The development of undergoer voice construction in Javanese

Abstract

This article aims to investigate the development of undergoer voice construction in Javanese from 10^th century to 21^st century. This research discusses the development of undergoer voice constructions by focusing on affix markers and the presence of agent arguments, especially agents in the form of first-person personal pronouns. The method employed in this research is a qualitative approach by utilizing corpus linguistic methods for data collection. The data sources used are taken from literary texts written include prose and poetry from the 10^th to the 21^st century. There are 12 texts that are used as data sources in this research. The literary works vary both temporally and regionally. The results claim that Javanese does not have completely the same type of construction as Filipino has the same type as Indonesian in respect to the presence of proclitic in the undergoer voice construction. The findings indicate that agents in the form of first and second-person personal pronouns are crucial in the study of undergoer voice construction in Austronesian languages as they determine the development of undergoer voice constructions. This study provides opportunities for similar research with data sources other than literary works to further strengthen conclusions regarding the disappearance of intransitive undergoer constructions in the Javanese language.

Keywords:

• Undergoer voice
• Javanese
• first personal pronoun
• diachronic
• Austronesia

Reviewing Editor:

• Jeroen van de Weijer, College of International Studies, Shenzhen University, China

SUPJECTS:

• Syntax & Linguistic Structure
• Historical & Comparative Linguistics

1. Introduction

This article aims to investigate the development of undergoer voice constructions from Old Javanese in the 10^th century to Javanese in the 21^st century. The term undergoer voice is commonly used to describe passive constructions in the Austronesian language family (Ross, 2002; Himmelmann, 2002; Hunter, 2018; Adricula, 2022). Besides undergoer voice, the term patient voice is also common with the difference lying in the semantic role of the argument. Patient voice relates to the semantic role that includes the patient’s focus, while the semantic role of undergoer voice includes not only the patient’s role, but also the target, goal, and object (Hunter, 2018). Other than patient voice and undergoer voice, there is also a focus system in the research of voice in the Austronesian family. The focus system is usually applied for languages that belong to the Filipino type (Himmelmann, 2002; Reid and Liao, 2004). According to Himmelmann (2002), Latrouite (2011), and Adricula (2022), the focus system relates to certain affixes used as voice markers morphologically and is used to distinguish between active and passive voice in English. However, according to these three studies, Austronesian language researchers prefer using voice to using the term focus in explaining active and passive in Austronesian languages. For the purpose of the research, the term undergoer voice is used in this article over other terms because it does not merely discuss patient-focused semantic roles. Further, the undergoer voice construction is analyzed by looking at the affix marker and the number of arguments available.

Based on the undergoer voice construction, there are two language types in the Austronesian language family. Himmelmann mentions that the two types of Austronesian languages are Filipino and Indonesian language type (2002, p 8). The aspect that distinguishes these two language types lie in their morphophonological features. In the Indonesian language type, person markers in a form of preposed clitic for the first and second person, and a combination of affixes with the suffix -i are available (Himmelmann, 2002; Ross, 2002). Meanwhile, in the Filipino language type, there are affix markers in and there are markers on nouns and pronouns as arguments for agents or undergoers (Latrouite, 2011). Himmelmann (2002, p. 8) explains that these differences are practical for investigating the Austronesian languages typology.

Morphologically, Old Javanese has affix markers in undergoer voice construction, such as the affix -in-. As explained by Himmelmann (2002), Latrouite (2011), and Adricula (2022), the affix -in- is one of the characteristics of the Filipino language type in the undergoer voice construction. Research on the affix -in- as an undergoer voice marker in Javanese has been mentioned by Zoetmulder (1983), Ogloblin (2008), and Adelaar (2011). According to them, undergoer voice construction with the affix marker -in- has not experienced changes from Old Javanese to Modern Javanese in literary texts. The following is an example of undergoer voice construction with a formal marker in the form of a verbal affix -in- in the Adiparwa text (10^th century).

1. Yatika in-alap de ning daitya

That (the holy water) UV-took by the demon

‘The holy water is taken by the demon’

1. In-alap mami ta ikang lĕmbu

UV-took =1.SG that buffalo

‘That buffalo is taken by me’

In example (a), the agent argument in the form of noun daitya ‘demon’ is present and preceded by the preposition de. Apart from nouns, agent arguments in the form of personal pronouns can also appear preceded by the preposition de. However, the personal pronoun forms that appear are bound forms, such as de mami and dengku. Unlike example (a), in example (b) the agent argument in the form of the first personal pronoun mami comes after the verb. The agent argument is on the right and is attached to the verb with the affix -in-. The personal pronoun mami comes in the form of an enclitic and is attached to the verb with the affix -in-. Although mami seems not attached to the verb, it is a bound form of the personal pronoun kami which cannot stand alone. In addition to first-person personal pronouns like mami, bound forms of second-person (-mu, -nta) and third-person (-ya, -nira/-ira) personal pronouns can also attach to verbs marked with the affix -in-. The agent argument attached to such verbs can only be in the form of original bound personal pronouns and cannot be nouns.

In Javanese texts from the 15^th century to the present day, the construction of undergoer voice with affix -in- with an agent argument in the form of a person pronoun attached to a passive verb is not found. The undergoer voice construction found involves verbs marked with the affix -in- and agent arguments in the form of nouns preceded by the preposition de. Besides, undergoer voice constructions with the affix -in-, in the text from the 20^th century, there are also undergoer voice constructions without the affix -in-. The following are examples of undergoer voice constructions in the texts from the 17^th and 20^th centuries.

1. Yata in-adĕg-akĕn dening watĕk dewata

Itu (peak of Mahameru mountain) UV-stand-aken by group of God

‘The peak of Mahameru mountain was stand by the group of God’

1. Kadang Pandawa badhe kula sirnakaken

Mr. Pandawa will 1.SG be destroyed

‘Mr. Pandawa will be destroyed by me’

Example (c) illustrates an undergoer voice construction with a verb marked with the affix -in- and an agent argument in the form of the noun watĕk dewata preceded by the preposition dening. No first and second-person personal pronouns as agent arguments preceded by prepositions are found in the data. It is found that only the use of the third-person personal pronoun as the agent argument in construction (c) is identified, that is dening dheweke. Meanwhile, example (d) represents an undergoer voice construction without the affix marker -in-. In this construction, the agent argument in the form of the first-person personal pronoun kula appears before the verb. This type of undergoer voice constructions are only present with agent arguments in the form of first-person and second-person personal pronouns. Through this feature, it is known that there are differences in the construction of undergoer voice with the affix marker -in- in Javanese texts from the 10^th century compared to the same construction in Javanese language from the 15^th century to the present day. The undergoer voice constructions (a) and (c) remain unchanged, while undergoer voice construction (b) undergoes a change. The difference in the changed construction is triggered by the first-person personal pronoun as the agent argument.

