Content uploaded by Jawharah Alasmari
Author content
All content in this area was uploaded by Jawharah Alasmari on Jun 14, 2019
Content may be subject to copyright.
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
1
Investigating the Rate of Agreement and Disagreement of
Tense and Aspect of Quranic verbs in Arabic to English
Translations: Experimental Results and Analysis
Jawharah Alasmari1, Janet C.E. Watson 2, Eric Atwell3,
1, School of Language, Cultures and Societies, University of Leeds, UK
& Princess Nourah bint Abdulrahman University. Saudi Arabia
2 School of Language, Cultures and Societies, University of Leeds, UK
3 School of Computing, University of Leeds, UK
1 ml14jsnaleeds.ac.uk, 2 j.c.e.watson@leeds.ac.uk 3 e.s.atwell@leeds.ac.uk
Abstract
The practice and denotation of tense and aspect differ in Arabic and English, so there is a challenge
when translating between the two languages, particularly when the appropriate translation depends on
a range of linguistic contexts
1
, comprising also the context of use. In this paper, the Qur’anic Arabic
corpus of verbs is used in Arabic with their English translations by building a sub-corpus of verbs.
The study uses a statistical method incorporating SPSS and Kappa feature of SPSS to investigate the
rate of agreement and disagreement of Quran Verb Tense and Aspect in Arabic to English
translations. The aim is to provide information that can be used to address some of the challenges that
arise when translating between Arabic and English. The SPSS results indicate the highest percentage
for past, present and future tenses of Quranic Arabic verbs; the progressive and perfective aspect has
the lowest percentage. Kappa must is used to estimate the disagreement between translations with a
stronger measure than the SPSS percent agreement calculation, while κ also takes into consideration
the possibility of the agreement occurring by chance. The results show a clear disagreement between
the original text, and its translations, while the agreement varies between strong and weak. This
indicates that there are difficulties when translating Arabic verbs into English.
Keywords: SPSS tool, Kappa rate, Arabic verbs, English translation
1. Introduction
Arabic is part of the Semitic language family (which includes, for example, Hebrew, Aramaic
and Amharic), which has a morphological system that differs from that of English or other
Indo-European languages (Sawalha, 2011). To explain Arabic morphology, verb conjugations
must be understood.
The verbs are conjugated in different tenses to reflect gender, plurality, voice, and other
aspects. The inflectional verbal morphology of Arabic distinguishes between a suffix
conjugation and a prefix conjugation. These are typically referred to by contemporary
modern linguists typically as the perfect and the imperfect (Neme and Laporte, 2013). The
suffix conjugation typically refers to the past tense, while the prefix conjugation typically
refers to the present tense.
The grammatical categories relevant for verbs are person (first, second, third), number
(singular, dual, plural) and gender (masculine, feminine). These categories are realised in the
1
Linguistic context = discourse that surrounds a language unit and helps to determine its interpretation
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
2
suffixes of the suffix conjugation and the prefixes (and suffixes) of the prefix conjugation, as
shown in the tables below:
Table 1: The Suffix Conjugation Takes the Following Suffixes
kataba “to write”
‘I write’
aktubu
‘we write’
naktubu
‘you (m.)
write’
taktubu
‘he writes’
yaktubu
‘you (f.)
write’
taktubiina
‘she writes’
taktubu
‘you (m. pl.)
write’
taktubuuna
‘they
(pl.)write’
yaktubuuna
‘you (f. pl.)
write’
taktubna
‘they (f.pl.)
write’
yaktubna
‘you (dual) write’
taktubāni
‘they (dual)
write’
yaktubāni
Table 2: The Prefix Conjugation Takes the Following Prefixes and Suffixes
kataba “to write”
‘I write’
aktubu
‘we write’
naktubu
‘you (m.)
write’
taktubu
‘he writes’
yaktubu
‘you (f.)
write’
taktubiina
‘she writes’
taktubu
‘you (m. pl.)
write’
taktubuuna
‘they
(pl.)write’
yaktubuuna
‘you (f. pl.)
write’
taktubna
‘they (f.pl.)
write’
yaktubna
‘you (dual) write’
taktubāni
‘they (dual)
write’
yaktubāni
On the other hand, In the English, there are many as sixteen different tense structures; there
are tense forms such as present, past, future, and each tense has four aspectual references:
simple, progressive, perfect and perfect progressive (Gadalla, 2002).
