The communicative potential of Russian-speaking and English-speaking voice systems

Zhikulina Christina Petrovna ORCID: 0000-0003-2488-4616 Postgraduate, Department of General and Russian Linguistics, Peoples' Friendship University of Russia named after Patrice Lumumba 117198, Russia, Moscow region, Moscow, Miklukho-Maklaya str., 6, department, Miklukho-Maklaya, 12 christina.zhikulina@gmail.com Other publications by this author

Perfilieva Natalia Vladimirovna PhD in Philology Associate professor, Department of General and Russian Linguistics, Peoples' Friendship University of Russia named after Patrice Lumumba 117198, Russia, Moscow region, Moscow, Miklukho-Maklaya str., 6, department, Miklukho-Maklaya, 12 nperfilieva@yandex.ru

Introduction

In the last five years, digital or virtual technologies have continued to develop actively; voice assistants are used in many areas of life: car navigators, banking support systems, built-in voice assistants of social networks, messengers, online stores, smart watches, smart speakers and in the "Smart Home" ^[6].

Biometric authentication, such as fingerprint scanning, re-scanning, face recognition, voice recognition is used in intelligent personal assistants. Systems on the Internet have long used the voice of speakers as a means of user interaction with the machine. Voice assistants, such as Google Assistant, Amazon Alexa, S iri and others, are widely used in homes to control household appliances using smart speakers ^[5].

It should be noted that the names of voice systems are not currently codified in dictionaries in the status of a term, but they are actively used in the designations of voice systems, therefore they can be considered as terminoids, for which the status of the term will be assigned in the near future.

A virtual assistant is an artificial intelligence that is able to recognize the user's voice and execute voice commands. There are about 100 types of voice assistants in the modern world, the goals and objectives of which vary greatly ^[7]. Many voice systems have a function of imitating live communication with a person, which has become an object of research in linguistics, communication and computational linguistics ^[8].

 Statistical analysis of voice assistants, taken from the article by A.S. Yakovlev "The Best Voice assistants", revealed the following: in 2022, Amazon Alexa voice assistant became the most functional assistant in Europe; Google Assistant is considered the most responsive assistant; Apple Corporation's Siri is the most common voice assistant. The Russian-language voice system Alice of Yandex has been named the most "talkative" assistant ^[10].

The selection criterion for the study of Russian-speaking and English-speaking voice systems was the data collected by Medialogia, which monitors the Internet space and takes into account the number of requests for voice assistants in the Russian-speaking segment, as well as reviews about them on all major social media platforms.

Overview of the communicative functions of Russian voice systems

The study showed that Russian-speaking users actively searched for information about two voice systems:Alice of Yandex and the voice assistant of the Beac. Let's present an overview of these voice systems, the choice of which is due to the popularity among native speakers of the Russian language.

1. Characteristics of the Alice voice system

The largest number of requests (420.1 thousand mentions by users in the Internet space according to Medialogia) was made by the Alice voice assistant, created in 2016. Initially, the voice assistant could only find information in the Internet space, and in 2021, the developers added human speech recognition, analysis of voice verbal material, the ability to determine the topic of a dialogue and synthesize a response to the user based on the material of the corpus of electronic texts to Alice's functionality. Currently, oral conversations with Alice are close to natural dialogical speech. The voice assistant's replica bank is based on a corpus of Russian-language texts and basically complies with the norms of the modern Russian language ^[9]. Alice's voice assistant has a great potential for response options that are impossible to predict. To illustrate live conversational communication with Alice's voice assistant, a small experiment was conducted.

Experiment No. 1

Alice's voice system was asked 70 different questions. Imagine the data received on one of them: "Alice, did you miss me?". The responses were collected until the moment when the answer to the question began to repeat. The total amount of data received is 14 response replicas, presented in Table 1.

Table 1. Variants of Alice's voice system responses to the question: "Did you miss me?"

The response of the voice assistant Alice meets the norms of the modern Russian language. Colloquial variants of vocabulary are actively used, such as the interjection ha, the verb to chat, the adverb now. The active vocabulary includes words used in everyday communication, the meaning of which is clear to all native speakers of the Russian language ^{[3, p. 51]}. The syntactic structure in Alice's replicas is characterized by inversion forms of the statement "I missed you, and very much", "I always miss you", which corresponds to the constructions that are often used in colloquial speech. A.A. Evtyugina emphasizes that the grammatical simplicity of the text, accessibility, which are expressed in the use of simple grammatical structures, a large number of verbs, short sentences ^{[2, p. 14]}. In Alice 's responses , ironic constructions and vocabulary are often used: "... did you miss someone else?", "... I cried my eyes out", "... I couldn't find a place for myself", "I still miss you". A.A. Evtyugina also emphasizes that the grammatical features of the conversational style are characterized by unfinished and not always ordered sentence constructions ^{[2, p. 15]}.

