Botanical Roots and Word Origins: A Systematic Reconstruction of Alor Plant Name Etymologies

Abstract

This research provides a systematic reconstruction of nine botanical names from the Papuan languages spoken on Alor Island. Although genealogical links among the Papuan Timor–Alor–Pantar languages have been previously investigated, linguistic micro-studies of the hypothesized ‘shared’ ancestor languages within languages of the individual islands remain sparse. This study has three main aims: Firstly, to provide etymological reconstructions of Alor–Pantar Papuan words on a micro-level, focusing on Alor Island and specifically on plant names, which represent a cornerstone of Alor culture and history. Adopting the Comparative Method comparing cognates of 15 local languages, this research postulates historical phonetic shifts and language alignment phenomena and proposes proto-forms of the words in proto-Alor. Secondly, to shed light on possible prehistoric language contact and settlement patterns. A key finding suggests that the initial consonantal shifts *b > b > f > p in the languages is a probable loanword feature originating from the Austronesians. The geographical spread could indicate the influence of external trade and/or settlement patterns within the Papuan populations. Lastly, to provide a backbone for future etymological research on Papuan languages in Alor (and beyond) by mapping out aspects of language origins and phonetic influences and establish features of the shared proto-language(s).

1. Introduction

1.1. Significance, Aim, and Motivation of This Study

The study of etymology, a fundamental aspect of historical and contact linguistics, is defined as the history of a word: its origins, as well as phonetic and morphological derivations and shifts, acting as a biographical ‘fingerprint’ in the history of a language (Malkiel 1957; Mailhammer 2014). Etymological research aims to explore uncertain aspects of word history (or pre-history), where hypotheses of original root-forms (and their meanings) must be postulated and reconstructed (Durkin 2009, p. 1). The focus of this study will be on the indigenous Alor–Pantar (AP) languages of the Alor Archipelago1, specifically those from the island of Alor.

Before the 19th century, the lack of shared language records posed a significant challenge in etymological research. Language studies had to be conducted by a synchronic philological approach, often relying solely on historical documents of a single known language (e.g., Old English → Middle English → Modern English). However, at the turn of the 19th century, the Comparative Method became increasingly popular in the field of etymology (Hale 2014; François 2014; Weiss 2014), with etymology gaining increased recognition as a crucial aspect of linguistic research and lexicography (Malkiel 1957; Dworkin 2015). The method adopts a diachronic approach by analyzing lexical correspondences across languages, identifying sound correspondences (cognates), and postulating the existence of shared ancestor (proto-)languages.

Comparative research is significant not only at the linguistic level but at the sociocultural level as well. As culture and language often have a symbiotic relationship, sociocultural insights of speaker populations can also be gathered through comparative studies (Epps 2014). Derived processes of language shift amongst languages within a genealogical family are important in the study of language contact (Mailhammer 2014; Lucas 2014; LaPolla 2010). It provides a historical outline of population settlement and migration dynamics and explains how ideas, technology, and religious concepts were introduced to a region in the course of history.

Although etymological studies on indigenous communities and languages have remained sparse for decades, the turn of the 21st century saw a growing interest in language documentation and maintenance of endangered languages (Henke and Berez-Kroeker 2016). As researchers rushed to conduct fieldwork on languages, such as those of Eastern Indonesia (Klamer 2012), a rich library of language data became readily available, aiding comparative linguists in formulating increasingly accurate reconstructions and hypotheses. In the context of Alor, however, the histories of the local cultures remain enigmatic, as historical documentation and archaeological data are still greatly lacking (Klamer 2012). As such, the study of etymology can provide valuable linguistic insights on the origins of Alor communities and their connections to neighboring Austronesian communities, such as elements of language influence or borrowing (Perono Cacciafoco and Cavallaro 2018).

This ties in with the main motivation of this study, which is to utilize existing linguistic data to provide insights on linguistic phenomena, such as language alignment and shifts, within the AP languages and with external language groups (e.g., Austronesian). This study also hopes to postulate historical migration patterns during prehistoric times. Although some attempts have been made at etymological reconstructions of AP languages, none have specifically looked at correspondences in botanical names. Botanical names are a crucial point of etymological research, as plants are known to be a core part of the culture and livelihood of local Alor people.