Compared to other languages such as French or Spanish, Arabic has a more complex system
of morphosyntactic agreementHoles, 2004; Habash, 2010). Alkuhlani and Habash (2001)
explain this complexity in Arabic as being partly because of its richness, and partly because
of its complex morphosyntactic agreement rules which depend on features not necessarily
expressed in word forms, such as lexical rationality and functional gender and number’’
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
3
Employing SPSS statistical software and its kappa function, used to measure the agreement
between two translations, a statistical method can be used to establish the average of
agreement and disagreement between the seven translations of Quranic Arabic verbs. In order
to do this, a dataset of the contexts of the Arabic verbs appearing in the verses and their seven
English translations will be used.
The objective of this method is to compare and consider the highest and lowest agreement
rates between the original text of the Quranic Arabic corpus, and the seven English
translations by means of quantifying verb tense and aspect. The statistical evaluation of
agreements helps to produce some descriptive details of Arabic verb tens and aspect, and its
translations in clear and accurate ways. Using a statistical method here could thus assistance
to provide a general picture of the use of verbs in the Quranic Arabic corpus, and their
translations. The analysis of these details can help to improve the translation of Arabic verbs
into English.
2. Literature Review
In their study, Al-Sohbani and Muthanna (2013) find that the main problem in Arabic–
English translation is poor language knowledge, and inadequate and undisciplined grammar
practice. To translate from Arabic into English, the correct aspectual references of each form
can be delivered by considering the context, and by recognising particles controlled in the
contexts.
Eades and Watson (2013) refute the idea of Arabic as possessing only two true tenses, and
argue that the role of aspect is also present. They suggest that understanding the role and
function of aspect is not simple. They also consider how, in some cases, the context can alter
the usual syntactic effect. Verb forms can be used in different meanings or as indications that
can only be understood by referring to current evidence and context.
Gadalla (2006, p.8) argues that “a good translator must fully understand the context of an
Arabic tense form before attempting to render it into English. Understanding the context
helps him to understand the meaning of each form, which is very important for translation”.
In this respect, Ibn-ˁali (2009) also states that an understanding of syntactical and
morphological instruction must go hand in hand. According to Schulz et al (2000, p. 286)
“[A]s the participles per se do not express a tense in Arabic language, it must be decided by
the context which temporal reference is given in particular cases”.
Al-Ṭabṭabā’i (1983) has identified another basis for this time-tense relationship. Al-
Ṭabṭabā’i considers Arabic tense with the existence ‘real tense’ or nonexistence ‘unreal tense’
regarding the action of verb indication.
3. Methodology
In this experiment, a dataset of Quranic Arabic verbs with their seven translations were used.
1616 examples were investigated of the perfect and imperfect aspect verbs translations of the
triliteral root qāf wāw lām ‘to say’ qāla using SPSS software. Subsequently, Kappa
feature was used to investigate the rate of agreements and disagreement of the seven
translations.
A corpus syntactical and morphological analysis of the Arabic Quranic texts were used to
consider statistical calculations of features such as verb aspect and mood in order to highlight
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
4
the extent of the difference between the uses of verb according to inflectional and syntactical
morphology. The value was recognised depending on the Quranic corpus morphology/syntax
analysis of the verbs as perfect, imperfect, imperative, subjunctive mood, jussive mood,
indicative mood, and passive perfect/imperfect which are needed to consider the use of the
verbs in the seven translations.
The Arabic perfect tense was translated into English as “say,” “said,” “saying,” “were
saying,” “are saying,” “says,” “will say,” “shall say,” etc., which are also used to evaluate the
verb translations in the software. V1, V2, V3, etc were used to refer to the seven translations.
The label was divided into chapters, verb tenses, verses, and their translations. Consider
Figures 1, 2, 3:
Figure 1: Entering and Saving Data in the Data Editor Window
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
5
Figure 2: Entering Data in the Data View
In order to estimate the average percentages of verb translations, the descriptive statistics
feature was used.
4. Results and Analysis of Using SPSS
In the statistical analysis for descriptive purposes, I compiled the following tables of
examples to highlight the frequency and percentage of the verb use in the Quarnic text and
the translation agreements.
