Functionalism and Accuracy in Human Translation vs. AI Translation of Arabic Podcast Transcript: A Contrastive Study

1. Introduction

0.1. Objectives of the Study

The objectives of this paper are:

1- To Compare the accuracy and naturalness of AI translated version and the human translated version of the most widespread Arabic podcast “How Relationship succeeds”, to gain some insights into the benefits of using AI tools for translation in general, and in the podcast industry in particular.

2- To explore the limitations and capabilities of AI technology in the field of translation and to put the podcast in its right position on the map of translation.

2- To classify certain linguistic patterns and features that contribute to the presence or lack of naturalness and to what extent AI translated and human translated podcasts match and affect listeners response.

3- To investigate the role of podcasts in promoting communication and enhancing culture exchange through translation. This evaluates the extent of the listeners engagement rates for podcasts and the effect of using various methods of translation.

5- To assess the fluency and coherence of the translated content in terms of comprehension and readability for the target audience. This is through examining idiom adaptation and cultural appropriateness to convey the intended meaning and tone of the original podcast.

0.2. Statement of the problem

Due to the growing appeal of podcasts across diverse societies significantly in recent years, the dire need for functional and precise translations has become indispensable for reaching wider range of audience. In the light of this, the effectiveness and accuracy of human translation versus AI translation in representing the content of Arabic podcast, while preserving hidden meanings, is an important area of study. Human translators can carry idiomatic expressions and cultural nuances with them in the target language, the thing that may not be easily interpreted through AI generated tools. In this context, it is critical to determine which technique is more efficient and practical that can carry the original intention of the speaker accurately.

0.3. Hypothesis of the study

It is hypothesized that human translation is more functionally efficient in providing accurate translations of Arabic podcasts compared to AI translation tools. This is based on the assumption that human translators possess a deeper understanding of the cultural context and nuances of the Arabic language, allowing them to accurately capture any hidden meanings. On the contrary, the advancements in AI technology offer cost and time-saving benefits. Therefore, a combination of AI translation with human editing is proposed as a solution for translating Arabic podcasts into English.

 0.4. Research Questions

The current research is an attempt to answer the following questions:

1- How do human and AI translations differ in their linguistic structure and use of cultural references in the context of translating the podcast transcript?

2- How do these differences impact the effectiveness of the translation in engaging the target audience?

3. What is the significant difference in the time taken to translate podcast episodes using AI tools versus human translators?

4- What are the potential limitations and challenges faced in translating podcasts using both AI and human methods?

0.5. Significance of the Study

       The study offers detailed and comprehensive functional analysis of some selections of the AI translated version and the human translated version of the Arabic podcast “How Relationships Succeed?”, and the impact of this translation on the reader. Data is examined manually and qualitatively to track the nature of translation by focusing on five translation assessment standards. These standards are the readability and naturalness of the target text, equivalence between source text and target text, cultural appropriateness achieved in word choices, accuracy and appropriateness to the audience, in addition to the assessment whether the translation is literal representation of the original text or if it captures the intended meaning in a more nuanced way.

0.6 Data Sources:

     The data sources under study comprise some selections from the three-hour podcast "How Relationships Succeed," where Yasser Hazaimi shares insights. These selections have been translated into subtitles a human translator, namely “Latifa Alharbi”, and analyzed in comparison to translations generated by ChatGPT (version 4), an AI tool. The selected portions vary in their content and tone, with some featuring formal questions posed by the announcer, while others include informal responses from the speaker. Specifically, the chosen segments range from minutes 0:00 to 0:40 and 10:27 to 11:35 and 15:40 – 16:15. These segments demonstrate how the content from those minutes directly relates to the differences between AI translation and human translation. It also serves as representative examples of the differences the research is trying to show.

1. Theoretical Framework:

        This section offers an elucidation of the theoretical framework within translation studies in general, along with highlighting the Functional Theory of Translation in particular. The main aim of this part is to highlight the concept of equivalence and functionalism in the translation of spoken casual language taking podcasts as sample. Moreover, it aims at presenting a comprehensive theoretical detail about the concepts by which the two translations will be analyzed and compared, such as accuracy, coherence, readability, equivalence, cultural appropriateness of the target contents compared to the source text, as well as its adherence to syntactic and linguistic frameworks.