Research on Javanese undergoer voice construction had been conducted by Poedjosoedarmo (2002), Nurhayani (2015), Puspitorini & Fitriana (2017), Surip & Mulyadi, (2018), Malihah (2018), and Asridayani and Suryani (2020). However, among these six studies, only Poedjosoedarmo (2002) and Puspitorini & Fitriana (2017) did research discuss the development of undergoer voice construction in Javanese. Both studies looked at the undergoer voice construction, not only based on verbal affix markers but also on the structure of the argument in relation to verbs. Poedjosoedarmo (2002) in his research concludes that there is a change in word order pattern in undergoer voice construction from Old Javanese to Modern Javanese. According to him, the undergoer voice construction in Old Javanese has the pattern of verb—enclitic agent—patient (V-A-P), then changes to patient—proclitic agent—verb (P-A-V). Poedjosoedarmo (2002) also explains that Old Javanese has the same characteristics as the Filipino language type, namely the presence of the affix marker -in- in the undergoer voice construction without the presence of a proclitic pronoun.

Similar with Poedjosoedarmo (2002), Puspitorini & Fitriana (2017) also presents the differences of undergoer voice construction in Old Javanese and also in Modern Javanese based on the position of the patient argument (P) and agent argument (A) in relation t the verb (V). Puspitorini and Fitriana’s (2017) research shows that there are three patterns of argument structure in the construction of undergoer voice in Old Javanese and Modern Javanese. The three patterns of argument order in Old Javanese are P-V-A, V-A-P, and V-P-A, while the three patterns of argument order in Modern Javanese are P-V-A, P-A-V, and A-V-P. Based on these patterns, there are differences in the patterns found by Poedjosoedarmo (2002) and Puspitorini & Fitriana (2017), not only based on the argument order patterns but also based on the number of argument patterns found.

The difference between Poedjosoedarmo’s (2002) findings and Puspitorini & Fitriana’s (2017) findings was influenced by the focus of the research. Poedjoseodarmo (2002) did not look at the presence of undergoer voice construction that was not required to present agent arguments and construction that was not required to present agent arguments. The difference in findings was also caused by the different data used. In his research, Poedjosoedarmo (2002) used data that were acquired from an Old Javanese grammar textbook (Zoetmulder & Poedjawijatna, 1992) and he did not use the original text. The grammar book cannot be used as a primary data source because the data included in the textbook does not represent the use of a whole text. Meanwhile, Puspitorini & Fitriana’s (2017) research used data sources in the form of texts, such as the Adiparwa text which represents the use of Old Javanese language in the 10^th century, the Pararaton text that represents Javanese language in the 15^th century, and novel texts from the 20^th to the 21^st century. Therefore, Puspitorini & Fitriana’s (2017) research shows the presence of more patterns of Javanese undergoer voice construction compared to Poedjosoedarmo’s (2002) research.

In addition to the use of appropriate data sources in the study of the development of undergoer voice constructions, there are two important considerations. Firstly, the use of verb markers in the form of affixes is consistently discussed as a main feature in undergoer voice constructions. However, undergoer voice constructions without verb markers receive less attention. This characteristic is related to the second aspect, which is the agent argument. Agent arguments are often overlooked in research on undergoer voice constructions (see Riesberg & Primus, 2015). Agent arguments can be oblique or core arguments. Oblique agent arguments have been extensively discussed in previous research, while core agent arguments receive less attention. However, it is these core agent arguments that are present in undergoer voice constructions without affix markers. The first-person personal pronouns that can function as core agent arguments influence the structure of undergoer voice constructions.

The analysis of undergoer voice constructions in Javanese language requires not only examining verb markers with specific affixes but also looking at its agent argument form, whether it is a noun or a pronoun. If we only focus on verb markers, the undergoer voice construction does not change from the 10^th century until now (see Zoetmulder, 1983; Ogloblin, 2008; Adelaar, 2011). However, initial observations of diachronic data in this research reveal the development of undergoer voice constructions with first-person personal pronouns as agent arguments (see example b). Understanding how the undergoer voice construction develops in Javanese becomes the focus of this research. First and second-person personal pronouns share similarities, while third-person personal pronouns exhibit a development similar to nouns (see example c). Agent arguments in the form of nouns generally do not undergo changes. Since first and second-person personal pronouns share similarities, this research focuses on first-person personal pronouns as agent arguments. The use of first-person personal pronouns can represent the usage of second-person personal pronouns.

Building upon this, the aim of this research is to identify the development of undergoer voice constructions, focusing on verbs both with markers and without markers, as well as agent arguments in the form of pronouns. This research limits the scope of the agent argument specifically to the first-person pronoun because the first-person personal pronoun as the agent is the trigger for the changes in the undergoer voice construction. This research also focuses on the position of the agent argument in relation to the verb and does not discuss the position of the patient’s argument in relation to the verb. The patient argument in this study is not discussed because the patient argument that functions as the subject can be to the left or right of the verb and might disappear due to certain contexts. In regard to the importance of data scope, this research uses data sources in the form of literary works namely prose and poetry dating from the 10^th to the 21^st century. In terms of Javanese texts from the 20^th to the 21^st century, this research uses data in the form of Javanese dialects of Jogja-Solo, Banyumas, and Banyuwangi. The Banyumas and Banyuwangi dialects were not used as data sources by Poedjosoedarmo (2002) and Puspitorini & Fitriana (2017). Both studies only used the Jogja-Solo dialect as their data source. The first-person pronoun varies based on regional variations. One of the characteristics of Javanese dialect can be observed from the variation of personal pronouns. Therefore, the use of Javanese dialect as the data is expected to show the development of undergoer voice construction in current Javanese. Data sources from a wider time span are required to sight the development of undergoer voice construction in this research. Based on the results of previous research conducted by Poedjasoedarmo and Puspitorini & Fitriana, it is known that the number of data sources and different time spans result in different research conclusions. Therefore, to obtain a description of the development of undergoer voice construction, a large number of data sources with a specific time span are crucial.

2. Methods

This research uses a qualitative method by utilizing some of the basic features of corpus linguistics. According to Croker and Heigham (2009), qualitative research is a type of research that utilizes a primary collection of textual data and tests it using interpretative analysis. Rasinger, as cited in Litosseliti (2010, p. 51), also explains that qualitative analysis is conducted to analyze text by considering its characteristics or qualities. In this research, the analysis is performed using data gathered from various texts by paying attention to each characteristic of the texts.

The use of corpus linguistics in this research is applied to the stages of data collection and processing. Corpus linguistics is an empirical method in descriptive linguistic analysis that utilizes examples from real-life data stored in corpora (Crystal, 1992, and Jackson, 2007, as cited in Cheng 2011). According to Mair (2012), there are three reasons for using corpora in linguistic research. These reasons are: (1) corpora can assist in data retrieval, making it more precise, systematic, and efficient; (2) the use of corpora can demonstrate authentic language data syntactically and within the context of discourse; (3) corpora can overcome issues related to the accuracy of usage frequency.