The verb aspects/ moods
Frequency
Percent
Valid Percent
Cumulative Percent
Valid
perfect
949
58.7
58.7
58.7
Imperfect/ indicative mood
207
12.8
9.2
67.9
Subjunctive mood
39
2.4
2.4
92.0
Jussive mood
20
1.2
1.2
93.2
Passive perfect
49
3.0
3.0
96.2
imperative
349
21.6
21.6
89.5
Passive imperfect
3
.2
.2
96.4
Total
1616
100.0
100.0
Table 3: Frequencies and Percentages for the Use of the Verb qāla Indicating Tense,
Aspect, and Mood Categories in the Quranic Text
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
6
Overall, the above table shows that only two aspectual forms, the perfect aspect and the
imperfect aspect verb, are used to describe past actions, and actions of the present or the
future. Furthermore, the indicative, jussive, and subjective moods, are indicated by only one
aspectual form.
The three tenses (relating to time/ when the action happens?), the past, present and future
in the two aspects, are established by the Arabic verb. The
progressive such as present continuous, and perfective aspects within ‘event time’ that gives
information about a prior action are provided as well by the two forms based on the context.
The impact of context effects is considered to be part of Arabic verb tenses.
The verb inflectional morphology of the perfect verb aspect forms, does not only indicate
tense or aspect as the corpus revels. It is clear that there are other factors in the contexts that
could affect verb tense and aspect. For examples, the past habitual aspect could be indicated
by the verb structure, and the particles or auxiliary verbs. Table 4 shows the percentage of the
perfect aspect verb forms translations to indicate tense and aspect. Consider the following
table: Table 4: Percentages of The translations of the Quranic Perfect Aspect Verb
Column1
Sahih
(%)
Pickthall
(%)
Yusuf
(%)
Shaki
(%)
Muhammad
(%)
Mohsin
(%)
Arrberry
(%)
Total
future
9.3
3.6
8.0
8.7
7.7
8.6
7.1
7.57
future passive
0
0
0
0
.1
0
0
0.02
infinitive
0
0
0
0
0
0
0
0
meaning
.1
.1
.8
.1
4.3
.3
0
0.83
Noun /Participle
.2
2.1
1.3
.4
4.7
.8
2.3
1.70
past
0
0
0
0
0
0
0
0
past continuous
0
0
0
0
0
0
0
0
past passive
0
0
0
0
.2
0
0
0.03
past perfect
1.1
.9
.9
.8
1.6
.9
.8
1.02
past perfect continuous
0
0
0
0
0
0
0
0
past simple
74.4
71.3
71.3
73.4
65.6
73.8
73.8
71.96
Perfect
Would- have been- used to
0
.1
.1
.1
0
0
0
0.05
present continue
0
0
0
0
0
0
0
0
present passive
.1
.1
.1
.1
.1
.1
.1
present perfect
.9
.5
.3
.2
5.6
.4
.6
1.23
present simple
13.9
21.2
17.1
16.0
10.0
15.0
15.3
15.49
100.00
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
7
The results indicate that the Arabic perfect verb aspect, which is usually built with suffixes, is
used to indicate multiple tenses/aspects, as the translations suggest. In the English
translations, it can be considered that as many as sixteen different structures; there are tense
forms such as present, past, future, and each tense has aspectual references such as: simple,
progressive, perfect and perfect progressive. The simple past is used in 71.96 percent of
cases, signifying the highest percentage; and the simple present in 15.49 percent of cases. The
future tense is realized in 7.57 percent of cases, and indicates the lowest percentage of using a
perfect verb. Perfective/imperfective aspects are indicated by the prefix conjugated verb as
past perfect with 1.02 %, perfect models with 0.05 %, and present perfect with 1.23 percent.
Other verb terms such as infinitives, nouns, or participles account for only 4.7 percent as the
highest percentage of using a meaning.
An investigation of the corpus translations reveals that Quranic Arabic verb structures are
used to indicate both time (absolute/relative) and action. The progressive and the perfective
aspects are expressed by the perfect aspect, which could also express the absolute and the
relative past tenses, or may express a completed and uncompleted action. Additionally, the
results show that the present perfect that is used to express a past event that has present
consequences is established as well.