1.1 Translation Theories:

         According to Venuti (2000), “A translation theory always rests on particular assumptions about language” (p.6). According to Larson (1984), translation involves the thorough examination of the vocabulary, syntax, and cultural references within the source language text. By deconstructing and analyzing these elements, translators can accurately convey the intended meaning in the target language, applying the appropriate linguistic and cultural nuances for maximum comprehension (p.3). Newmark (1988) comments on the translation process saying “rendering the meaning of a text into another language in the way that the author intended the text"(p.5). There are many translational theories that should be applied in the process of translation to ensure effectiveness and accuracy. Some of these theories include Equivalence Theory, Descriptive Translation studies, Cultural Translation Theory, and Skopos Theory, which is a part under the umbrella of Functionalism and Functional Theories. Then theorists go beyond this to assess the function and the impact of this translated text on the audience, from which Functional Theories begin.

1.1.1 Functionalism and Functional Theory:

           Functionalism means the purpose or the goal of the translation. In other words, as Reiss & Vermeer (2014) refers to this “A translational action is governed by its purpose” (p.85). They confirm this by saying that “each utterance is an expression of the cognitive state and the intentions of the speaker” (p.85), the thing that can be hardly interpreted by an AI translation tool. They assume that in the process of an adequate translation, the priority goes directly to Functionality (p.86). They refer to the translation action, from functional point of view, saying that each action, and its interpretation presupposes a ‘theory’ about the action and its conditions. An action can be regarded as successful if the producer’s theory about the action and the recipient’s theory about the interpretation do not differ. According to Wang (2021), “Functionalism means focusing on the function of texts and translation.” (p.623). According to functionalists, as Reiss & Vermeer (2014) mentioned “Every action can be assigned a value which for the producer depends on its intention (intended function) and for the recipient on its interpretation” (p.88). According to Arbabi & Farahani (2019), “Translation functionalism is classified into those theories of translation which define translation as a product-oriented process, in accordance with the requirements of the commissioner and the target text receiver” (p.66).

       Generally, Functional Theories in translation highlight the importance of the communicative function of a text over its linguistic form. These theories argue that the final target of translation is not simply to reproduce the words of the source text in a simple way, but to convey the intended meaning and message to the target audience in a way that is culturally and communicatively appropriate to the community they are in.

1.1.1.1. Skopos Theory:

       Skopos Theory is a specific approach within functional theories that places particular emphasis on the concept of "Skopos", which is a Greek term meaning "purpose" or "goal”. It was developed in the late 1970s and early 1980s by Hans J. Vermeer, a German scholar of translation studies. According to Skopos Theory, the major goal of a translation is to fulfill a certain purpose or meet a particular function for the target audience. It is based on functionalism, and it claims that translation is guided by extra-linguistic factors, which is the purpose of the translation (Reiss & Vermeer, 2014). The translator's task is to produce a target text that achieves the intended purpose or function, even if this means deviating from the linguistic form or style of the source text. At this point, this research is an attempt to determine if AI technology is able to accurately understand and convey the intended message of the speaker without misinterpretation or misunderstanding.

       Functional Theories, specially Skopos, permits for more flexibility and compliance in the translation process, as the translator is encouraged to prioritize the intended function of the target text over strict adherence to the form of the source text. Thus, the rationale of the present study is to explore how functional linguistic approaches can enhance our understanding of translation quality by examining how well translations meet their intended purposes and functional requirements.

1.1.1.2. The Concept of Equivalence in Translation:

       Translation theorists and functionalists explore the idea of equivalence and its different types, such as word, textual, grammatical, and pragmatic equivalence. According to Baker (2011), analyzing word-level equivalence is the first step in the bottom-up approach to translation. Grammatical equivalence can be challenging due to language differences, while textual equivalence aids in maintaining cohesion between the source and target texts. Pragmatic equivalence involves understanding implicatures to address cultural differences. Catford (1965) sees equivalence as situational, while Nida (1964) views it as communication-focused.

