.
62 individuals responded and requested the materials to participate in the study. Of those, 28
completed the questionnaire and translation exercises and returned them to me. Of the others, 11
notified me that they had decided to quit the study without completing it, and 23 never responded
to my follow-up emails.
Of the 28 participants:
Two were or had been professional linguists; eleven others had studied linguistics at
university
Four were bilingual from childhood
All participants spoke at least two natlangs, with two participants speaking a total of
eight apiece; the average was slightly over four
Fourteen spoke at least one auxlang with some degree of fluency
Eighteen had constructed at least one artlang (although sketchlangs were included in
this count)
For the purposes of the study, the participants self-evaluated their competency level in the
natlangs and auxlangs that they spoke. In terms of the languages constructed by the participants,
it was clear from the responses in the questionnaire that some of the listed languages were
significantly developed while others were mere sketches. I was not sure that the raw number of
artlangs constructed by each participant would result in meaningful comparisons. Donc, je
created a scale summarizing the “skill level” of the artlangs listed by the participants, using the
following rubric:
0 = no artlangs constructed
1 = only sketchlangs
2 = at least one significantly developed artlang
3 = at least one significantly developed artlang plus other incomplete artlangs
4 = at least two significantly developed artlangs
5 = more than two significantly developed artlangs
This “skill level of artlangs” was used as a variable in the statistical analysis, as well as the raw
count of artlangs constructed by the participant.5
5 It is acknowledged that the assessment of this skill level is a subjective judgment by me.
Maʻalahi
dix
There were two sets of translation exercises: a set of five sentences to translate from Maʻalahi to
Anglais, and a second set of five translations from English to Maʻalahi. Each set was arranged so
that the first exercise was easy and the last one was difficult, with an increasing level of
difficulty for the intermediate exercises.
Most of the participants completed the entire questionnaire and, to the best of their ability, tous les
translation exercises. A few of them, due to noted lack of time or interest, completed only a
portion of the translation exercises.
Each translation was scored as follows:
3 = correct, or only minor errors
2 = “medium” errors
1 = major errors
0 = not attempted
A perfect score for all ten translation exercises was 30. Four of 28 participants (14%) achieved
perfect scores. Only six participants (21%) scored 20 or less. The average score was 23.7.
The data was analyzed using Epi Info 7, a free statistical analysis package developed for
epidemiologists.
Quantitative Analysis
The results of the study are presented in relationship to the hypothesis, reiterated below:
The ease of learning a language is not significantly related to the simplicity or regularity
of its grammar; plutôt, it is related to three factors:
Cognacy: The similarity of the lexicon, and to a lesser extent, the grammar, to a
language already spoken by the learner;
Polyglottism: The number of different languages, especially from diverse language
families, spoken by the learner; et
Metalinguistic Awareness: The degree of linguistic knowledge of the learner and the
ability to conceptualize languages abstractly.
To evaluate the “ease of learning” two factors were used. One was the participant’s reported ease
of learning after having studied the Maʻalahi language and completed the translation exercises.
As this was a subjective factor, the correctness of the translation exercises was also used as a
proxy for the ease of learning.
Maʻalahi
11
These two factors correlated closely. [Correlation coefficient r = 0.51. Spearman’s p = 0.0066.]
Donc, throughout the remainder of the analysis, the correctness of the translation exercises
was used to evaluate the factors related to the ease of language learning.6
The supposition that the simplicity of a language’s grammar accounts, dans une large mesure, for ease
of learning, was not supported by the study. Comme il est bien connu, most learners of Esperanto (et
similar European-language-based auxlangs) report that it is easy to learn. As discussed above,
the grammar of Maʻalahi is roughly as simple as Esperanto’s. Donc, if the simplicity of
grammar was the overriding factor to determine ease of learning, the expectation would be that a
significant proportion of the participants would report that learning Maʻalahi was similarly easy.
Toutefois, from inspection of Chart 1, it can be seen that the reported ease of learning varied
greatly. For a constructed language with a simple grammar, one would expect a much narrower
distribution with a mean close to the left-hand side of the chart. Considering that the participants
in the study have a much higher degree of competency and interest in languages and linguistics
than the general population, this shows that grammatical simplicity or regularity for these
participants were minor factors in the ease of language learning.