5. Results and Analysis of Using COHEN'S KAPPA Statistic
Cohen’s kappa (also referred to as kappa) was used in the statistical analyses of agreement
across the seven translations. While SPSS was used to compare between the original text of
Quranic Arabic corpus, and the seven translations to show their compatibility and the extent
of differences in the expression of the verb in terms of tense/aspects between them, Kappa
was used to estimate the agreement and disagreement between translations with a stronger
measure than the SPSS percent agreement calculation, while κ also takes into consideration
the possibility of the agreement occurring by chance. Using Kappa helps to show strength or
weakness of agreement between the translations, which may be further evidence indicating
the differences between them. This disagreement may be the result of various challenges
translators encounter when translating Quranic Arabic verb tense/aspect.
Kappa is primarily used to measure the agreement between two individual sounds of the same
type. It is generally thought to be a stronger measure than a percent agreement calculation, as
κ also subtracts out agreement due to chance. Accordingly, Steven explains (online), “Kappa
measures the percentage of data values in the main diagonal of the table and then adjusts and
these values for the amount of agreement that could be expected due to chance alone.”
The kappa value is equal to 1 or less. Perfect or complete agreement is a value of 1, while
zero or less than 1 indicates moderate or low agreement. Consider the following tables which
provide details of the process of the experiment and its results. The tables 5, and 6 show the
difference of results between kappa and the percent agreement calculation:
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
8
Table 5: The Percentage calculation of Agreement between the translations
Table 6 Calculation of Cohen’s Kappa
Sahih
International
Pickthall
Yusuf
Ali
Shakir
Muhammad
Sarwar
Mohsin
Khan
Arberry
Sahih
International
1
Pickthall
0.62
1
Yusuf Ali
0.55
0.53
1
Shakir
0.68
0.67
0.66
1
Muhammad
Sarwar
0.41
0.41
0.45
0.49
1
Mohsin
Khan
0.58
0.62
0.73
0.68
0.57
1
Arberry
0.65
0.54
0.53
0.62
0.42
0.59
1
Subjects = 199
Raters = 7
Kappa = 0.566
As a data corpus of the seven translations is used to compile the results shown above, it can
be noted that there is a moderate agreement between the Sahih International and other
translations, and a substantial agreement with the Shakir translation, which has the highest
value at 0.68. The lowest value of agreement, 0.41, is found with the Muhammad Sarwar
translation, which yields moderate agreement. According to the value of agreement between
the Muhammad Sarwar translation and every other translation, the results provide the lowest
value, with a moderate agreement between 0.57 and 0.41.
Overall, the results show a clear disagreement between the translations, while the agreement
varies between strong and weak. This indicates that there are difficulties when translating
Arabic verbs into English.
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
9
6. Discussion of the finding
According to the results, it is clear that the descriptive statistical analysis has shown that
while Arabic has only two aspectual forms, perfect and imperfect, English has more
grammatical classes for tenses. Additionally, there is variation in the agreement between the
seven translations in terms of verb tense/aspect. It was clear that the whole context and
technical attentions by the translators have an effect on the Quranic verb translations.
In Arabic, the use of inflectional morphology aspects and the use of syntactical morphology
may had a reasonable effect on verb tense and aspect, indicating that some translators were
seemingly unable to translate verbs by recognising such as the prefixes and affixes that are
added to the verb forms, and some particles or auxiliary forms.
Tucker (2010) confirms when suggesting that English tenses do not follow the same patterns
as Arabic tenses. Shamaa (1978, pp. 32–33) also discusses potential reasons for the difficulty
encountered when translating Arabic tenses into English: “Temporal contrasts in Arabic are
less systematic, i.e., they are not clearly marked by verb forms. […] Temporal reference in
Arabic is expressed by means of verb forms in conjunction with time adverbials and other
lexical items. It is, however, the context which […] finally places the action or event in its
true temporal and aspectual perspective. But since context may not provide the same clear-cut
and easy determinations afforded by some European [e.g. English] tense systems, it is
therefore a source of occasional ambiguity”.
It is without doubt that each translator has his/her own way of translating, as the results show.
Muhammad Sarwar translations, for example, may base meaning in most translations cases
on contexts without recognizing the role of grammatical elements such as particles and
auxiliary verbs which can affect the verb tense/aspect in the context. In this regard, Gadalla
(2006, p. 244) makes the following observation “Certain verbs such as
kāna ‘to be’ and
certain particles such as
qad ‘already’ combine with these two forms of the verb to convey
various meanings’’. Thus, one of the major problems facing English–Arabic translators is
identifying the Arabic verb form and the verbs or particles that can be combined within it in
order to convey a particular English tense. Therefore, morphological and syntactic issues
have to be addressed when translating Arabic verbs.