This study aims to thus (1) investigate the linguistic evolution of Alor botanical names and (2) reconstruct botanical root-forms in proto-Alor (pAlor)—the hypothesized ancestral language of Alor languages. The Comparative Method will be utilized to study the cognate forms of 15 Alor languages. Additionally, the paper will also consider existing hypotheses by prominent scholars of pAP/TAP languages and provide critiques on these existing reconstructions.

1.2. Historical Background of Alor Island and Its People

Alor (Indonesian: Pulau Alor), coordinates 8°15′ S 124°45′ E, is located in Eastern Indonesia. It is the largest island in the Alor Archipelago, situated at the eastern end of the Lesser Sunda Islands, as shown in Figure 1.

Due to sparse historical records, the pre-colonial past of the Alor–Pantar Archipelago remains much of a mystery. The area caught the interest of Portuguese and Dutch colonial powers from the 16th century and eventually became a part of the Dutch East Indies in the mid-19th century, although they perceived the islands as unprofitable, and it was thus an often-neglected colonial territory. Up to the 1930s, the islands were among the places in the world that were least influenced by Western culture, and the Western world deemed the inhabitants as uncivilized and primitive (Wellfelt 2016, p. 13).

In the mid-20th century, Indonesia entered the tight authoritarian regime of the Indonesian ‘New Order’ era (1965–1998) under President Suharto, who avidly promoted anti-colonial sentiments in the country. Upon Suharto’s resignation in 1998, there was a crucial period of decentralization, and as people adjusted to a post-authoritative nation, Alor communities became interested in their own history and unique languages, separate from the country’s collective history (Wellfelt 2016, p. 13). At the start of the 21st century, these communities began to embrace their local oral legends and myths, which they hailed as sacred and an important part of their cultural identity.

Today, researchers are earnestly recording data on the history of the islands’ communities, along with language documentation and preservation efforts of the local AP languages. These language documentation attempts will be further detailed in the next section.

1.3. Language Documentation Efforts and Linguistic Background of Alor–Pantar Languages

Extensive language documentation efforts by researchers from Australia, the Netherlands, and Singapore from the 1990s have greatly expanded the database of grammatical and lexical information of at least 25–30 languages out of an estimated 200–250 languages spoken in Eastern Indonesia (Klamer 2012). Early researchers began by gathering words from the languages based on ‘universal concepts’ of the Swadesh list. Some of the words consistently recorded include body parts, pronouns, and generic environmental features, which include names of plants.

Languages of Eastern Indonesia have been grouped by linguists into two main families—Austronesian and Papuan. It is important to note that ‘Papuan’ is a blanket term based on geographical considerations, covering all linguistically non-Austronesian languages spoken on the western Pacific Island of New Guinea, including neighboring islands like Timor–Alor–Pantar (TAP), and do not imply inherent genetic relatedness (Hammarström 2014). However, lexicostatistic and glottochronological studies on the TAP languages have established a genealogical relatedness amongst the AP languages (Holton and Robinson 2014). As such, specific regional language groups, such as languages spoken in the Alor–Pantar Archipelago, are specified to be ‘AP Papuan languages’. Research on Austronesian languages in AP showed that prominent Austronesian language Bahasa Alor ‘Alorese’2 had less than 5% lexical similarity to the surrounding AP Papuan languages (Klamer 2012).

Historically, the AP Papuan languages could date back at least 40,000 years, spoken by prehistoric settlers, whereas Austronesian presence only appeared around a mere 6000 years ago, as indigenous inhabitants of Taiwan traveled to Southeast Asia, either for purposes of migration or trade, a phenomenon known as the ‘Austronesian Expansion’ (Klamer 2012; Perono Cacciafoco and Cavallaro 2018). Pronominal evidence could suggest an alternative possibility, in that the TAP languages were introduced by settlers from New Guinea and are linked to the Trans-New Guinea languages (Pawley 2005). However, this has not been firmly proven to be the case (Holton and Klamer 2017).

Language classification of the TAP languages branches the descendent languages into three subgroups: the proto-Alor–Pantar subgroup, the proto-Eastern Timor subgroup, and Bunaq (Schapper 2017).

The AP language family then branches out to a multitude of individual languages spoken in different parts of Pantar and Alor by separate communities, the languages are grouped under 2 branches: Alor and Pantar.