Chart 1
t
est
tu
o
C
Ease of Learning (1 = “very easy” to 5 = “very difficult”)
6 Throughout the study, Spearman’s correlation analysis was used instead of the more common Pearson’s analysis,
since Spearman’s analysis is more tolerant of non-normal data, such as found in this study.
Maʻalahi
12
To investigate cognacy, the participants were split into three groups:
(1) Those who spoke any Polynesian language
(2) Those for whom none of their native languages was a European language
(3) Everyone else
This analysis could not be performed. Only one participant did not list a European language as
one of his/her native languages. No participants had more than a minimal familiarity with any
Polynesian language. Although seven (25%) of the participants had a slight degree of exposure
to a Polynesian language, either in their studies or during travel to the area, this casual exposure
was insufficient to distinguish them from the other participants.
Significant correlation was found between the ease of learning of Maʻalahi and polyglottism.
Analysis of the number of childhood native languages was not fruitful, because no participant
listed more than two. The other factors (the number of non-native natlangs spoken, the number
of families to which those natlangs belong, and the reported competency level in those natlangs),
were all shown to correlate with the reported ease of learning and the correctness of the
translation exercises. Charts 2 and 3 show examples of these relationships.
Chart 2
s
est
o
J’ai
t
l
À
s
est
À
le
J
f
o
s
s
e
est
t
c
e
le
le
o
C
Non-Native Natlang Families
Correlation Coefficient: r = 0.39
Spearman’s Correlation: p = 0.0419
Maʻalahi
13
Chart 3
s
est
o
J’ai
t
l
À
s
est
À
le
J
f
o
s
s
e
est
t
c
e
le
le
o
C
Total Natlang Competency Level
Correlation Coefficient: r = 0.37
Spearman’s Correlation: p = 0.1067
Metalinguistic awareness, as defined in this study, was not found to be related to the ease of
learning of the Maʻalahi language. Some of the factors used to assess metalinguistic awareness
did not correlate. Par exemple, no statistically significant relationship was found between the
level of formal linguistics study and reported ease of language learning.
Toutefois, one factor, en particulier, did correlate highly. This was the number of artlangs
constructed by the participant. The number of artlangs, and the assessed skill level (the degree of
completeness to which those artlangs were developed), were predictive of the correctness of
translations and the reported ease of learning in this study. Charts 4 and 5 show, respectivement, le
relationships for the number of constructed artlangs, and the skill level of same.
Maʻalahi
14
Chart 4
Number of Artlangs Constructed
Correlation Coefficient: r = 0.40
Spearman’s Correlation: p = 0.0130
Chart 5
s
est
o
J’ai
t
l
À
s
est
À
le
J
f
o
s
s
e
est
t
c
e
le
le
o
C
s
est
o
J’ai
t
l
À
s
est
À
le
J
f
o
s
s
e
est
t
c
e
le
le
o
C
Skill Level of Constructed Artlangs
Correlation Coefficient: r = 0.42
Spearman’s Correlation: p = 0.0073
Maʻalahi
15
De plus, if the number of constructed artlangs is converted to a binomial variable (no artlangs
constructed versus at least one), the correlation is even stronger. [Correlation coefficient r = 0.47.
Spearman’s p = 0.0019.] Clairement, the act of constructing even a single artlang is related to one’s
ability to learn Maʻalahi, and probably any language (although whether this is a causal
relationship is undetermined).
Donc, within the limitations of this study, as discussed below, it is concluded that simplicity
or regularity of the grammar is not significantly related to the ease of learning that language, mais
rather that the major factors are the level of the learner’s polyglottism and the extent to which
he/she has been involved in language construction.