The data sources used are written literary texts (prose and poetry) from the 10^th to 21^st century varied both temporally and regionally. These texts include the Jogja-Solo, Banyuwangi, and Banyumas dialects. Linguistically, most of the data sources used in this research are manuscript-transmitted texts which include literary works from the 9^th to 18^th century. The term ‘manuscript-transmitted texts’ in this research refers to text that has undergone transliteration or transference from one script to another (Robson, 1998). In this case, the text used is the result of transliteration from Javanese script to Latin script. This research employs manuscript-transmitted texts published in diplomatic editions. According to Robson (1998), diplomatic editions are used to present the text as it is and not alter what is written in the manuscript. Each letter in the Javanese script is represented by a different letter, making it identical to the original manuscript. With a diplomatic edition, linguistic researchers can discover that the Latin-script accurately reflects the writing in the Javanese script.

In addition to using data sources originating from 10^th to 21^st century texts, this research also utilizes corpus features available at http://sealang.net/ojed/ and http://sealang.net/java/dictionary.htm. These two websites not only function as dictionaries but also provide corpus features that can be utilized to obtain data on the use of undergoer voice construction in sentences in Old Javanese and Javanese. Furthermore, the data sources originating from the 20^th to 21^st century are modern literary works written in the form of novels, short stories, and poetry. The following are the texts used as data sources along with the number of words in each period.

Table 1 presents the data sources used in this research that contain the information of the period and the number of words in each century. These texts are taken to represent a particular period without limitation on the number of words from each period.

Table 1. Data sources and the number of words.

No	Texts	Century	Number of Words
1	Adiparwa	10	50.877
2	Arjunawiwaha	11	12.027
3	Tantri Kamandaka; Bharatayuddha	12	44.172
4	Arjunawijaya	14	19.608
5	Gita Sinangsaya	15	7.419
6	Sri Tanjung, Pararaton, Het Boek van Bonang	16	33.002
7	Bismaprawa and Tantu Panggelaran	17	32.139
8	Babad Kraton	18	258.625
9	Literary Text (Sastra.org)	19	2.789
10	Javanese Corpus (Jogja-Solo)	20–21	2.252.307
11	Banyumas Dialect	20–21	183.941
12	Banyuwangi Dialect	20–21	62.401
TOTAL			2.959.307

All data sources used in this study are arranged in plain text (txt) format. Subsequently, the data is processed using Antconc by utilizing its basic features such as concordance. According to McEnery and Hardie (2012) concordancer allows to search a corpus and retrieve from it a specific sequence of characters of any length—perhaps a word, part of a word, or a phrase. Concordance is used to investigate the syntactic structure of the undergoer voice construction with the agent in the form of the first-person pronoun. In the processing stage, the first-person pronouns found in each text are used as a node in a concordance to see their use in the undergoer voice construction of Javanese from the 10^th to the 21^st century. The observation results on the concordance are then grouped together based on the form of the first-person pronoun that becomes the agent argument and based on the markers of the undergoer voice found. Based on the form, personal pronouns are grouped into two, namely free and bound form. Meanwhile, based on the markers, the undergoer voice construction is grouped based on the affix used with the argument of the agent in the form of first-person pronoun. A concordance search is carried out in each period. The use of first-person pronoun as the agent argument in undergoer voice construction is also grouped based on that period. The data analysis is conducted in stages. First, the analysis involves identifying the position of the agent argument against verbs with undergoer voice affixes. In this stage, the data are analyzed by considering the affixes as undergoer voice markers and the types of first-person personal pronouns (free forms or bound forms) as agent arguments. Subsequent analysis is then carried out on undergoer voice constructions that come without undergoer voice markers. In presenting the data, this research complies with procedures of diacritics of the Old Javanese-Indonesian Dictionary (Zoetmulder, 2006) by adjusting the form of ŋ to become ng, such as ŋhulun into nghulun.

3. Result and discussion

Old Javanese and Javanese have undergoer voice construction with and without affix markers. In the undergoer voice construction with affix markers, Old Javanese and Javanese have different markers. The undergoer voice markers in Old Javanese are the affixes ka- and -in- (Zoetmulder and Poedjawijatna, 1992; Hunter 2018; van der Molen 2015). The difference between the two affixes lies in their ability to present agent arguments. The affix ka- has low characteristics in presenting agents, whereas the affix -in- is a very agentive affix (Puspitorini, 2015; Hunter, 2018). Meanwhile, the undergoer voice markers in Javanese are the affixes di-, dipun, tak-, dak-, kok-, ka-, and ke- (Ogloblin, 2005a; Adelaar, 2011; Nurhayani, 2015). The differences in affix are not only based on language varieties, such as ngoko (di-) and krama (dipun-), but also on the agent argument, namely no- for the first-person agent argument, and kok- for the second-person agent argument. The markers in the form of affixes in undergoer voice construction can be observed in Table 2 below.

Table 2. Undergoer voice affixes.

Old Javanese Affixes	Javanese Affixes
-in- ka-	di- dipun tak-/ dak- kok- ka-/ ke-

Although there are several affix markers for the undergoer voice construction, not all affixes in Javanese are discussed. Only affixes that present the agent argument in the form of first-person pronouns are discussed in this research. In Javanese, the undergoer voice marker affix which presents the agent in the form of the first person is tak-/dak-. In contrast to construction with affix markers, the undergoer voice construction without affix markers usually comes with the first-person pronoun that serves as the agent argument. In Old Javanese and Javanese, the presence of first-person pronouns is different because it is determined by its development in each period.

The development of undergoer voice construction in Javanese since the 10^th century until now has been exhibited in surviving undergoer voice constructions (continuous) although some undergoer voice have been lost (discontinuous). The development of undergoer voice construction is seen based on the changes that occur in the presence of the agent argument in the form of the first-person pronoun. Based on its development, the undergoer voice construction with affix markers undergoes a change in presenting the agent argument. The lost undergoer voice construction is a construction with an affix marker with the presence of the agent argument in the form of the first-person pronoun. Whereas the undergoer voice construction that tends to remain the same and does not undergo change by the presence of the argument of the first-person pronoun agent is the one without affix markers. The difference between the undergoer voice construction with and without affix markers lies not only in its ability in presenting the agent argument, but also in relation to the position of the agent argument (A) in relation to the verb (V). In the undergoer construction with an affix marker, the agent argument comes after the verb (V-A), whereas in the undergoer voice construction without an affix marker the agent argument precedes the verb (A-V).

3.1. Undergoer voice construction with affix markers

In Old Javanese, the undergoer voice construction with markers is indicated by the presence of the affixes -in- and ka-. The difference between the two affixes lies in their feature of presenting the agent, as seen in the example below.