The Abui (abz) language (under the Alor branch), in particular, is one of the most widely spoken Papuan languages in Alor, with approximately 16,000 speakers (Perono Cacciafoco and Cavallaro 2018). Nanyang Technological University, Singapore (NTU) has been actively conducting research on the Abui community and their language. Pioneered by the team led by Assistant Professor František Kratochvíl, extensive fieldwork was conducted on Abui and other Alor languages from the early 2010s. He has also produced a comprehensive Abui–English dictionary (Kratochvíl and Delpada 2014) as well as a Sawila–English dictionary (Kratochvíl et al. 2014).

Subsequently, the NTU team shifted their research focus to botanical medical humanities, helmed by Dr. Francesco Perono Cacciafoco and Associate Professor Francesco Cavallaro. They actively gathered first-hand data on Abui botanical names and investigated the significance of plants in Abui culture: exploring their myths, legends, local herbalism beliefs, and links to micro-toponyms (Kratochvíl 2017); Lieu and Perono Cacciafoco 2021). Under the guidance of local consultant Mr. Benediktus Delpada, the data gathered were collated in the ‘Abui Botanical Corpus’ (Perono Cacciafoco 2018—ongoing), an online database created exclusively by the team. To date, 53 plant names have been recorded and detailed, along with their cultural significance in the Abui community, such as traditional medical properties and local legends passed down through the generations. The Corpus can be found on https://blogs.ntu.edu.sg/abui/collection-2/ (accessed on 1 November 2024).

1.4. The Alor Languages and the Analyzed Languages

This study focuses on fifteen Alor–Pantar languages spoken on Alor Island, and grouped under the Alor branch of languages, specifically, Adang (adn), Kabola (klz), Klon (kyo), Kafoa (kpu), Kui (kvd), Hamap (hmu), Abui (abz), Suboo (sub), Papuna (pap), Tiyei (tiy), Kiraman (kira), Wersing (kvw), Kamang (woi), Kula (tpg), and Sawila (swt), listed from west to east of Alor. The general geographical range of the languages, from Perono Cacciafoco et al. (2015, p. 40)3, is shown in Figure 2. The languages and the relative fieldwork locations retrieved from the LexiRumah database (Kaiping et al. 2019) are shown in Figure 3.

The following provides a brief background of the Alor languages (from Perono Cacciafoco and Cavallaro 2018; Hammarström et al. 2021). Information on the Agglomerated Endangerment Status (AES) of languages are retrieved from Glottolog 5.1 public repository at https://github.com/glottolog/glottolog (accessed on 1 November 2024).

Adang is a language spoken in west Alor, around the Adang–Buom area, located on the western Kalabahi border. It is closely related to Hamap and Kabola. AES status: ‘shifting’. ISO 639-3: adn; Glottolog: adan1251.

Kabola is spoken around the north of Kalabahi. AES status: ‘not endangered’. ISO 639-3: klz; Glottolog: kabo1247.

Klon (or Kelon) is a western Alor language spoken around the Kalabahi bay (Perono Cacciafoco and Cavallaro 2018). It is closely related to Adang. AES status: ‘shifting’. ISO 639-3: kyo; Glottolog: kelo1247.

Kafoa (or Jafoo) is spoken in the areas between Klon and Abui (Perono Cacciafoco et al. 2015, pp. 40–41). Native speakers do not refer to the language as ‘Kafoa’, though the local name used is still unknown. AES status: ‘threatened’. ISO 639-3: kpu; Glottolog: kafo1240.

Hamap is spoken in western Alor, around Moru village. AES status: ‘threatened’. ISO 639-3: hmu; Glottolog: hama1240.

Abui is spoken in the central part of Alor. Abui tanga is the language’s native name, which means ‘mountain language’. AES status: ‘shifting’. ISO 639-3: abz; Glottolog: abui1241.

Suboo is a language spoken east of the Abui area (Schapper 2014). ISO 639-3: N/A; Glottolog: N/A. Code sub is used for ease of reference.

Papuna (pap) is a language that is likely related to Abui (Schapper 2014)—the two populations likely split at some point. It is spoken east of the Abui areas. ISO 639-3: N/A; Glottolog: N/A. Code pap is used for ease of reference.

Tiyei (or Tiayei, or Tiee) is spoken towards the north of the Papuna-speaking region (Schapper 2014). It is often classed as a dialect of Kamang. ISO 639-3: N/A; Glottolog: N/A. Code tiy is used for ease of reference.