Qualitative Analysis
At the end of the questionnaire, the participants were invited to offer additional comments. 24 of
28 chose to do so. From analyzing these responses, some conclusions can be drawn:
1. Twelve of the participants volunteered that they felt that the materials (the grammar
description and the lexicon) were inadequate for learning the language. Il y avait
suggestions to:
Incorporate more examples
Include interlinear glosses
•
•
• Merge the lexicon and the grammar reference aid
• Explain in more detail unfamiliar concepts, such as directionals or modal verbs or
passive intransitives
• Add a series of progressive lessons
Even though the letter soliciting volunteers stated that the material was “written for linguists
or conlangers,” and the instructions said that a primer was not available, it seems that for half
of the participants the description of the grammar was overly terse. From other comments, il
appeared that some of the participants, rather than studying the grammar, merely skimmed it
and then turned directly to the translation exercises, during which they paged through the
grammar using it as a reference. This was true even for participants who indicated that they
had studied linguistics at the university level. Évidemment, this made completing the
translation exercises more time consuming, although it didn’t decrease the correctness of the
traductions.
The take-away from this is that creators of future studies which ask volunteers to learn a test
language should consider preparing progressive lessons similar to what is found in a
Maʻalahi
16
language textbook, even if the population from which the participants are drawn consists
mostly of conlangers or polyglots.
2. As is common among the conlang and auxlang community, many participants freely offered
helpful, constructive criticism of Maʻalahi, and how it ought to have been designed. Le
most prevalent comment was that the use of the ʻokina as a letter, and of the macron as a
diacritical mark indicating a long vowel, was “problematic” and made lexical lookup
difficile. I doubt that these same individuals have similar problems with the circumflex or
breve as used in Esperanto, or for that matter, the umlaut of German. This raises the question
about whether their perceived difficulty lends additional weight to my hypothesis that it is
familiarity with a language one already knows that contributes most strongly to the ease of
learning of a new language (although I had not initially considered that this hypothesis would
extend to a language’s orthography).
3. Other critical comments focused on the phonology, specifically the small size of the phonetic
inventaire, which leads to a large number of homonyms, and words that differ only by a short
versus a long vowel, or the presence versus the absence of an ʻokina, (which are distinctive in
Hawaiian). I wonder if the phonetic inventory had been larger than is common among
European languages, similar comments would have been forthcoming complaining that the
large number of phonemes made the language unnecessarily difficult.
4. Of the 28 participants, seven of them (25%) recognized Maʻalahi as a Polynesian language
(identified either as Hawaiian, or some suggested it was based on Tahitian or Niuean). None
of these participants, malheureusement, were fluent or competent speakers of any Polynesian
langue, which would have led to additional useful results for this study.
5. et enfin, six of the participants volunteered that participating in the study was fun,
whereas three participants said that it was not fun.
Discussion of Results
Before discussing the results of the present study and their implications, it must be noted that a
limitation is the nature of the sample. It is not a representative sample of the general population
whatsoever. It is a “convenience sample” drawn from a subpopulation of language enthusiasts.
De plus, all the participants were fluent in English, and most knew at least one other European
langue. Half of the participants spoke an auxlang with a reasonable degree of competency.
Many had studied Latin and/or classical Greek. A more extensive and thorough study would
include participants selected from diverse regions of the world, including those whose spoke no
European language, and who possessed varying degrees of linguistic interest.
Maʻalahi
17
An open question is whether the results of this small study can be extended to the general
population. I believe the answer is yes. Considering that the sample of participants was drawn
from among language enthusiasts, the results could reasonably be even more extreme for the
general population. This would indicate, donc, that the focus of most auxlangers, sur le
development of ever simpler and more regular grammars, may be misplaced. Most auxlangs
today are based on European languages (leading to the somewhat pejorative term: “euroclone”).
The purported ease of language learning and usage claimed by many auxlangers may be
attributable to nothing more than that the learners of those auxlangs natively speak a European
language and are benefitting from the cognacy of the auxlang to their own native language, ou
possibly to some other European language in which they are fluent.
The present study implies that the learning of a European-language-based auxlang would be
much more difficult by a monolingual speaker whose native language was Quechuan, ou
Hawaiian, or any non-European language. Those who attempt to construct a “worldlang” (un
auxlang whose target is everyone throughout the world) may be more successful if they
downsize their ambitions to working on a regional auxlang, such as Afrihili [Annis] ce qui est
aimed at a target of sub-Saharan African speakers.