1. Huwus ka-hañjur^V de ni nghulun^A krodhabahni (Ad-10)

already UV-carry by I anger

‘My anger has been carried by me’

1. sirekana p-in-et^V mami^A (RY-9)

he UV-look I

‘It’s him I am looking for’

The undergoer voice construction (1) comes with the affix ka- as a marker. In this construction, the agent argument in the form of the first-person pronoun nghulun appears to the right of the verb and is preceded by the preposition, de. The preposition de indicates that the presence of an agent argument in the form of a personal pronoun is not required or is optional/oblique. The presence of this oblique agent argument is marked by the preposition de ‘by’. Accordingly, (Ross, 2002: 28), the oblique is usually characterized by a special structure that is present in the form of ad-positional phrases (prepositional or postpositional). The presence of an oblique agent argument is referred to as an intransitive undergoer voice construction. This construction is intransitive because it only has one argument that must be present, namely krodhabahni ‘anger’.

Meanwhile, undergoer voice (2) comes with an affix -in- as a marker. In this construction, the agent argument in the form of a personal pronoun is present as clitic and is attached directly to the verb without being preceded by a preposition. An agent argument that attaches directly to verbs with passive markers without being preceded by a preposition is a semi-core argument. According to Arka (2005), the agent argument in the form of an enclitic pronoun attached to a verb marked with undergoer voice is a semi-core argument. Arka (2005, p. 39) provides an example of an enclitic personal pronoun (-nya) attached to a verb with a passive marker di- in Indonesian. According to him, enclitic -nya is a semi-core agent argument. This undergoer voice construction is also called the semi-transitive undergoer voice construction. This construction is semi-transitive because the agent argument is semi-core.

The construction of the undergoer voice marked with ka- and the construction of the undergoer voice marked with -in- have different developments. The following describes the development of the undergoer voice construction with ka- and -in- markers with the agent argument in the form of the first-person pronoun.

3.1.1. Affix ka-

In the undergoer voice construction marked with the affix ka-, the agent argument in the form of the first-person pronoun is always to the right of the verb (V-A). In addition, the agent argument is also always preceded by the preposition (de/dening). The agent argument is in the form of the first person-pronoun that appears in the undergoer voice construction marked with the affix ka-, namely nghulun, ngwang, sun, and enclitic -ku and -mami. The agent argument in the form of the pronoun nghulun appears in the undergoer voice construction in the text from the 10^th to 12^th century. The agent argument in the form of the pronoun sun is present in the undergoer voice construction in the text from the 16^th to 17^th century. Below are examples of the undergoer voice construction with the agent arguments in the form of personal pronoun nghulun and sun.

1. ikang tribhuwanāmaṇḍala ka-waçākna^V de ni nghulun kālih^A (Ad-10)

Those three worlds UV-(will) rule by I two

‘Those three worlds will be ruled by the two of us’

1. ka-tinghalan^V rahasya nipun de ning sun^A (Par-16)

UV-see secret her/his by I

‘The secret is seen by me’

In the two examples above, the agents’ arguments (nghulun and sun) are preceded by the preposition de/dening with a verb marked by the affix ka-. According to Hunter (1988) and Puspitorini (2015), affix ka- has low features in presenting agents. Therefore, in the construction of undergoer voice with a verb marked with the affix ka-, the agent argument is always preceded by the preposition de. Due to its low feature in presenting the agent, the agent argument in undergoer voice construction marked with the affix ka- is not always required to be present. The agent argument preceded by the preposition de is optional (Puspitorini, 2015, p. 160). The undergoer voice construction marked with ka- is intransitive because the agent argument is optional. This construction only requires the presence of one argument, namely the target argument (ikang tribhuwanāmaṇḍala and rahasyanipun).

In the undergoer voice construction (3) and (4), there is ligature particle ni between the agents (nghulun and sun) and the preposition de. This ligature particle serves to connect the nouns on the left with those on the right. According to Ogloblin (2000, p. 183), agents in the form of nouns come with a combination of ligature de and ni/ning. In this case, de and the pronouns nghulun and sun retain the nominal characteristics which are indicated by the presence of ligature ni. Pronoun nghulun is the first-person pronoun derived from nouns (Fitriana, 2022). In Old Javanese, de means ‘cause, reason; action, manner’ (Ogloblin, 2000; Adelaar, 2009). However, de in the undergoer voice construction functions as a preposition to mark the agent (Ogloblin, 2000, p. 181). During its development, the nominal feature of de is lost, while its function as a preposition has survived to this day.

Apart from agents in the form of personal pronouns nghulun and sun, the undergoer voice construction marked with ka- also present as an agent argument in the form of enclitic personal pronouns -ku and mami. Both pronouns appear as agent arguments in undergoer voice construction of the 10^th to 14^th century, as seen in the examples below.

1. ikang sukha ka-gawe^V de mami^A (Ad-10)

That happiness UV-make by I

‘That happiness is made by me’

1. ka-ton^V pĕjah-mu de-ngku^A (AW-11)

UV-see death-you by-I

‘Your death is seen by me’

In examples (5) and (6), the agent arguments -mami and -ku are attached to the preposition de ‘by’ with a verb marked with affix ka-. The agent arguments mami and –ku which are directly attached to the verb marked with ka- are not found. This is closely associated with the low characteristics of the affix ka- in presenting the agent. Therefore, the agent argument in the form of enclitics mami and -ku cannot be attached to verbs marked with affix ka-. The enclitics mami and -ku can be attached to verbs marked with the affix -in- because this affix is very agentive. The agent argument mami and -ku are optional because they come before the preposition de. This construction is intransitive because it only requires the presence of one argument, namely the target argument (ikang sukha and pĕjahmu) and they can be positioned to the right or left of the verb. The undergoer voice construction with a V-A pattern and with agents mami and –ku attached to the preposition de is only found in texts from the 10^th to 14^th century and is not found in current Javanese texts.

Based on the order of the arguments (example 3–6), the agent arguments are always to the right of the verb (V-A). Although the agent argument is not required to be present, it cannot be positioned to the left of the verb, and remains to the right of the verb. This is different from the mandatory argument (target argument) in the undergoer voice construction with the affix marker ka- where the argument can be to the right or left of the verb.

In the texts from the 18^th to 21^st century, the affix ka- was still used as an undergoer voice marker. Agent arguments are also still not required to be present. If there is an agent argument present in the undergoer voice construction which is marked with the affix ka-, that argument is not the first-person pronoun. There is no undergoer voice construction marked by the affix ka- which presents the agent argument in the form of first-person pronouns in texts from the 18^th to the 21^st century.

3.1.2. Affix -in-

The affix -in- is an affix that marks the undergoer voice construction in Old Javanese. According to Poedjosoedarmo (2002), Old Javanese has a similar type of undergoer voice construction as Filipino. This is because the undergoer voice construction in Old Javanese presents a verb marked with the affix -in- rather than presenting a proclitic.

In the undergoer voice construction marked with the affix -in-, the agent argument in the form of first-person pronoun appears in the enclitic forms –ku and mami. The other first-person pronouns, such as nghulun, sun, kula are not found serving as an agent argument in the undergoer voice construction with the affix marker -in-. According to Ogloblin (2000, p. 182–183) in the undergoer voice construction, the first-person pronoun agent is usually present in the enclitic form -ku/-mami. The agent may be attached to the preposition de directly without being followed by the ligature particle ni and also be attached to verbs marked with affix -in-.