Kiraman is closely related to Kui. ISO N/A; Glottolog: kira1248. Code kira is used for ease of reference.

Wersing (or Kolana) is spoken in a scattered pattern around the coast of Alor. AES status: ‘threatened’. ISO 639-3: kvw; Glottolog: wers1238.

Kamang (or Woisika) is spoken to the north of Suboo-speaking areas. It has multiple related dialects: Lembur, Sibo, Kamang, Tiyei (or Tiayei, or Tiee), Watang, and Kamana-Kamang. AES status: ‘shifting’. ISO 639-3: woi; Glottolog: kama1365.

Kula (or Kola) is spoken on the Eastern side of Alor. Its dialects are Kula, Kulatela, Watena, Kula Watena, Iramang, Larena, Sumang, and Arumaka. AES status: ‘threatened’. ISO 639-3: tpg; Glottolog: kula1280.

Sawila (or Tanglapui) is spoken in the southeastern part of Alor. Its dialects are Sawila, Lona, Salimana, Lalamana, and Sileba. AES status: ‘shifting’. ISO 639-3: swt; Glottolog: sawi1256.

2. Literature Review

This literature review will focus on three different aspects of existing research: Firstly, the extant literature on existing etymological research and methods, as well as concepts of language change, will be outlined. This is to provide a foundation for the rigorous systemic research method of comparative analysis that will be utilized in the present study. The second part will address the research gaps and highlight the current literature on the lexical repository of Abui botanical names, including the study of language reconstruction in the Timor–Alor–Pantar region. The third category will review prior reconstruction attempts of the hypothesized pAP language and the overarching pTAP language.

2.1. Methods of Etymological Reconstruction and Concepts in Language Change

Mailhammer (2014) describes etymology as the diachronic study of a language by tracking the origins of linguistic elements, such as lexical items (lexical etymology) and structure (structural etymology), accounting for specific changes it has undergone. He highlights the intricacies of lexical etymology, including the study of phonological forms, inflectional morphology, word formation details, word meaning, and syntax. Crucially, he also provides categorical grading on the quality and plausibility of etymological reconstructions based on the strength of the relations derived from the comparative analysis of lexical cognates. Stronger matching of formal and semantic elements among cognates are ranked as high quality, while reconstructions based on uncertain or largely unsubstantiated/assumed connections are graded lower. Lastly, he delves into contact etymology, establishing its importance in highlighting important historical information about language communities and in accounting for unexplained linguistic aspects of languages.

Garrett (2014) analyzes the mechanism and scholarly debates surrounding studies in sound change. He raises the point that sound change does not simply entail phonologization. He summarizes the debate on the Neogrammarians’ theory of sound change, with the concept of a lack of exceptions coming into question. He also notes a recent/modern shift in focus in the field of language change towards a more socially contextualized (sociohistorical) perspective. He highlights issues of actuation and embedding problems in the field of language change. Actuation in the reasons and process of how language change is initiated and the question of why it happens at a certain time in one language and not another. Embedding problems are the issues in the relations between an element of language change and the social and linguistic context. Specifically, how phonological, morphological, and lexical aspects of a language affect sound changes. The paper provides a glimpse into the complexities of sound change in historical linguistics and emphasizes the need to focus not only on obvious technical aspects but also to look at the ways social and linguistic elements in a language could have impacted and initiated these changes.

Fowler (1957) presented on the vitality and applicability of the Historical–Comparative Method in the context of linguistics, detailing aspects of how its popularity in the early 19th century catapulted the field from merely a subset of classical studies to a well-regarded natural science—the scientific study of language. He also outlined further discoveries of sound laws within the Indo-European languages (IE), such as Grimm’s law, which introduced rules (sound laws) in proto-Indo-European that were consistent without exceptions (Neogrammarian Hypothesis). Scholars in the 1950s started to focus heavily on language structure, extending the Comparative Method to include the analysis of morphemes, instead of simply phonemes. This novel progression essentially brought about new opportunities for research on the ‘genetic’ relationship and links (language genealogy) among different languages. He called for further comparative studies in ancient languages, such as Anatolian and Hittite. Even though this paper reflects a stage of historical–comparative linguistics in its relative infancy, it nevertheless provides valuable insights into the Comparative Method—its usage, relevance, and impact on the field of linguistics.