The results of this small study (and someday, hopefully, of a more extensive one), should be of
significant interest to those who are involved in the construction of auxlangs. The discord and
schisms between aficionados of one auxlang or another have perplexed outsiders over the
seemingly minuscule grammatical or orthographic differences that are the source of the
arguments. What this study shows, mais, is that these tiny differences are irrelevant to the ease
of learning of the auxlang. Since any auxlang which is deemed to be successful must be easy to
learn so that its usage may expand, it would be advisable for auxlangers to desist from these
disputes and turn their focus to the experimental investigation and objective determination of the
factors which are more likely to lead to the dissemination of their languages.
This study implies, considering the vast grammatical, lexical and phonetic differences that exist
between the languages of the world, that the adoption of a true international auxiliary language
cannot be driven by features or attributes inherent to the language, but rather by external
economic, politique, or religious factors. The 19th century dream of a true international auxiliary
langue, untainted by the history of conquest and domination [Écologique], cannot be achieved unless
its propagation is forced, or strongly encouraged, by such external factors.
The Use of Conlangs for Experimental Linguistics
Most conlangers are familiar with the two main categories of constructed languages: artlangs (À
conlang created for an artistic purpose, such as literature, film, or for deepening the reality of a
science-fiction world), and auxlangs (a conlang developed to be spoken as an auxiliary language
Maʻalahi
18
by a region or by the entire world). There are also engelangs (engineered languages developed to
investigate the capabilities and limitations of human cognition or linguistic processing).
A subset of engelangs could be denoted as a “testlang," c'est, a conlang developed primarily for
the purpose of testing a particular linguistic hypothesis. Ainsi, its proper domain is confined to
the research laboratory; there is never an intent to promote it as a candidate IAL nor to have it
spoken by characters in a fantasy novel.
The most well-known testlang is Loglan/Lojban. Originally it was developed by James Cooke
Brown to test the Sapir-Whorf hypothesis. To my knowledge, no such experiment of Sapir-
Whorf has ever been performed, perhaps due to the schism between early speakers of the
langue. That failure is no reason, cependant, to desist from creating testlangs whose purpose is
to probe the intricacies of neurolinguistic processing, or address any other linguistic hypotheses.7
The original intent behind the construction of Maʻalahi was not as a testlang. Plutôt, I created it
merely because I wanted a simple language that sounded as much as possible as Hawaiian.
Toutefois, as I was working on one of the later drafts of Maʻalahi, I realized it could function as a
test vehicle to answer the question of whether grammatical simplicity/regularity or lexical
cognacy were more important factors in the ease of learning of an auxlang.
I would like to take this opportunity to encourage my fellow conlangers, especially those who are
professional linguists in university settings, to consider the craft of testlanging as a relatively
unused experimental technique.
There have already been some cognitive-psychological studies which have purported to use
“artificial languages” or “artificial grammars.” [Folia] Toutefois, these so-called artificial
languages are little more than sequences of symbols, which may or may not conform to an
arbitrary “grammatical” rule created by the experiment designer. In my opinion, these studies do
not actually examine the type of neurolinguistic processing that occurs with natural language
apprentissage. I propose that more fully developed artificial languages would function better to tease
out the intricacies of language learning.
It is my belief that the conlang community is positioned to offer help in the custom design of
constructed languages to function as testlangs.
The Future of Auxlangs
The foregoing may give the impression that I am opposed to auxlangs, or believe, Au moins, ce
7 Perhaps, someday, a testlang will be developed to examine Chomsky’s hypothesis of deep structure and
transformational grammar.
Maʻalahi
19
the pursuit of a true international auxiliary language is a pointless exercise. Nothing could be
further from the truth.