       Accordingly, Catford’s (1965) refers to the concept of equivalence in translation saying, “In total translation SL [source language] and TL [target language] texts or items are translation equivalents when they are interchangeable in a given situation” (p.49). Afterwards, House (1977) created her method using functional linguistics theories, emphasizing the importance of having a specific function for the translated text. House (1977) stated, “translated texts should not only match their source texts in function, but employ equivalent situational-dimensional means to achieve that function” (p. 49). She focused on ensuring the target translation closely matched the source text while also suggesting that translators should filter for cultural nuances to achieve equivalence. House (1977) believed that translation should serve a communication purpose rather than just transferring information. Ultimately, House's approach aimed to organize the assessment of translation.

       In this respect, Functional equivalence, from translation studies point of view, inspects how particular cultural communication interacts with translation goals. At this point, Functional theories in general, and Skopos' theory in particular, confirm Reiss and Vermeer (2014) concept that “A translational action is governed by its purpose” (p.122)

1.2. Standards for Assessing Translations

       There are many standards to assess a good translation among which is equivalence in meaning between the original text and the target language, ensuring that the intended message is accurately conveyed. It should also be readable and easily understood by the target audience, making sure that the translation flows smoothly. Accuracy is essential in maintaining the integrity of the original text, while also considering the cultural nuances and context of the target language. Additionally, a good translation must take into account the specific needs and expectations of the audience, adapting the language and style accordingly to effectively communicate the message. Overall, a successful translation should encompass all these factors to provide a high-quality and effective reflection of the original text.

1.2.1. Coherence

       Coherence, in translation, is meant to be semantic and pragmatic relations that give the text unity, meaning and purpose (Colina, 2015). On the other hand, in translation, for a message to be considered as 'understood', it must be coherent both within itself and with the situation of the recipients. In simpler terms, the message must be logical and relevant to the recipients in their specific context. This concept emphasizes the importance of the message being coherent with the situation in which it is received, rather than being universally understandable. Essentially, understanding involves connecting the message to one's own situation and the relevant background knowledge it entails. Thus, in translation, achieving coherence is not just about the translator’s adherence to literal meanings, but also about maintaining the overall sense of the text and considering the background knowledge of the audience. Coherence, in translation, is thus established by connecting cultural and factual elements with linguistic understanding. Without considering these factors, a translation may simply become a literal rendering of the text. Consequently, Reiss &Vermeer (2014) comments, “A failure to achieve coherence in translations may be due to the translator’s carelessness (we shall not consider ignorance as a possible reason at this point), but it may also occur if the translator, while sticking to the assumed meaning of a word, loses sight of the sense of the text as a whole.”. (p.100). They add saying that another reason for coherence loss in translation “may be that the translator does not take into account the differences in the background knowledge of the source and target recipients. As we know, coherence is established by linking cultural and factual with linguistic knowledge” p. (100-101). Accordingly, it can be said that AI tool cannot match the quality of translation that can be achieved by following these instructions given to human translators.

1.2.2. Audience

       The audience of a translated text plays a crucial role in the process of translation as they are the ones who will ultimately consume and interact with the translated material. It is crucial to consider the audience's language proficiency, cultural background, and specific needs in order to ensure that the translation is effective and meaningful for them. According to Colina (2015), the audience are the receivers to whom the text is addressed including characteristics of those addressees: their knowledge, gender, age, educational, cultural, and socioeconomic backgrounds.

One way to put the audience into consideration in the translation process is to conduct thorough research on the target audience before beginning the translation, the thing that can never be achieved by an AI tool for electronic translation.

       According to Reiss & Vermeer (2014), “The intended audience (‘addressees’) or recipient may be described as a specific kind or subset of skopos. How an interaction is carried out depends, among other things, on the relationship between the parties to an interaction.” (p.92). The audience makes the translation functional by determining whether it effectively conveys the intended message and achieves its communicative purpose. By tailoring the translation to suit the needs and preferences of the audience, the translation is more likely to be coherent, engaging, and impactful for the target readers or viewers. Colina (2015) comments on this saying “A component of evaluation that examines how well the translation achieves the goals, purpose and function of the text for its target audience and purpose.” (p. 296).

       This approach prioritizes the end result of the translation, focusing on meeting the expectations and needs of the client and audience of the translated text. This view sees translation as a service that aims to accurately convey the message and meaning of the original text.