Perono Cacciafoco et al. (2015) advocate for the experimental convergent methodology in the study of language reconstruction in undocumented languages of Southeast Asia, specifically focusing on the Abui language. The method involves the analysis of toponyms, hydronyms, and oronyms through the study of geography, cartography, landscape archaeology, and historical semantics, taking into account the speaker’s cultural, spiritual, and pragmatic interpretations of a place, in order to discover the ancient origins of a language and produce a ‘remote stratigraphy’ of the place names.

2.2. Research Gaps

To provide some background, one of the main obstacles in the study of Alor–Pantar history is the largely absent, or yet undiscovered, historical and archaeological data. Linguistic comparative studies are therefore critical in allowing for historians to gather data on the islands’ history and settlement patterns. As words would have retained elements of their prior forms, delving into the linguistic ‘past’ of the local languages is an important step in achieving this goal.

For the past twenty years, research on AP languages has been centered on earnest attempts at language documentation and data collection to, in essence, cover the bases of Eastern Indonesian languages that have been neglected for most of history. More recent developments in linguistic studies of the Alor–Pantar Archipelago include the grouping of languages into typological categorization and language families. Some attempts have been made at reconstructing the roots and features of pAP and pTAP (see Holton et al. 2012; Holton and Robinson 2014; Heston 2017; Perono Cacciafoco and Cavallaro 2018); however, extensive analysis is lacking, especially considering the large variety and quantity of language data that have been collected thus far from the islands.

One group of lexemes that has been mostly neglected is plant names. Plant taxonomy (phytonyms) among communities in the Alor–Pantar Archipelago has recently been an emerging point of interest. On the sociocultural significance of vernacular plant names, Binns (1976) states that it is ‘important to link the printed scientific names with the oral knowledge of its peoples; otherwise, in the long run, their reliable vernacular experience becomes irretrievably lost.’ Being an integral part of the geographical landscape, phytonyms are a practical choice for comparative linguistic studies, as they are likely to have been identified since ancient times and should be relatively stable in their linguistic form.

As the local people relied heavily on oral transmission of information for most of their history, the need for accurate botanical species distinction for herbal applications would have logically made it unfeasible for drastic phonological adaptations and shifts throughout history. Therefore, even as the population dispersed across the island in early history, forming individual communities, and eventually developing their own language variety with distinct features, the botanical forms would have a higher probability of retaining original linguistic features, more so than other conceptual lexeme categories. The main goal of this paper is to, thus, address this clear research gap by delving into the etymology of AP botanical names. As some plants were recorded to be non-native, originating from other linguistic groups, this research also attempts to uncover previously undiscovered insights on the historical impact of external language contact on the Alor languages.

2.3. Abui Lexical Research

Kratochvíl et al. (2016) studied the toponymy of the Abui community by building a repository of Abui place names (toponyms). The authors found that the toponymic naming patterns of the Abui community are mostly transparent and linked to agricultural and horticultural elements in the environment and landscape. The authors also studied the social and cultural functions of place names, as well as their connections with community dynamics.

Lieu and Perono Cacciafoco (2021) introduced Abui as a fairly undocumented and endangered language. They utilized field linguistics and language documentation methods, with the assistance of local language consultants, to build an Abui Botanical Corpus with 54 plant names. They gathered data on local names of plant species and the cultural significance associated with the plants. The paper focuses heavily on the relationship between plants and the culture (oral traditions) of the Abui people. A limitation of the study was raised by the authors, mainly that photographic or laboratory verification of the plants was not conducted, and collection methods were highly reliant on the local correspondents’ identification and input. Future studies should aim to provide more rigorous verification methods to aid in the verifiability of the collected data.