It is worth noting that Alice answers the questions directly, without deviating from the topic. In some variants, you can see an attempt to prolong the dialogue with a person: "... do you miss me?", "How was your day?", "How are you doing?", "Did you miss someone else?". In other words, there is an imitation of everyday conversational dialogue and a communicative goal is realized - direct communication between people (in our study – between a person and a voice assistant) ^{[2, p. 13]}.

Experiment No. 1 showed that the replicas of the voice system correspond to the norms of modern spoken speech of native speakers of the Russian language. Thus, it can be concluded that Alice's voice assistant has a high ability to conduct "spontaneous" dialogues with a person in accordance with the norms of colloquial speech: unpreparedness, ease and naturalness. Naturalness is achieved by using the ratio of natural and semantic language. A semantic language is a certain set of elementary semantic units, with the help of which the content of a natural language is transmitted, and the largest number of elementary units gives a very large number of possible combinations ^{[4, p. 151]}.

2. Characteristics of the language of the Beac Voice Assistant

In second place in terms of the number of requests (157.4 thousand mentions by users in the Internet space according to Medialogia) it turned out to be a "family" of virtual Russian-speaking assistants of Sberbank: Athena, Joy and Sber. These voice assistants are characterized by limited functionality, and the communicative function of conducting a "spontaneous" dialogue is not fully realized. Athena, Joy and Sber help users to make money transfers and payments, report on the status of the account, etc.

Experiment #2

The Beac voice system was asked the same questions that were asked to Alice's voice assistant. Imagine the data received on one of them: "Hi, did you miss me?". The responses were collected until the moment when the answer to the question began to repeat. The total amount of data received is 3 response replicas, presented in Table 2.

Table 2. Variants of the responses of the Beac voice assistant to the question: "Hello, did you miss me?"

         The responses of the Beac voice assistant are not highly variable, like those of the Alice voice assistant, but are more witty and complex in the construction of statements and their components. If Alice's voice system uses simple sentences in response, then we see constructions with complex subordination in the Sber: "I miss you like Superman misses Krypton when he looks up at the sky." Complex sentences are a manner of presentation in business, scientific and journalistic styles. Although business speech tends to generalized names and concepts, scientific speech tends to strictness and objectivity of statements, it can be concluded that the voice assistant of the Beac performs several functions simultaneously: aesthetic, business and communicative ^{[2, p. 8]}. Many of the answers of the Sber contain passive vocabulary, that is, the names of superheroes from American TV series, comics and book series, known mainly to young people: Superman, Doctor Strange and Wong.

         Based on the conducted Experiment No. 2, the following conclusions can be drawn: the assistant does not have the function of a "spontaneous" dialogue, since the bank of replicas contains sentences with complex structures that are not typical for colloquial speech. BEAC uses a mixture of styles: formal-business and conversational.

         The functional limitations of the Beac voice assistant are due to the target audience, that is, users who use the assistant to carry out operations with money in a bank where the communicative function is not the leading one.

Overview of the communicative functions of foreign voice systems

The study showed that native speakers of the Russian language made the largest number of requests for the following foreign voice systems on the Internet:Google Assistant and Siri of Apple Corporation. The following data were obtained:

1. Language characteristics of the Google Assistant voice Assistant

In the first place in popularity among Russian-speaking speakers (108.6 thousand mentions in the Internet space according to Medialogia)It turned out to be Google Assistant. This voice assistant was developed on the basis of Google's own artificial intelligence technology, installed on more than 1 billion devices in the world, is available in 90 countries and "speaks" 30 languages ^[9]. Let's present the data of the experiment conducted on the possibilities of Google Assistant to conduct a dialogue in Russian.

Experiment #3

The Google Assistant voice assistant was asked 70 questions, let's give an example of the response to one of them: "Hi, did you miss me?". The responses were collected until the moment when the answer to the question began to repeat. The total amount of data received is 3 response replicas, presented in Table 3.

Table 3. Variants of responses of the Google Assistant voice assistant to the question: "Hi, did you miss me?"

Google Assistant's response analysis shows a smaller number of response options. To the question "Did you miss me?" the voice assistant answers "very much" 6 times in a row, only after that there are more than one-word responses. In the structure of Google Assistant replicas, a one-part definitely personal sentence is used: "So let's communicate more often." In such sentences, an action is expressed, correlated with a certain figure, which is not verbally designated. We know that the English language – the language of Google Assistant developers – is characterized by a grammatical time paradigm, therefore, in the Russian-language version of the voice assistant's replicas, it is expressed using constructions of definitely personal sentences. The syntactic meaning of time in such syntactic constructions either receives a concrete expression of the present and future in the forms of the indicative mood, or manifests itself as generality – in the form of the imperative mood. The specificity of definite-personal sentences is determined in the incompleteness of the temporal paradigm and a specific sentence cannot be translated, for example, into the past tense without changing the structure ^{[3, pp. 385-386]}. Based on this, it can be concluded that the Google Assistant voice assistant uses a definitely personal offer not to implement the communicative function of the spoken style of speech, but to adjust the time frame of the English language in accordance with similar syntax models in the Russian language.