1. ya h-in-ilyakĕn^Vku^A ring bañu (Ad-10)

they UV-wash away-I in water (river)

‘they were washed away (by) me in the river’

1. satwa ika in-ĕntasakĕn^V mami^A (TK-12)

animal that UV-taken out of the water I

‘the animal was out of the water (by) me’

In the undergoer voice constructions (7) and (8), the agent arguments in the form of enclitics -ku and -mami are to the right of the verb (V-A) and are attached directly to the verb marked with -in- without being preceded by a preposition. The agent argument in an enclitic form attached to a verb marked with affix -in- is semi-core. This semi-core feature of the agent argument happens because it is not preceded by a preposition. Further, the undergoer voice constructions (7) and (8) are also semi-intransitive because they only require one argument, namely the patient argument, with a semi-core agent argument.

Besides being attached directly to verbs marked with the affix -in-, enclitic pronouns -ku and mami may also be attached to verbs marked with affixes -ěn and -akna.

1. rarah-ěn^V ku^A kamu wulik-ěn iṅ wanāśrama (Awj-14)

hunt-UV I you search-UV into forest of hermitage

‘you’ll be hunted by me (and) searched into the forest of hermitage’

1. tinggal-akna^Vngku^A kita (Ad-10)

leave-UV I you

‘I will leave you’

1. pintan-ěn^Vmami^A ikang lěmbu (Ad-10)

ask for-UV I that buffalo

‘that the buffalo I ask for’

1. wӧr-akna^V mami^A kita kabeh. (Ad-10)

fly-UV I you all

‘I will fly you all’

Examples (9) to (12) are undergoer voice constructions with passive markers -ěn, and -akna with agent arguments in the form of enclitics -ku and mami attached directly to the verb. According to Ogloblin (2005b) and Hunter (2018), the suffixes –ĕn and -akna are suffixes that are present in the irrealis mode. In Proto-Austronesian, the affix –in- is the undergoer voice marker for the past tense, while the affix –ĕn is a marker of the undergoer voice marker for the non-past tense (Blust, 2002, p. 59). In the undergoer voice construction, the suffix -akna is not accompanied by the affix -in-. Similar with the construction (7) and (8), agent arguments in the form of personal pronoun enclitics in constructions (9) to (12) are always to the right of the verb marked with an affix (V-A). This construction is also semi-intransitive because the agent argument is semi-core.

Besides being attached directly to verbs marked with -ěn and -akna, enclitics -ku and mami can also be attached to prepositions with verbs marked by the affix -akna.

1. Ary-akna^V ko de-ngku^A (Ad-10)

leave-UV you by-I

‘you will not be left by me’

1. ikang wwang tan wineh ĕntas-akĕna^V de mami^A (TK-12)

that person not be allowed take out the water-UV by I

‘that person was not allowed (to be) taken out of the water by me’

In examples (13) and (14), the undergoer voice construction comes with an affix marker -anna. In this construction, the agent argument is in the form of enclitic personal pronouns -ku and mami which appear to the right of the verb (V-A) by attaching to the preposition de. In the undergoer voice construction, the agent argument is optional because it is preceded by the preposition de. This construction is intransitive because it only requires the presence of one argument, namely the patient argument which can be positioned to the left or right of the verb (ko and ikang wwang). While the agent arguments in constructions (13) and (14) are arguments that are not required to be present.

The undergoer voice construction with agent arguments in the form of enclitic personal pronouns -ku and mami attached to the right of the passive verb marked with a passive affix (V-A) or attached to the preposition de is only found in Old Javanese texts from the 10^th to 14^th century. The personal pronoun enclitics attached to prepositions and verbs marked with affixes are not found in the current Javanese. In the current Javanese, the passive affix -in- is still used in literary texts that have archaic forms (Ogloblin, 2008). However, the undergoer voice construction with the agent arguments –ku and mami attached to verbs with the affix –in-, -ěn, and -akna are not found in contemporary Javanese. The absence of enclitics –ku and mami in the undergoer voice construction with a V-A pattern is related to the development of personal pronouns in Javanese. According to Uhlenbeck (1968, p. 473) enclitic personal pronouns attached to verbs are only found in Old Javanese and are not available in contemporary Javanese. This is different from the bound forms –me and mami which are attached to nouns. Both are enclitics which are used to express the possessive (genitive) case, which are found in both everyday Javanese and in archaic literary texts.

The loss of -ku and -mami enclitics in the undergoer voice construction cannot be resolved in this article. It requires bigger data and more evidence to explain this matter. However, when viewed from early observations, the loss of the enclitic -ku and -mami attached to verbs with the affix -in- in the texts from the 15th century coincides with the presence of the affix den-. The affix den- is the affix marker of undergoer voice which began to appear in texts from the 15^th century. According to Zoetmulder (2006), den is used as a prefix in the undergoer voice construction. The prefix den has the same function as depun, demu, denta, de nira and dera. In the Old Javanese dictionary, den is included in the subentry of de which means action, ways, cause, and indicates the presence of an agent (by).

3.2. Undergoer voice construction without affix markers

The undergoer voice construction in Javanese is not only marked by affix markers but may also appear without affix markers. The absence of a marker in the form of an undergoer voice affix is replaced by the presence of a personal pronoun, such as the first-person pronoun. This construction is transitive because the agent argument in the form of the first-person pronoun must be present in the undergoer voice construction. Based on the language type, the construction is included in the Indonesian type. Several languages have the undergoer voice construction that belongs to the Indonesian type, such as Indonesian (Purwo, 1988; Umar, 2003), Pendau (Quick, 2002), and Javanese (Poedjosoedarmo, 2002).

According to Poedjosoedarmo (2002) there is a pattern change in the undergoer voice construction from Old Javanese to Javanese. According to him, in Javanese the marker for the agent argument appears in proclitic form. Proclitic appears as the development from Old Javanese to Modern Javanese. The order of argument patterns in Old Javanese changes in Modern Javanese, namely Verb (undergoer voice) + Agent (enclitic) + Subject becomes Subject + Agent (Proclitic) + Verb (undergoer voice). Based on observations made by Poedjosoedarmo (2002) it appears that there is a change in the order of the agent arguments in relation to the verb, namely from verb-agent (V-A) to agent-verb (A-V).

This research shows different findings from the explanation by Poedjosoedarmo (2002). The undergoer voice construction with A-V pattern does not only exist in Javanese, but it already existed in Old Javanese in the 10^th century. In this construction, the first personal pronoun as the agent argument is to the left of the verb, so the argument sequence pattern formed between the agent argument (A) and the verb (V) is A-V. The difference between Poedjoseodarmo’s findings and this research’s findings is because Poedjosoedarmo only sees affixes as an undergoer voice marker.

The undergoer voice construction with the A-V pattern is not the same as the active construction with the A-V pattern. The relationship between the agent and the verb in the undergoer voice construction shows a rigid structure and other elements cannot be inserted. Meanwhile, the relation between agents and verbs in active construction is more tenuous, as seen in the example below.