2.4. Studies on Proto-Timor–Alor–Pantar and Proto-Alor–Pantar

Klamer (2012) provides a case study of micro-level historical reconstruction of the language Bahasa Alor ‘Alorese’ of Eastern Indonesia. In contrast to the linguistic research and typological ‘fingerprint’ studied in macro-level linguistic research, exceedingly comprehensive cultural data, such as ethnographical observations and historical oral stories, allow for the micro-linguistic study of languages in the region by combining both linguistic and cultural information. She found a 60% lexical similarity between Alorese and the nearby Austronesian language Lamaholot spoken on Flores Island, although Alorese has lost most of its original morphology. She concluded that the two languages came from the same ancestral language and hypothesized a timeline of speaker migration based on oral stories and legends. This suggests that Alorese speakers are fairly new to the region, migrating from the Lamaholot-speaking regions (e.g., Flores) to the Alor–Pantar Archipelago only around 700 years ago ante quem. Although the scope of the paper is limited to the Austronesian Alorese group, it signals an impressive breakthrough in Alor–Pantar linguistics. As language and typological data are increasingly supplemented to provide fine-tuned oral and ethnographic studies based on individual communities, research on AP languages is no longer simply confined to macro-level studies.

Holton et al. (2012) and Holton and Robinson (2014) studied 12 Papuan languages and reconstructed above 100 pAP lexical items and consonants and proposed sound laws across the languages (shown in Figure 4 and Figure 5). The Comparative Method was applied to examine sound correspondences within primary lexical data from these languages, using form-meaning pairing in basic cognate sets. The authors also concluded that there was no linguistic evidence to link AP languages to the trans-New Guinea family, thus deeming it a separate language family of its own, which they termed the Timor–Alor–Pantar (TAP) language family. The study proved to be a significant breakthrough in AP linguistics, finally solidifying theories of an ancestral link between Papuan languages of the Alor–Pantar Archipelago and the parent language that came prior—proto-Alor–Pantar (pAP).

Heston (2017) attempted to reconstruct vowels in the pTAP language by studying languages in the TAP family. By analyzing sound correspondences among the languages, he reconstructs a total of five cardinal vowels and schwa: *a, *e, *i, *o, *u, and *ə. He found, however, that there were nonconforming correspondence sets that cannot be easily explained from the six vowels reconstructed. Some limitations of this study, therefore, include the lack of coverage in certain aspects of vowel length, as well as the possibility of additional/alternative vowel phonemes in pTAP, which would account for the nonconforming sets.

Perono Cacciafoco and Cavallaro (2018) conducted a comparative study of common words in nine Alor languages (words such as ‘sky’ and ‘moon’) and provided reconstruction attempts of the historical phonetic features of pAP based on recorded forms gathered from extensive fieldwork. The authors acknowledged that the etymologies provided cannot be fully ascertained or proven without a doubt. However, they expressed the importance of reconstruction attempts in postulating prehistoric population movements and the value of providing a framework for the historical study of Alor communities.

3. Methodology

3.1. The Comparative Method

In this study, the Comparative Method in linguistics will be used to attempt the etymological reconstruction of nine Alor plant names. According to Rankin (2017), the Comparative Method is a set of techniques, developed across a period of around a century and a half, that allows for us to recover elements of early linguistic (mainly unattested) stages in a group of languages within a language family (i.e., languages descending from an ancestor language or ‘proto-language’ and branched out into the many languages today). The method identifies patterns and distinctions from cognates as products of either a shared genealogy or shared responses to specific historical conditions (Griffiths 2017), which will allow for us to theorize early linguistic forms and map out the links between the language systems.

The current study focuses on applying the Comparative Method to Alor plant name cognates, as phonologically, the names share the same set of initial consonants (labial initials), which enables the combination of the Comparative Method and investigation of extralinguistic factors such as archaeobotany and historiography.

3.2. Materials and Data

Two primary sources were used to acquire lemmas for comparative research: the (1) LexiRumah database and (2) Proto-Alor Lexicon Database.

The LexiRumah database (Kaiping et al. 2019) is an online database containing a collation of lexical data from survey wordlists and published sources on the languages of the Lesser Sunda Islands of Eastern Indonesia and Timor-Leste, including the Alor–Pantar Archipelago. The database was created and is maintained by the team at Leiden University, the Netherlands. It is part of the NWO Vici project ‘Reconstructing the past through languages of the present: the Lesser Sunda Islands’, under a grant from the Netherlands Organization for Scientific Research. The lexical terms are listed with their conceptual matches, along with an English definition. Each word is also associated with three types of transcriptions: IPA standardized transcriptions, orthographic transcriptions (for languages with an established orthography), and the original form of the word recorded in the field notes of the researcher(s).