1.2.3. Accuracy and Readability

       Assessment of high-quality translation should also involve the concept of accuracy. An accurate translation is one that is easy to read, and well-received. The evaluation of how accurately the original text is conveyed into the new language should involve evaluating the accuracy of message translation always requires comparing the original text in the source language with the translated text in the target language. In essence, comparing the source text with the translated text is a key aspect of assessing translation accuracy.

       Thus, one of the key standards in evaluating the quality of a translation is assessing its accuracy and readability and the ability of the audience to read it in a smooth way. A successful translation should effectively convey the original message into the target language while maintaining fidelity to the source material. This evaluation process extends beyond the mere judgment of the researcher and necessitates a thorough comparison between the source text and the translated version. By scrutinizing how well the essence and meaning of the original text are preserved in the translation, a more comprehensive assessment of accuracy can be achieved. Moreover, an accurate translation should not only mirror the content of the source material but also it should be easily digestible and well-received by the intended audience.

1.2.4. Equivalence

       In Koller's 1989 work, the concept of equivalence in translation is explored, specifically focusing on connotation and denotation equivalence. High quality translation is based on two sorts of equivalence when translating meaning from one language to another. The first one is the denotative equivalence that ensures accuracy in transferring the literal meaning of words. Subsequently, human translator goes beyond word-to-word translation and involves capturing the hidden connotations of the original text. At this point human translator is the only one that can preserve connotation equivalence and convey the full depth of cultural and contextual meanings in translation

1.2.5. The Process of Translation is culturally Bound

       Translation also involves the adaptation of language and cultural context to ensure that the message is accurately conveyed to the target audience. This process requires translators who have a strong command of both languages, and a deep understanding of the cultural nuances that can impact the meaning of the text. According to Reiss & Vermeer (2014), “Through the process of their socialization, (‘enculturation’) humans become members of cultural community and often of other communities as well” (p.23). When they grow up in certain community, they adopt the ideas and concepts of such society and culture the thing that appears a lot in the art of translation. According to Reiss & Vermeer (2014) “A language is part of a culture. Cultures use language as their conventional means of communicating and thinking. Culture encompasses a society’s social norms and their expression” (p.23). The art of translation is a crucial skill that requires a human touch, especially when dealing with cultural nuances, as the case with podcasts. Podcasts often discuss social life and use everyday language that may not easily translate word-for-word. A human translator can effectively convey the intended message, tone, and cultural references with accuracy and fluency, ensuring that the content is properly understood by the target audience. This human touch is essential in bridging the gap between languages and cultures, making the translation of podcasts a truly nuanced and specialized task that requires a deep understanding of both the source and target languages.

According to Reiss & Vermeer (2014)

Culture specific and individual views about the worlds are, in a way, ‘frozen’ in language as a means of communication and thought. ……The values assigned to objects and phenomena vary from culture to culture and from individual to individual. All these refractions can cause translation problems.” (p.23)

They comment on this point saying “A language is part of a culture. Cultures use language as their conventional means of communicating and thinking. Culture encompasses a society’s social

norms and their expression.” (p.26)

1.3. Podcasts

       Nowadays, podcasts play a significant role in shaping our media consumption habits and provide a platform for diverse opinions and perspectives. Podcasts have become so widespread and common in all languages because of being accessible. Podcasts can be easily listened anywhere allowing people to consume content while driving, exercising, or doing household chores. This accessibility has contributed to the popularity of podcasts as a convenient form of entertainment and education. Podcasts cover a wide range of topics, specially what is related to personal development. This diversity allows listeners to find podcasts in a meeting point with their interests and preferences, leading to a wide audience for podcasts in all languages. Creating a podcast is relatively easy and affordable, compared to traditional media formats like TV or radio. Overall, the combination of accessibility, diverse topics, low barriers, and personal connection has contributed to the widespread popularity of podcasts in all languages. As a result, translating podcasts is crucial for expanding their impact and reach to gain global audience.