1. Nghulun^A mang-atĕr-akĕn^V i sang Pāṇḍawa (Ad-10)

I to escortPāṇḍawa

‘I escort Pāṇḍawa’

1. Mang-rĕngӧ^V nghulun^A i bhagawān Kaṇwa (Ad-10)

Hear I about bhagawān Kaṇwa

‘I heard about bhagawān Kaṇwa’

Examples (15) and (16) are active constructions. In both examples, the verb is present with active markers in the form of prefixes maN-akĕn and maN-. In example (15), the agent argument in the form of the personal pronoun nghulun is to the left of the verb (A-V). In this example, the subject is the pronoun nghulun, not sang Pandawa. This is proven by the presence of the particle i which precedes sang Pandawa. The i particle serves to mark that the word following it is not the subject. According to Puspitorini (2015, p. 145), the identification of the subject is made through the impossibility of the subject being preceded by a particle or preposition i. The verb in the active construction comes with the active marker prefix maN- followed by the suffix -akĕn. The sequence A-V in the active construction forms a clause with a Subject-Predicate (S-P) pattern. Therefore, the relationship between the agent (nghulun) functioning as the subject, and the verb (mangatĕrakĕn) functioning as the predicate, is distant. This is proven by the ability to move the position of the subject to the right or left of the verb. In example (16), the agent argument is the first personal pronoun nghulun which functions as the subject. The word bhagawan Kanwa does not function as the subject because it is preceded by the particle i. In example (16) the agent argument that functions as the subject is to the right of the verb. The versatile movement of the subject indicates a distant (flexible) relationship between the subject and the verb. This is also supported by Puspitorini (2015, p. 146) who explains that based on its position in a construction in Old Javanese, the subject can be positioned to the right or left of the verb.

The distant relationship between agents and verbs in active construction does not occur in undergoer voice construction. In the undergoer voice construction without affix markers, the relationship between the agent argument and the verb is rigid, as seen in the example below.

1. Nghulun^A adĕg-akĕn^V ratu rasika (Ad-10)

I be crowned king he

‘He is crowned as a king by me’

1. * adĕg-akĕn^V nghulun^A ratu rasika (Ad-10)

Be made I king he

‘I made him king’

In example (17) the agent argument in the form of the pronoun nghulun is to the left of the verb adĕgakĕn. Syntactically, the personal pronoun nghulun is tied to the verb. Therefore, the pronoun nghulun does not function as the subject. The subject of the construction is rasika. The A-V sequence in the active construction (example 15) forms a Subject-Predicate (S-P) clause, while the A-V sequence in the undergoer voice construction (example 17) does not form a Subject-Predicate (S-P) pattern clause. The sequence A-V in the undergoer voice construction does not form a Subject-Predicate (S-P) pattern clause because the agent (nghulun) in example (17) does not function as the subject.

In example (17) there is only the suffix -akĕn of the verb without an undergoer voice marker. According to Zoetmulder and Poedjawijatna (1992) and Molen (2015) the suffix -akĕn cannot simply be attached to the root. The suffix -akĕn must rather be added to the root with an active marker affix (a-/ma-, aN-/maN-, -um-) or a passive marker of undergoer voice (-in-). The agent argument (nghulun) in the undergoer voice construction does not function as the subject but appears as a clitic and replaces the passive formal marker -in-. Therefore, the relationship between agent (nghulun) and verb (adĕgakĕn) in the undergoer voice construction with an A-V pattern is rigid. The relationship between agents and verbs in the undergoer voice constructions is rigid because the agent’s arguments are clitic and cannot be moved to the right of the verb (see example 18).

In the undergoer voice construction without a marker, the agent argument is always attached to the left of the verb and cannot be positioned to the right of the verb. Semantically, the undergoer voice construction without a marker is transitive because the agent argument in the construction is mandatory and cannot be omitted. While syntactically, the construction is intransitive because pronouns are bound and attached to verbs. Even though it looks like a free form, the pronoun nghulun is syntactically bound to the verb. Therefore, the agent argument in the form of a personal pronoun does not serve the syntactic function independently.

The development of undergoer voice construction without a marker has been going on since the 10^th century to the present day. The development of the undergoer voice construction can be distinguished based on the agent argument, namely in the form of first-person pronoun or first-person pronoun deictic particles. The following describes the development of the undergoer voice construction without affix markers based on the agent argument.

3.2.1. First-person pronouns

In the undergoer voice construction without affix markers, the agent argument in the form of the first-person pronoun is always to the left of the verb. The agent arguments for the undergoer voice construction are nghulun, ngong, sun, and kula. The differences in personal pronouns as agent arguments depend on the presence of personal pronouns in each period. The agent argument in the form of the pronoun nghulun is present in the undergoer voice construction from the 10^th to 14^th century. The agent argument in the form of the pronoun sun appears in the undergoer voice construction from the 15^th to the 21^st century. The agent argument in the form of pronoun ngong appears in the undergoer voice construction from the 16^th century. Meanwhile, the agent argument in the form of the pronoun kula appears in the undergoer voice construction from the 18^th to the 21^st century.

Below is the example of the undergoer voice construction with the agent argument in the form of the personal pronoun nghulun.

1. mah prang! nghulun^A pějahi^V yan kamu wīra rodra. (Awj-14)

in war I kill if you fighter cruel

‘In war, if you are a cruel fighter, I will kill you’

In example (19), the agent argument in the form of personal pronoun nghulun is to the left of the verb pejahi. The verb in the undergoer voice construction is marked only by suffix -i and is not marked by passive affixes. The pronoun nghulun is syntactically bound with the verb pejahi and functions as a passivity marker. The pronoun nghulun that becomes the agent argument for the undergoer voice construction with the A-V pattern appears in texts from the 10^th to 14^th century. Nevertheless, there is also another pronoun form, ulun, which appears in texts from the 18^th century that serves as the agent argument for the undergoer voice construction with an A-V pattern.

1. Babarkatan sun pundhut, ulun^A bekta^V dhateng Matawis (Bkr-18)

blessing I take I bring to Matawis

‘I take the blessing; I bring it to Matawis.

In example (20) the pronoun ulun becomes the agent argument positioned on the left and is syntactically bound to the verb bekta in the undergoer voice construction with an A-V pattern. The pronoun ulun is a transformation of the pronoun nghulun (Fitriana, 2022). Even though the pronoun as an agent argument undergoes form transformation, the construction of the undergoer voice with the A-V pattern does not change.

In the texts from the 15^th century, the personal pronoun that appeared as an agent argument is the pronoun sun. The absence of the pronoun nghulun as an agent argument in the undergoer voice construction without a marker is related to the development of the first-person pronoun in Javanese. In the texts from the 15^th to 21^st century, agent arguments in the form of the pronoun sun are on the left and attached to verbs.