The Proto-Alor Lexicon Database is a database that aims to be an etymological dictionary of the targeted Papuan languages. It was developed at the NTU Linguistics and Multilingual Studies Programme (LMS) by František Kratochvíl and Francesco Perono Cacciafoco. The database is in the form of an Excel file containing a total of 2000–4000 forms among nine Alor languages, sorted by semantic areas. Most of the data were collected independently by the team through extensive periods of documentative work, as well as with supplementary input from researchers of other universities. To ensure the accuracy of each lemma, every entry has been double-checked and approved by native language consultants and matched against existing lexicons and grammar sources.

3.3. Methods

The IPA form of specific botanical words (corresponding to a specific concept, e.g., ‘Mango’) were extracted and chosen from the two databases. The criterion were based on (1) the availability of the plant name in the wordlists and (2) a preliminary comparison of the local form for each plant. For (1), the plant should be recorded in at least 10 languages with possible cognates consisting of forms from at least 6 languages. For languages with multiple forms per lexeme (from different wordlists), as long as the entry appears in one of the sources, it is considered available. This is to ensure consistency and accuracy of the reconstructions. For (2), if there is a significant indication of a lack of phonetic similarity between the forms, the plant is not chosen for analysis.

Taking these factors into account, the plant names chosen were ‘Betel Nut’, ‘Corn/Maize’, ‘Mango’, ‘Coconut’, ‘Jackfruit’, ‘Bamboo’, ‘Sweet Potato’ (with mentions of ‘Banana’), ‘Gum Tree/Eucalyptus’, and ‘Rice (uncooked)/Rice Plant’.

Next, the lexemes were then compiled in an Excel sheet (refer to Supplementary Materials), starting with a column containing the plant name, followed by the proto-form proposed by Holton et al. (2012), if available. The cognates were then added to the subsequent columns, arranged in a general trend from west to east of Alor. All cognates, even those that appeared more than once in a particular language due to different fieldwork data, were recorded in the Excel sheet to build a statistically rich pool of data for current and future studies. The word forms provided by the NTU Botany Team were listed in the Excel sheet but were not utilized for analysis, as no IPA transcriptions are available (given forms are written in orthography).

For ease of cognate comparison, only selected forms were included and listed directly in the analysis portion of this paper, with the specific fieldworker cited. The rest were assessed to most likely be non-cognates. In the case where there was a notable difference among lexemes in a single language relevant to the analysis, it was highlighted explicitly. This was implemented to accommodate the extensive data in this study, which required the careful selection of relevant data to prevent an overload of unnecessary information.

Comparative analysis was then conducted based on the cognates. Initial stages included the comparison of consonants and consonant clusters, identifying phonemic patterns, suggesting probable patterns of linguistic development, and accounting for deviations. Next, prior pAP forms proposed by Holton et al. (2012), if applicable, were also critically assessed. The final steps included a proposal of a proto-form of each word and other general insights that could be gathered from the comparison.

4. Results and Discussion

4.1. Betel Nut

Holton et al. (2012) proposed *bui as the reconstructed pAP form for ‘Betel Nut’ (Scientific name: areca catechu).

Possible cognates: adn buː; klz buʔ; kyo buːi; kpu fʊi; kvd bui; hmu N/A; abz fu; sub N/A; pap bʊ; tiy N/A; kira bui; kvw pui; woi N/A; tpg pi; swt pu.

Lexemes from (fieldworkers) used are adn (Robinson 2010), klz (Robinson 2010), kyo (Choi 2015), kpu (Baird 2003), kvd (Holton 2010), abz (Schapper 2010), pap (Delpada 2011), kira (Holton 2010), kvw (Holton 2010), tpg (Williams 2010), and swt (Choi 2015).

From this dataset, it is clear that the initial consonantal sounds are consistent with Holton et al.’s (2012) proposed sound laws: *b > p in the east Alor languages kvw and swt and *b > f in abz (see Figure 4 and Figure 5).

Scholars believe that the ‘betel nut’ seed (from the areca palm) is non-native to the Alor archipelago and that the Papuan Alor forms may have been borrowed from Austronesian languages via language contact (Kratochvíl 2017). The plant was initially domesticated in the Southeast Asia region and possibly introduced to the Alor archipelago at a later point (Yen 1977; Holton and Robinson 2014), though there is a lack of archaeological information to provide an exact timeline of its arrival. However, available linguistic and archaeological findings on proto-Austronesians located in today’s Taiwan showed that ‘betel nut’ was present in the region and regularly sold to other communities (Lichtenberk 1998).