       The podcast “How Relationships Succeed”, and its translation, in particular is chosen for being widely spread. It is provided through the famous Arabic podcast “Fnjan’ and is the most-watched podcast episode. It is one of the tops in global YouTube charts as it exceeds 100 million views. According to Reporter “Saudi podcast Fnjan has amassed over 100m views in just over a year on its episode featuring relationship and communication skills trainer, Yasser Al Hazaimi”. It is recorded that 21% of the audience watched the episode with English subtitles, highlighting its widespread international appeal. Many viewers refer the episode’s success to Al Hazaimi’s presentation style and society’s growing interest in understanding the social relations, particularly in the face of contemporary life’s challenges. Overall, "How Relationships Succeed" is important because it offers valuable guidance and insights on how individuals can attain meaningful relationships in their lives.

2. Analysis

Example1:  Audio  from 0:00 to 0:40 

       In this example, AI translates the Arabic structure as it is but with no mistakes. It negates the presence of relationships without needs in a simple way. It is a copy of the Arabic structure. On the other hand, human translation confirms that ALL relationships are for need and this translation ignores the use of “negative structure”. It goes beyond the simple structure of the speaker and refers indirectly to the name of the podcast which is “How relationships succeed”. So, it gives quick solution, all relationships are based on needs. This seems to be more functional and accurate. It answers the question under study in the podcast. It here focuses on the audience and the message and the tone conveyed by the speaker and sent to the audience.

       In general, confirmation can be used to assert a belief or statement as a universal truth. Human translator confirms a bold assertion about the nature of relationships and emphasizes the pervasive influence of need in human interactions, the thing that can never achieved using AI tool. Ultimately, the choice between negation and confirmation to convey meaning depends on the context, purpose, and audience of the message. Both approaches can be effective in different situations, but here the human translator considers the specific message being communicated before deciding which method to use.

       Another difference between AI translation and human translation is in the choice of vocabulary. AI uses very simple and straight forward word-to-word translation like the words “relationship” and “rule”. In the AI translation, the word choices are very direct and the words in the translation have their equivalents.

        The human translation also uses the same wording but sometimes resort to more deep words such as the usage of the verb “drive” in the passive voice in the structure “relationships are driven by need”. The word “driven” focuses on the meaning of motivation and the strong impact of this need that leads people to relationships. Human translator is more elaborative to send the tone, body language and the overall sense of the context. Conversely, the AI translation is word to word translation. In general, the human translation is more complex and sophisticated while the AI translation is basic and simpler and easier which may make it accessible to wider audience.

       Concerning Sentence Structure and Readability, AI translation uses a direct and literal approach, translating each utterance separately and presenting them in individual lines. This allows for a clear breakdown of each point being made and affects readability, with representing each line a distinct idea. This approach may be more suitable for a line-by-line analysis. Human translation combines the meanings of the original text and uses semicolons to connect related information and presents it as one cohesive sentence. This approach supports the readability of the text and condenses the information. It also provides a unified interpretation of the text by linking related ideas together.  This approach emphasizes the desired outcome and appears to be more effective in conveying a holistic interpretation of the text.

       Concerning Cultural references, in the Arabic “  "هو المستغني سبحانه"Allah" is the Arabic word for “God” used by Muslims. In the Islamic faith, it is customary to refer to “God” as "Allah" in all contexts. On the other hand, "God" is a term used in English and is commonly used by Christians and people of other religions to refer to the supreme being. It is also used in translations of the Bible and other religious texts.

       AI translation uses "God" which is more universal and does not account for cultural differences. Conversely, the human translation using "Allah" considers cultural differences, especially if the speaker is Arabic. While the translation may accurately reflect the Arabic text, it may not resonate with the English-speaking audience.

Similarly, the cultural difference appears in translation of     "سبحانه"

AI translation uses “He is self-sufficient and perfect” as it is even without capital letter for the adjectives “sufficient “and “perfect”. It only uses capital for the pronoun “He”. Based on this, AI provides a literal translation of the term without considering the cultural context in which the term is used. Conversely, the human translation takes into account the cultural significance of the term in Arabic language and Muslim beliefs and translated it as “SWT” that stands for "Subhanahu Wa Ta’ala, “. It is a term used by Muslims to show respect when mentioning Allah. Therefore, in this case, the human translation may be more appropriate for conveying the full meaning and significance of the term by the speaker but it ignores the understanding issue to the listeners.
The researcher’s point of view is confirming the fact that good translation is the one that captures the intended meaning of the speaker as well as considering the audience.
A more comprehensive translation that takes into account both the Arabic cultural context and the understanding of the English-speaking audience such as  “Allah SWT (God who is Self-Sufficient and Perfect)” would be ideal.