1. Jayasantika ing Kudus, lawan si Sestrajaya, iku sun^A gawé^V papatih (BK-18)

in andthey I made governor

‘Jayasantika in Kudus and Si Sestrajaya, I made them governors’

1. Sun^A turuti^V panjaluke Aini (BYW-21)

I comply with request

‘I complied with Aini’s request’

The undergoer voice construction without a marker with the agent argument in the form of personal pronoun sun in Javanese appears in the Javanese dialect of Banyuwangi. In the Jogja-Solo Javanese dialect, the use of sun is limited to classical literary texts.

In the texts from the 16^th century, the personal pronoun ngong appears as an agent argument in undergoer voice construction. The personal pronoun ngong is the transformation of the pronoun ngwang (Fitriana, 2022). In the texts from the 10^th to 14^th century, the pronoun ngwang as an agent argument appears to the right of the verb with the affix marker ka- (V-A). The pronoun ngwang that appears to the left of the verb without an affix marker is not found in the text. The personal pronoun that can appear to the left of the verb without an affix marker is ngong.

1. ni Sri Tañjung mangke, lah ngong^A-ate˘rake˘n^Vmulih (ST-16)

now I’m-taking home

‘Now I'm taking Ni Sri Tanjung home’

Example (23) shows the personal pronoun ngong as an agent argument positioned to the left of the verb. Syntactically, the personal pronoun ngong is bound to the verb on the right.

Pronoun kula, serving as the agent argument of the undergoer voice construction without a marker, appears in texts from the 18^th century to the present. Besides the pronoun kula, the pronoun kawula that serves as the agent argument is also found.

1. Kuda nuli kawula^Abekta^V (Bkr-18)

Horse then I bring

‘then (that) the horse I bring’

1. Dawegane ajeng kula^A caosaken^V ngge sampeyan (Korpus BJ)

The coconuts will I be prepared for you

‘The coconuts will be prepared by me for you’

In examples (24) and (25) the agent arguments in the form of personal pronouns kawula and kula are both on the left and attached to the verb on the right. In the undergoer voice construction without a marker, the agent argument in the form of the pronoun kula is not only used in Javanese dialect of Jogja Solo, but also in other Javanese dialects. This happens because the use of pronouns at the level of krama speech tends to be the same, namely the use of kula as the first-person pronoun. The agent argument in the undergoer voice construction without a marker does not show a form reduction even though it is a bound form and attached to the verb. The bound feature of these pronouns only appears in the undergoer voice construction without a marker and has survived in the texts from the 10^th to the 21^st century.

Pronouns that serve as agents in the undergoer voice construction without markers show different grammatical characteristics from the original Javanese pronouns, such as aku and kami. For example, the pronoun nghulun appears in the form of a nominal phrase to express possessiveness. Meanwhile, to express personal possession, aku and kami appear in the bound form that are attached to the noun.

1. Ibungku (Ad-10)

‘my mother’

1. ibu ni nghulun (Ad-10)

‘my mother’

To indicate the notion of possessiveness, the bound form -ku is attached to the noun ibu (Example 25). The particle -ku/-ngku is a bound form of aku (Uhlenbeck, 1968; Zoetmulder & Poedjawijatna, 1992). This is different from the pronoun nghulun which constructs a nominal phrase with the noun ibu to express the notion of possessiveness (Example 26). In example (26), the pronoun nghulun is preceded by the ligature particle ni. Particle ni serves to connect the noun ibu with nghulun. The nominal feature of nghulun is indicated by the absence of form reduction such as aku which becomes -ku and is present in the ligatured-possessive nominal phrases.

The agent argument in the undergoer voice construction without a marker is always in the form of personal pronouns; no noun form (Fitriana, 2022). This is what distinguishes the characteristics of the undergoer voice construction with affix markers and the one without affix markers. In the undergoer voice construction with affix markers, the noun can be the agent argument. Another difference between the construction of the undergoer voice with and without markers lies in the pronoun type. In the undergoer voice construction with markers, the agent argument may be in the form of the original Old Javanese pronoun (aku, mami) with its bound form and the personal pronoun derived from the noun. While the undergoer voice construction without a marker, the agent argument is only filled with personal pronouns derived from nouns namely nghulun, ulun, ngong, kawula, and kula (Fitriana, 2022).

3.2.2. Deictic particle

Old Javanese has a deictic particle that connotes the first-person pronoun. According to Zoetmulder (2006), in Old Javanese, ndak/dak is a deictic particle which indicates the first person. In the undergoer voice construction without an affix marker, the agent argument can also be a deictic particle ndak/dak. In connection with these particles, Poedjosoedarmo (2002) made a hypothesis regarding the form of dak-. According to him, dak may be derived from the Malay language, hendak ‘want, will’, whose original form is not a pronoun. The form dak- appears as an agent in the proclitic form in Modern Javanese.

In contrast to Poedjosoedarmo’s opinion, the deictic particle dak- is already present in Javanese texts from the 10^th century, as seen in the example below.

1. Ndak^A warah^V ta kita (Ad-10)

I be told you

‘You’ll be told by me’

In the undergoer voice construction (28), ndak is the agent argument to the left of the verb. In the undergoer voice construction, ndak/dak is attached to the left of the verb and refers to the first person. In this construction, ndak does not function as the subject, because the word kita ‘you’ is the subject. This is proven by the presence of ta particles which serves to mark that the word on the left is a verb and the word on the right is the subject. The order of the predicate ta as subject is the order that is commonly found in Old Javanese (Puspitorini, 2015). Similar to the construction with agent arguments in the form of the first-person pronoun, construction with deictic particles is transitive. This happens because the agent argument in the form of a deictic particle must be present in the undergoer voice construction. This particle is syntactically attached to the verb and cannot be separated from the verb.

The order of the agent argument in the form of a non-deictic particle in relation to the verb is A-V. The presence of the agent argument in the form of a deictic particle to the left of the verb (A-V) proves that the argument order in an A-V pattern has existed in Old Javanese since the 10^th century. The A-V pattern is not a new form that has just appeared in Modern Javanese. This pattern has survived in text written from the 10^th century to the 21^st century by experiencing form transformation in its deictic particles.

In the texts written in the 17th century, the undergoer voice construction without an affix marker with the agent argument in the form of a deictic particle does not undergo change. The particle is attached and is to the left of the verb without a marker. Though based on the construction it doesn’t change, the development lies in the form of the present particle, namely dak-. In the texts from 10^th to 14^th century, the forms that appear are ndak-/dak-, whereas in the text written in the 17^th century, only dak- is found. The use of the form dak- has survived until the 21^st century.

1. dak^A tukune^V mās maṇik arih (TP-17)

I be bought gold gems

‘that gold and gems will be bought by me’

Besides dak-, there are also tak- and tek- forms that serve as agent arguments positioned to the left of the verb in the undergoer voice construction without affix markers. The forms tak- and tek- are present in the undergoer voice construction in texts written from the 20^th to the 21^st century. The difference between the two forms lies in the area of use. The form tek- is present in Javanese texts written in Banyumas dialect.