En fait, I believe that sometime in the future (after I, and all those who are alive today, are long
gone), an IAL to be used by the world’s populace will be designed, selected and authorized by
the world’s governments. Toutefois, this process will be brought about by external factors, basé
on the widely perceived benefits of peace and political and economic cooperation and
communication. No language, no matter how cleverly designed, nor how aesthetically attractive,
nor how easy to learn, peut, on its own merits, overcome the political inertia and provincial
nationalism that pervades the world of today.
The conclusions presented in this study, mais, may be a guide to those who wish to pursue the
craft of auxlang design. They indicate that it may be fruitless to pursue a goal of designing a
worldlang which will be easily learnable by all. It may be possible, mais, to create a regional
auxlang, such as Afrihili [Annis] or Lingua Franca Nova, whose lexical and grammatical source
languages all belong to one, or several closely related, linguistic families, and the learning of
which may be perceived as relatively easy by the intended target audience.
Cela ne veut pas dire, I wish to reiterate, that there never will be an IAL, but rather that its
adoption will not be an easy or painless process. No such language, in an environment of
linguistic diversity, can ever be driven solely from the grassroots; its promulgation must also be
supported economically and politically by the major sovereign powers of the world.
Outside of the personal enjoyment one may experience in constructing an engineered language,
there appears to be no reward in the activity of auxlanging (though this could be said of
artlanging as well). Worldlang development today can thus best be perceived as a playful
investigation into the parameters of language design, whose time of fruition has not yet arrived.
This investigation, néanmoins, can produce prototypes that may serve as examples for that
future time, when the world political structure will be finally ready for linguistic unity.
Remerciements
My appreciation to David Peterson for his encouragement in this project and his reviews and
critiques of early versions of the Maʻalahi language, to Philip Newton for detailed proofreading
and the correction of several errors, and to Joan Jensen for the statistical analysis in this study
and proofreading the write-up.
Maʻalahi
20
Glossary
Artlang
Auxlang
Conlang
Engelang
IAL
L1
L2
LCS
“Artistic Language” – Any conlang that is created primarily for aesthetic
reasons, par exemple, to increase the realism in literary works or films.
“Auxiliary Language” – Any conlang that is purposefully created as a
means of communication between people who do not share a common
langue, comme l'espéranto, Ido, Sambahsa, etc..
“Constructed Language” – Any language that has been consciously
devised (as opposed to natlang).
“Engineered Language” – Any conlang designed to fulfill specific
objective criteria, such as Ithkuil or Toki Pona. Most auxlangs could be
considered to be a type of engelang.
“International Auxiliary Language” – Any auxlang whose target is all the
peoples of the world, as opposed to a regional auxlang.
Same as “native language.”
Any non-native language spoken by a person, c'est, any language that
was learned after the age of six years (sometimes referred to as a “foreign
language”).
“Language Construction Society” (http://conlang.org/) – The premier
organization for those active or interested in the art and science of
constructing languages.
Native Language
Any language spoken by a person that was learned from birth (or before
the age of six years, according to some linguists).
Natlang
SAE
“Natural Language” – Any language that is not purposefully constructed,
but that has arisen “naturally,” such as English, Chinois, Hindi, Espagnol,
etc. (as opposed to conlang).
Abbreviation for “Standard Average European,” a term used to describe
auxlangs derived exclusively or primarily from Indo-European source
langues.
Sketchlang
A “sketchlang” or “sketch” is any conlang that is not fully developed,
perhaps only an idea or a concept for a conlang.
Maʻalahi
21
ALORS
TAM
Testlang
Abbreviation for Subject/Verb/Object, as applied to word order.
Abbreviation for Tense/Aspect/Mood, as applied to the grammar of verbs.
“Test Language” – Any engelang designed to test or demonstrate a
linguistic hypothesis, such as Loglan (or Maʻalahi).
Worldlang
Same as “IAL.”
Bibliographie
Annis, William S. Afrihili: an African Interlanguage. Fiat Lingua, April 2014.
[http://fiatlingua.org/wp-content/uploads/2014/04/fl-00001F-00.pdf]
A description of the regional auxlang, Afrihili, created by K. UN. Kumi Attobrah in 1967
as a lingua franca for sub-Saharan Africa.