       This approach ensures that the richness and nuances of the Arabic language are preserved while making the meaning accessible and meaningful to a broader audience. In general, the human translation is more complex and sophisticated while the AI translation is basic and simpler and easier which may make it accessible to wider audience.

Example 2:   Audio from 10:27– 11:35

The speaker here in this sample audio is talking about the difference between the presence of God in Muslims and non-Muslims lives and shows to what extent God’s presence affects people’s everyday life.

       Unlike the previous example, in the AI translation, the reference to God is translated as “Allah”, which is more appropriate than “god” that is used in the human translation, but it is used preceded by the article “the”, which is not right as the word Allah is well-known for Muslims and frequently used to refer to their God and is not in need to be used with article.

       On the other hand, the human translation refers to God using the  common noun “god” even without Capital letter which is very appropriate in this context, as the speaker’s intention here wants to provide a general fact about the presence of the idea of god, even if it is not to Muslims, and then he applies this on Muslims life in the second sentence. At that time, the human translator refers immediately toالاله" as Allah which is meant to be The God of Muslims specially and not the idea of having God in general.

        This shows to a great extent the efficiency of the human translator to convey the intended message of the speaker, and shows also that the AI translation is literal and translates every word and sentence on its own regardless its intended message and the overall context.

       Another example appears in the same sentence is the redundancy and of using the word “absent”. In the Arabic context, the speaker uses “غاب الاله, غاب الرقيب, غاب الشاهد, غاب مقام الاحسان" “the stress on the word غاب is to show that the absence of every item is a consequence to the absence of the other and they are result of each other.

       In the AI translation, this is translated by repeating the word “absent” which is not the meaning, on the other hand, it is highlighted in the human translation by using the word “Then” which indicates that everyone is a result or a consequence to the other. This may show that AI translation is accurate and insists to reflect every word in Arabic but in a very literal way regardless the intended meaning that appears vividly from the tone of the speaker and is deeply understood by the human translator.

      The translation of the Arabic phrase “مقام الاحسان “is translated in the AI translation in a very literal way. The word “مقام” is translated as “place” and the word "الاحسان is translated as “excellence”. In the human translation, it is translated as a whole term that exists in the Muslims culture as “Beneficence”, which is considered as a kind of respect to the culture of Muslims, and that they have their own terms. Concerning denotative meanings, “Beneficence” in Oxford dictionary means “ the act of doing good; kindness”, or “a charitable act or gift”. On the connotative level, it reflects the intended meaning of the speaker and meets the culture of Muslims.

       In this sentence, the AI translation seems to resort more to the simple rather than complex or compound sentences. It is full of numerous to-the -point simple sentences. This facilitates the communication and clarifies that the text is clear and to the point, but this lacks the tone of the speaker. Direct and simple sentences are more readable to the English speaker and avoids the reader’s confusion. This helps the smooth flow quickly that engages and invites the reader. Furthermore, it resorts a lot to the overuse of punctuation markers and the excessive use of commas that separates the small simple direct sentences and this makes it more readable to the reader. In the human translation, the overuse of commas and punctuation is to reflect the pauses of the speaker which are not present in the AI translation. This is because the AI translation is that of written text.

       In this part, the difference between the translations lies in the difference in lexical choices in the translation of the word “تعاملي”. It is translated in the human translation as “the way I treat you” and the AI translation as “interactions”. Actually, it appears from the whole context that the speaker in Arabic means the way people deal with each other. The choice of the verb “treat” is better because it is commonly used in this context, while “interactions” is commonly used within the chemical reactions and interactions between substances.