1. kowe tak^A dhelengi^V ngguya-ngguyu terus (KBJ)

you I be looked at smiling constantly

‘you are being looked at by me constantly smiling’

1. cengkolak kayu tek^Abuang^V (BYM)

bow wooden I throw away

‘the wooden bow I throw away’

Based on its form, the deictic particle ndak does not undergo change of form from Old Javanese to current Javanese. The form transformations are ndak/dak > dak > dak/tak/tek. Referring to this condition, the transformations can be observed from the loss of sound [n] in ndak, then the loss of voiced sound [d] to the voiceless sound [t]. Besides sound transformation, tak also has form variations, such as tek that is used in the Banyumas dialect. In the current Javanese, dak serves as a prefix. However, dak- does not reflect its characteristic as a clitic because it refers to the first person.

According to Poedjosoedarmo (2002) and Asridayani & Suryani (2020), dak in the current Javanese can be categorized as an agent-prefix because it functions not only as a passive marker prefix but also as an agent. Dak can be categorized as an affix because dak/tak in current Javanese is similar with the passive prefix di- (Adelaar, 2011). Dak/tak can be substituted with the passive prefix di-. Meanwhile, in Old Javanese, ndak cannot be substituted with passive affix -in- or ka-. The transformation of deictic particles to affixes indicates the grammaticalization of dak. The particle dak undergoes change in function from deictic particles to clitic and then transforms into affixes (deictic particles > clitic > affixes).

The following table below presents the occurrence of the undergoer voice construction based on the V-A and A-V pattern in each century.

Table 3 shows the development of the undergoer voice construction in Javanese texts from the 10^th to the 21^st century. The table shows that there are two patterns of order of agent arguments against the verb, namely V-A and A-V. The construction with the V-A pattern is the undergoer voice construction with an affix as a marker and is intransitive and semi-intransitive. The construction with the V-A pattern is more common in texts written from the 10^th to 14^th century. This construction is not found in Javanese texts from the 15^th to the 21^st century. In the texts from the 16th to 17^th century, there is only construction with the agent argument in the form of the pronoun sun. The development of the construction with the V-A pattern contrasts with the development of the undergoer voice construction with A-V pattern. The construction with A-V pattern is an undergoer voice construction without affix markers. This construction is transitive because the agent argument must be present and syntactically bound to the verb. This construction appears with an agent argument in the form of a pronoun derived from a noun and deictic particle. The construction with the A-V pattern still appears in 21^st century Javanese. The development of this construction is closely related to the development of the first personal pronouns of Javanese.

Table 3. The development of UV construction.

Pattern			Century
			10	11	12	14	15	16	17	18	19	20–21
V-A	affix ka-	nghulun	✔		✔
		ngwang	✔
		sun						✔	✔
		(de) mami	✔		✔
		(de)ngku	✔	✔	✔	✔
	affix -in-	ku	✔	✔	✔	✔
		mami	✔		✔
		(de)ngku	✔	✔	✔	✔
		(de)mami	✔		✔
A-V	nghulun	✓	✔	✔	✔	✔
	ulun	✓								✔
	ngong	✓					✔
	sun	✓					✔	✔	✔	✔	✔	✔
	kula	✓								✔	✔	✔
	ndak-	✓	✔	✔	✔	✔
	dak-	✓			✔				✔	✔	✔	✔
	tak-	✓									✔	✔
	tek-	✓										✔

4. Conclusion

Research on the undergoer voice construction with the agent’s argument in the form of the first personal pronoun in Javanese shows a development. This finding is different from Ogloblin’s (2008) claim about the absence of changes of undergoer voice construction in Javanese literary texts. Ogloblin (2008) pays attention only to the formal markers of undergoer voice construction, namely the affix -in-, to observe whether there is a change or not. This research also uses literary texts as the data source. These formal markers are indeed found in literary texts from the 10^th to the 21^st century. However, this research shows that the undergoer voice construction has indeed undergone development. The difference in these findings occurs because Ogloblin (2008) did not pay attention to the agent argument which is a crucial element in observing the undergoer voice construction of languages belonging to the Austronesian family. The agent is the determinant in distinguishing intransitive and transitive undergoer voice construction. The agent argument is also the determinant for showing the development of the undergoer voice construction in Javanese.

The results also claim that Javanese does not have completely the same type of construction as Filipino, as stated by Poedjosoedarmo (2002). While Javanese does have the affix -in- as a formal marker of the undergoer voice construction, which is found to be a characteristic of Filipino too, this research, however, demonstrates that Old Javanese also has the same type as Indonesian in respect to the presence of proclitic in the undergoer voice construction. The proclitic in the undergoer voice construction of Javanese emerging in the A-V pattern has existed since the 10^th century. Thus, Poedjosoedarmo’s (2002) claim which asserts that the A-V appears as the result of changes in the the undergoer voice construction with V-A pattern, is also proven to be false. Based on the type of undergoer voice construction, this research proves that it is possible for a language to have two types, namely Filipino and Indonesian. Old Javanese is a language that has two types of undergoer voice construction, namely Filipino and Indonesian. The two types of undergoer construction exist simultaneously and experience different developments. The results of this research can be used as a consideration to look again at the type of the undergoer voice construction in the Austronesian language family.

Research on the development of undergoer voice construction in Javanese needs to be extensively continued by investigating more varied data, both in terms of period and region, and furthermore on text type. Moreover, this research found agents that are oblique, such as dening kula and dening inyong, which are scarce in number (only one). Therefore, data expansion is required to strengthen or refute the conclusions drawn from this research.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Universitas Indonesia under Grant of PUTI number NKB-716/UN2.RST/HKP.05.00/2020.

Notes on contributors

Atin Fitriana

Atin Fitriana is a lecturer at the Javanese Study Program, Department of Linguistics, Faculty of Humanities, Universitas Indonesia. Atin completed her doctoral studies in 2022 with a research focus on the development of Javanese personal pronouns from the 10th century to the 21st century. Atin has been involved in several research projects such as the Javanese corpus, the Old Javanese corpus, the etymology of Old Javanese, the Javanese dictionary, and the Indonesian dictionary for Foreigner.

Dwi Puspitorini

Dwi Puspitorini was born in Magelang on October 11, 1964. Since 1991 until now she has been teaching at the Javanese Study Program, Universitas Indonesia. She completed her doctoral studies in 2015 with a research focus on Old Javenese. Several works have been produced, including Indonesian Grammar Practice Book for advanced level (2006), Jalan Bahasa Volume 2, Indonesian Grammar for Foreign Speakers (2007), Beginning Indonesian (2007), BIPA Textbook for beginners (with introductory language Japanese) (2007); PILAR: Textbook of Indonesian as a foreign language for beginners (2008). Scopus ID 57203359482

Myrna Laksman-Huntley

Myrna Laksman-Huntley born in Jakarta on January 5, 1961, received her doctorate in spoken communication (phonetics) from Stendhal University in France. She has been a faculty member of the French study program at Universitas Indonesia since 1991. From 2012 to 2015, she was the coordinator of the French study program at Universitas Indonesia. Scopus ID 57216618512