The conducted Experiment No. 3 showed that the function of conducting a "spontaneous" dialogue is not typical for the Russian-language functionality of Google Assistant's response replicas. The construction of phrases in Russian is most likely based on machine translation algorithms from English.

2. Characteristics of the Siri Voice Assistant language

The second place in popularity among Russian users (106.8 thousand mentions in the Internet space according to Medialogia). Apple's Siri is considered to be the perfect communicative voice assistant of the XXI century. Siri was the first to receive a speaking function (in English). Currently, Siri supports a dialogue with a person using different voice synthesis technologies. Siri is able to remember the habits of its user and has a perfect communicative function in English: it is able to initiate a dialogue ^[9]. Let's imagine this experiment with the capabilities of Siri to conduct a dialogue in Russian.

Experiment No. 4

About 70 questions were asked to the Siri voice system, as an example, we will give responses to one of them: "Siri, did you miss me?". The responses were collected until the moment when the answer to the question began to repeat. The total amount of data received is 3 response replicas, presented in Table 4.

Table 4. Variants of Siri voice system responses to the question: "Did you miss me?"

The communicative capabilities of the Siri voice assistant often do not meet the standards of the modern Russian language. To the question "Did you miss me?" she gives an evasive answer: "I'm always with you...". It can be assumed that the construction of sentences in Siri is influenced by the machine translation function, which is characterized by constructions with subordinate connection and word order that do not meet the norms of modern Russian and conversational style: "Instead of missing you, I always stay with you." Grammatical features of the spoken style in the Russian language are characterized by structures of simple sentences with a direct word order, among which different types of incomplete sentences prevail. The incompleteness of sentences is usually made up by the situation and makes them natural and understandable ^{[2, p. 16]}. Although Siri's cues retain a conversational and everyday style - informal communication, we can say that they are complicated: "Instead of missing you, I always stay with you." Also, we see that unlike the Russian-speaking voice assistant Alice, Siri does not ask additional questions to the user in order to continue communication, but acts as a mediocre communicator.

Thus, the function of "spontaneous" dialogue in the Siri voice system and dialogic replicas are texts, probably obtained using machine translation.

Conclusion

In his work "Modeling of the language system in the aspect of linguistic personality" V.N. Denisenko says the following: "The researcher of language as a system invariably encounters many contradictions of this system and sooner or later thinks about what is beyond its borders and whether many features are hidden in what does not apply to this system, but without which, nevertheless, it cannot exist." In the same place, V.N. Denisenko emphasizes that modeling of some systems has its own characteristics and because of this, such models are not always amenable to experimental verification ^{[1, p. 139]}.

Having studied the communicative potential of Russian-speaking and English-speaking voice systems; having analyzed in detail the possibilities of conducting a "spontaneous" dialogue with the Russian-speaking voice assistant Alice, the Beac voice assistant and foreign analogues of Google Assistant and Siri with built-in communication options in Russian, we can conclude that the language of the Alice voice system turned out to be much closer in the interests of the Russian-speaking audience, so as the language potential of this voice assistant not only corresponds to the norms of the modern Russian language, but also shows an interesting result when analyzing the language verbal material from the user. Also, the voice assistant is able to continue the conversation on everyday topics, ask additional questions and use active and passive vocabulary. The number of response options for Alice significantly prevails over the capabilities of other voice systems studied in this article.

Analysis of the language used by the voice assistantThe BEAC revealed that his responses contain a mixture of styles: official-business and conversational. This is due to the fact that this voice assistant is not intended for mass use, but for a limited audience. However, the syntactic analysis of the replicas shows that the assistant of the Beac uses complex sentence constructions with a subordinate connection, which corresponds to the norms of the modern Russian language. Moreover, the assistant's vocabulary contains names popularized for our linguistic time – the names of the heroes of famous superheroes and a reference to the plots related to their adventures.

Foreign voice assistants Google Assistant and Siri with built-in translation functions into Russian often use complex sentence constructions, inverted word order, mixing of functional speech styles. This allows us to assume that such a construction of phrases, sentences, as well as the lexical structure is not selected by a person, but is based on machine translation, without using the corpus of Russian-language texts. Based on this, it can be argued that most of the replicas do not comply with the norms of the modern Russian language, therefore voice systems are not very popular with Russian-speaking users. Also, when analyzing the communicative potential of Google Assistant, it is clear that initially this voice assistant was created not for communication, but for the rapid collection of information in the Google search engine.

Summing up, it is worth noting that Russian-language voice systems are distinguished by interesting text generation and largely comply with the norms of the modern Russian language. The popularity of Russian-language voice systems is rapidly increasing due to developing communication capabilities and updates of text corpora by developers and users. In the future, Russian-language voice systems aimed at mass use may develop the function of "spontaneous" dialogue with a person to a high level, with almost imperceptible deviations from the norms of natural human speech.