Besides, the lexical choices of the word “الرقيب” which is one of the 99 Names of Allah. In the AI translation, it is translated as “the observer “which is commonly used as the translation of the word” المراقب” and not “الرقيب"  which is translated in most of the Qur’anic verses translations as the Most Watchful. Here the human translator preserves the Muslims culture and beliefs and uses the translation commonly used in the Quran, unlike the AI that translates blindly a kind of word – to – word literal translation

       In this part, the human translator completes the insistence to preserve Muslims’ culture. The translator uses for “الله عز و جل “as “Allah AWJ” which is the usage of the first letter of every word in Arabic and uses them in English.

From the writer’s point of view, this is typically correct if the reader is Arabic speaker, but if the reader is English speaker and most commonly not Muslim, this is not readable and not accurate. The translator here should have used (AWJ) to preserve the Muslims culture but should have also clarified the meanings of these acronyms between brackets to make the translation readable and understandable.

        The AI translation in this part ignores the culture of the Muslims and skipped “عز و جل “from the translation which confirms the literal translation strategy adopted from the beginning.

Then we have the translation of the word “عرض” which is translated in the Human translation as “he offered us the trust” but is literally translated in A I translation as “presented” which is totally far from the intended meaning. The AI deals with the word as if it is a “presentation” and uses its verb. It sticks to the subject and object used in the Arabic but the human translation resorts the usage of passive voice to solve this problem.

       Within the stream of speaking, the speaker is talking about how can people forgive each other, and he suddenly says " الله يقول "and then completes by giving an example about what Aisha (May Allah be pleased with her). In this case, the cognitive abilities of the human being skipped this because it is out of context, but is written in the script and the AI translation tool never skipped as it lacks the human cognitive abilities that can filter what is meant to be and what is not.

       Furthermore, AI tool deals with the whole example as a quotation and is put between brackets.  This is only because it follows "الله يقول" and it misunderstands that the rest of the example is like Quran said by God.

       It appears clearly from this example that the human translator clarifies first the situation between Aisha and her maid before mentioning the quotation so as to facilitate to the reader that this quotation is based on certain situation, unlike the AI translation that resorts to the quotation first although it is not understandable without understanding the quotation first.

Besides, the human translator understands from the context that Aisha by saying “يوم يقوم الناس لرب العالمين " means that we as Muslims will take our reward in the Day of Judgement, which is well-known in our religion and our culture and is not in need to be described. That’s why the human translator uses a proper noun using capital letters, and here appears the outstanding role of punctuation markers that refers to information and imply hidden means to be clarified to the reader but it cannot be used by any AI tool.

At the end, “"انا اسكت this is also clarified by the translation of the last words but it is understood from the context that Aisha will stop and will not blame her maid, and it does not mean “silent” as the way it is translated by the AI tool which is a very literal way.

Example 3: Audio from 15:40 – 16:15

In this part the announcer is asking about the importance of spiritual aspect on human beings in reality.

       In this part, the announcer means to take the theoretical ideas and apply on our lives. Functionally, this is what appeared from the human translator when she translates “ننزل شوية للواقع "and uses verb “to apply” which is totally convenient to imply the intended meaning by the speaker. On the other hand, AI translation sticks to the literal translation and uses “go down to reality” which is not the meaning and which is not functional to let the reader understands the meaning.

       In the first sentence, the announcer is trying to apply on the real life in our time and he said:” الواقع اليوم”. The human translator follows functional approach and translates the whole meaning and translates it into “Nowadays”, while AI translates it literally as “today” which is commonly used when talking about simple actions in our exact day not our age or generation as meant by the Arabic context.

       The utterance "الغرب و الغربيات"  is mentioned by the announcer by which he means The western countries and the ideas related to their culture, and to what extent the Arab countries are influenced by them. This is efficiently conveyed and translated by the human translator as one whole using “the influence of the west is present”. This translation is correct, accurate, and equivalent to the intended meaning of the original text. It also reflects the tone and the intention of the speaker clearly. Conversely, it is translated by AI as “the West and Western women”, the thing that shows that this tool adopts literal translation and word-to-word direct equivalence. It ignores the total context of the sentence and translates "الغربيات" as the feminine plural for western people.

       In this part, the announcer is referring to everything that has been said previously by the speaker and wants to ask a question in different area. He used فكل الكلام اللي انت قلته"”. "الكلام " here refers to the ideas and sequence of opinions presented. It is totally understood by the human translator and conveyed within context when she uses “What you have said”. On the contrary the AI tool translates in a literal way and uses “words” as a synonym to "الكلام". which is neither accurate nor equivalent.