Écologique, Umberto. The Search for the Perfect Language. Trans. James Fentress. 1995. Puits noir
Publishing.
The history of language construction in Europe from the middle ages to the present time.
Elbert, Samuel H. & Pukui, Mary Kawena. Hawaiian Grammar. 1979. University of Hawaiʻi
Presse.
The most complete grammar of the Hawaiian language. It covers all parts of speech in
detail, differing speech patterns between 19th century, tôt, and late 20th century usage,
rare constructions, and different linguists’ opinions on disputed areas of Hawaiian
grammaire.
Folia, Vasiliki; Uddén, Julia; de Vries, Meinou; Forkstam, Christian & Petersson, Karl Magnus.
Artificial Language Learning in Adults and Children. Language Learning, vol. 60, supplement 2,
Dec. 2010, pp. 188-220.
[http://pubman.mpdl.mpg.de/pubman/item/escidoc:541621:6/component/escidoc:541620/Folia-
LanguageLearning10.pdf]
A report on experiments in which adults and children are trained in an artificial
(constructed) language and tested on their ability to distinguish grammatical from
ungrammatical syntax. The neurological foundation of syntax learning for the artificial
language is tested using fMRI and TMS scanning. The results indicate that syntax
learning of an artificial language uses the same neural apparatus as for a natural language.
Maʻalahi
22
Lindstedt, Jouko. Native Esperanto as a Test Case for Natural Language. Festschrift in Honour
of Fred Karlsson, pp. 47-55. SKY Journal of Linguistics, vol. 19:2006 supplement.
[http://www.linguistics.fi/julkaisut/SKY2006_1/1FK60.1.5.LINDSTEDT.pdf]
A study of the usage of Esperanto among L1 speakers with particular attention to changes
introduced by the “nativization” of Esperanto. The author finds that the changes are
minimal and due mostly to crossover from the subjects’ other native languages. Il
concludes that the lack of change is due to the fact that Esperanto grammar is not learned
prescriptively as a formal grammar, but from the reading of Esperanto texts, qui
allows for the natural and unconscious learning of grammar, similar to natural languages.
Montúfar, Adelina Mariflor Solis. The Contemporary Esperanto Speech Community. Fiat
Lingua, Jan. 2013. [http://fiatlingua.org/wp-content/uploads/2013/01/fl-0000dix-01.pdf]
A description of today’s Esperanto speech community, focusing on why individuals
choose to learn Esperanto, the demographics of Esperanto speakers, and the factors that
lead to their coalescence into a speech community.8
NeSmith, R. Keao. Tūtū’s Hawaiian and the Emergence of a Neo-Hawaiian Language. ʻŌiwi
Journal—A Native Hawaiian Journal, vol. 3, 2005. [http://www.traditionalhawaiian.com/Oiwi-
Journal-_3-1-09_.pdf]
A description of the evolution of “Neo-Hawaiian” among L2 speakers with limited
exposure to L1 speakers, and examples of the differences between Neo-Hawaiian and
traditional Hawaiian.
Pukui, Mary Kawena & Elbert, Samuel H. Hawaiian Dictionary. Revised edition, 1986.
University of Hawaiʻi Press.
The most complete Hawaiian-English / English-Hawaiian dictionary. Also see the
excellent online dictionary: Nā Puke Wehewehe ʻŌlelo Hawaiʻi
[http://www.wehewehe.org/]
Roehr-Brackin, Karen & Tellier, Angèle. Enhancing Children’s Language Learning: Esperanto
as a Tool. 5th Essex Language Conference for Teachers. 29 June 2013.
[http://www.essex.ac.uk/langling/documents/elct/2013/esperanto_tool.pdf]
A PowerPoint presentation on the use of teaching Esperanto to middle-school children to
increase their metalinguistic awareness, and the benefits this had for their subsequent
performance on learning a foreign language.
8 Of particular relevance, for this study, is the fact that Asian learners of Esperanto had a good degree of fluency in
English previously, which may indicate that the ease of learning of Esperanto is due partly to its cognacy with
English and other European languages.
Maʻalahi
23