7. Conclusion
The descriptive statistical analysis has shown that while Arabic has only two aspectual forms,
perfect and imperfect, English has more grammatical classes for tenses. It was clear that the
whole context and technical attentions by the translators have an effect on the Quranic verb
translations. From the experiment discussed here, it can be concluded that using a statistical
method is a useful way to collect, analyse, and make inferences from the described data.
Choosing appropriate English translations of Quranic Arabic verb tense and aspect using a
quantitative approach and statistical methods can supplement the qualitative analyses of the
data. Further, the data can be classified and counted in order to explain the observed facts,
which can then be used to improve the translation of the suffixes and the prefixes conjugation
of verb systems between the two languages.
International Journal on Islamic Applications in Computer Science And Technology, Vol. 6, Issue 1, March 2018, 01-10
10
8. References
Alasmari, J., & Atwell, E., Watson, J.2017. Using the Quranic Arabic Corpus for
Comparative Analysis of the Arabic and English Verb Systems. International Journal on
Islamic Applications in Computer Science and Technology. Vol 5. Available at:
http://www.sign-ific-ance.co.uk/index.php/IJASAT/article/view/1683
Alasmari, J., & Atwell, E., Watson, J.2017. A Comparative Analysis Of Verb Tense And
Aspect In Arabic And English Using Google Translate. International Journal on Islamic
Applications in Computer Science and Technology. Vol 5. Available at:
http://www.sign-ific-ance.co.uk/index.php/IJASAT/article/view/1683
Alkuhlani, A. and Habash, N. A Corpus for Modeling Morpho-Syntactic Agreement in
Arabic: Gender, Number and Rationality. Proceedings of the 49th Annual Meeting of the
Association for Computational Linguistics. Portland, Oregon. Pages 357–362.
Al-Sāmrā’i, I. 1983. Al-Fiҁl Zamānuh wa ‘abnīyathu. Bagdad. Maṭbaāt Al-Āni.
Al-Sohbani, Y. & Muthanna, A. 2013. Challenges of Arabic-English Translation: The Need
for Re-Systematic Curriculum and Methodology Reforms in Yemen. Academic Research
International, 4(4), p.442.
Eades, D., & Watson, J. C. 2013. Tense and Aspect in Semitic: A Case Study Based on the
Arabic of the Omani Šarqiyya and the Mehri of Dhofar. Boston: Brill.
Gadalla, A. H. 2006. Arabic imperfect verbs in translation: a corpus study of English
renderings. Meta: Translators’ Journal, 51(n° 1), pp. 51–71.
Gadalla, H. & Abdel-Hamid, A. (2000). Genitive constructions in English and Arabic: a
contrastive study. The Bulletin of the Faculty of Arts, 6. Assiut University, Egypt. pp. 1-
64.
Habash, N. Y. (2010). Introduction to Arabic Natural Language Processing. UK: Morgan &
Claypool Publishers.
Holes, C. 2004. Modern Arabic: Structures, Functions, and Varieties. Washington, D.C.:
Georgetown University Press.
Ibn Al-’Aṯῖr, Ḍ.1983. T: Aḥmad, A., and Badawi, T. Al-Maṯal Al-Sā’ir fi ‘Adab Al-Kātib wa
Al-Šāir. Dār Al-Rafāi.Al-Riyaḍ. Al-Mujalad 2.
Neme, A. A., & Laporte, E. (2013). Pattern-and-Root Inflectional Morphology: the Arabic
Broken Plural. Language Sciences, Elsevier, 40, pp.221-250.
Sawalha, M. S. S. (2011). Open-source Resources and Standards for Arabic Word Structure
Analysis: Fine Grained Morphological Analysis of Arabic Text Corpora. PH.D. Thesis,
University of Leeds.
Schulz, E., et al. 2000. Standard Arabic. Cambridge University Press.
Shamaa, N. 1978. A Linguistic Analysis of Some Problems of Arabic to English Translation.
PhD. Dissertation, Linacre College, Oxford.
Simon, S. What is a Kappa coefficient? (Cohen's Kappa). [Online]. Available at:
http://www.pmean.com/definitions/kappa.htm [Accessed 2 March 2017].
Tucker, A.M. 2010. The Morphosyntax of the Arabic verb: toward a unified syntax-prosody.
University of California, Santa Cruz, Qualifying Exam Paper.95064-1077.