      Likewise, AI translates "ديني بحت" as “purely religious” while the human translator absorbs the meaning and simply translates as “religious”. Finally, during this live interview, the announcer repeatedly asks a question about the impact of religion on people's lives, using the Arabic word "او" to clarify and emphasize one point. This repetition, that is frequent in live podcasts, is correctly understood by the human translator, who conveys the overall meaning with a single, concise translation. In contrast, AI translates each repeated question literally, resulting in a translation that lacks fluency and coherence.

3. Conclusion and Findings

Throughout this detailed analysis, the key differences between human and AI translation of the Arabic podcast "How Relationships Succeed," can be summarized in 5 points based on the standards of assessment previously mentioned,

1- AI translations often exhibit a literal, word-for-word approach, failing to capture the contextual meanings inherent in human communication. Human translators, on the other hand, possess a deeper understanding of the underlying message, tone, and intention of the speaker and can convey it accurately and effectively.

2- Concerning the audience, human translator sometimes considers the intended audience. This is because human translators can collect the meaning and purpose of an utterance, even when the text is repetitive, as in a live session. However, in other cases and instances, both AI and human translators may sometimes fail to effectively communicate with the audience. For instance, when translating concepts related to Muslim culture, human translators might use language that is only understandable to Arabic speakers, even if the intended audience is English-speaking.

3- Regarding the concept of equivalence as a third point of comparison between human and AI translation, AI translation algorithms aim to achieve equivalence between the source and target languages, their approach often results in word-to-word translations that may lack the nuances present in the original text. On the other hand, human translators s tries to achieve a deeper level of equivalence that encompasses not just linguistic accuracy but also cultural appropriateness and semantic fidelity and faithfulness.

4- The fourth point highlights the role of cultural sensitivity in translation. Human translators are able to navigate through the cultural references of the source language while ensuring that the translated content remains culturally relevant for the target audience. In the case of translating the Arabic podcast "How Relationships Succeed" into English, human translators are better equipped to preserve the cultural context and nuances of the Arab culture. However, human translators may at times prioritize cultural authenticity to such an extent that they overlook the needs and preferences of the English-speaking audience.

5- Moving on to the final point, coherence plays a crucial role in determining the quality of a translation. Human translators excel in creating a coherent narrative that flows naturally and resonates with the target audience. In contrast, AI translation systems often struggle to maintain coherence, especially when faced with the nuances of spoken language. AI tends to break down the text into individual components, leading to disjointed translations that may not convey the intended message effectively.

       In conclusion, the comparison between human translation and AI translation of the Arabic podcast "How Relationships Succeed" confirms the irreplaceable role of human translators in capturing the nuances, emotions, tone, and cultural references embedded in the original text. Although AI translation systems offer speed and efficiency and saves time, they fail to achieve the human touch and contextual understanding and cognitive abilities that human translators can bring. When it comes to translating content that revolves around complex topics such as relationships and cultural dynamics, human translation remains the preferred choice for ensuring a meaningful and engaging cross-cultural communication experience and accurate outcome along with the assistance of AI translation tools to save time and effort.

4. Recommendation for Future AI Translation Models

       Future AI models should hold extra abilities and to undergo technical and linguistic modifications to be able to compete with human translators later on. For instance, it needs to be consistent and effective in using punctuation markers to preserve communication. It is also important to work on the area of reducing redundancy. This is because AI models often repeat information unnecessarily. This can be addressed by refining algorithms to identify and eliminate redundant phrases. Moreover, sentence structure modification is a must to enhance the fluency and readability and accuracy of translated text and to construct more sophisticated and grammatically correct sentences characterized by equivalence on the level of intended meaning and the speaker’s tone. In addition to this, AI models should be trained to deal with figurative language and implied meanings and metaphors and tones more than the abilities available in the present models to be able to translate spoken language. It should also be updated with the facilities that understands context and to translate contextual situations rather than word-to-word literal translation.

By focusing on these areas, AI translation models can bridge the gap between AI and human translation, providing more accurate, nuanced, and engaging translations.