Austronesians were known for their sailing capabilities in ancient times, often selling exotic and unique goods to other regions of the world. They also had a long history of trading in the Alor archipelago, with some Austronesians even migrating permanently to the Alor–Pantar Archipelago. Linguistic traces of this migration are still seen today, such as in the Austronesian language Alorese spoken in parts of Alor. Cognates in Austronesian languages that are phonetically similar to the AP names include Tetun bua ‘betel’ and Tokodede buo ‘betel’ (Holton and Robinson 2014). Although Holton et al. (2012) acknowledge the possibility of borrowing, they chose to proceed with the pAlor reconstruction of the word.

The current cognate list is sorted from west to east Alor languages. Analysis of the dataset appears to show a pattern that could indicate a spread of the lexeme from Pantar Island to west to east Alor, with the only exceptions being the pap and kira forms (mapped in Figure 6). A random sampling of Papuan wordlists from Pantar Island showed that multiple languages spoken in Pantar villages directly west of Alor, such as Sar, Deing, Kaera, and Teiwa, represent ‘betel nut’ as bui. It can, therefore, be postulated that the name could have spread eastward and adapted to the sound change laws of the specific Alor languages pAlor *bui > bui > fui > pui.

It is also notable that there is a loss of medial (syncope) or final vowel (apocope) in some of the forms.

Loss of medial vowel: tpg pi;

Loss of final vowel: adn bu, klz buʔ, abz fu, swt pu.

Vowel lowering is seen in some of the forms in nucleus position u > ʊ in kyo bʊi kpu fʊi pap bʊ abz fʊ (alternative lexeme documented by Kratochvíl 2017). However, /u/ and /ʊ/ are similar-sounding. They are both high rounded back vowels, differing mainly in vowel tenseness. As such, an alternative possibility could be an inaccuracy of the vowel sounds recorded by the fieldworkers.

In summary, with the current linguistic data available, Holton et al.’s (2012) reconstruction of pAP root *bui is indeed the most plausible. An additional insight gathered in the current study is that the AP word forms could have originally been loanwords from the Austronesian language, which gradually spread from west to east of the island, adapting to the sound laws unique to some of the languages. This explains a likely shift in initial consonant pAlor *b > b > f > p from the western to eastern languages on the island.

Some limitations to this analysis are that ‘areca nut’ and ‘betel nut’ may refer to separate species of plants, though it is mostly classified together by fieldworkers. The NTU Botany team found possible differences in the species of the two nuts, acquiring two different forms from the local Abui people (abz language): betel vine nut ‘meeting’; areca nut ‘fu’. Future fieldwork on the AP languages should take this into account and obtain possible differing terms for the two species to provide a more accurate botanical classification. Another issue is that most wordlists from field studies of the Austronesian languages of Flores do not contain the lemma for ‘betel/areca nut’. More lexemes obtained in this region can help cast a wider net to assess the validity of the ‘east to west spread’ theory and map out a possible direct movement path.

4.2. Corn/Maize

The sound change theory proposed in the ‘Betel Nut’ dataset can be supported by findings on ‘Corn/Maize’ (scientific name: Zea mays).

Possible cognates: adn ˈbate; klz botiʔ; kyo bat; kpu N/A; kvd batar; hmu N/A; abz ˈfɑt; sub patiː; pap baːtɪ; tiy pati; kira bati; kvw peˈter; woi patei; tpg pəˈte; swt pata.

Lexemes from (fieldworkers) used are adn (Choi 2015), klz (Robinson 2010), kyo (Choi 2015), kvd (Holton 2010), abz (Kratochvíl 2017), sub (Han 2014), pap (Delpada 2011), tiy (Delpada 2011), kira (Holton 2010), kvw (Holton 2010), woi (Schapper 2011), tpg (Choi 2015), and swt (Choi 2015).

Holton and Robinson (2014) acknowledge that the forms could have been borrowed from Austronesian languages. As the crop was first introduced to the region in the 15th–16th century by the Dutch, the AP words for ‘maize’ are likely cognates with Old Malay batari ‘sorghum’ (Fox 2003; Holton and Robinson 2014). It is postulated that the Austronesian word for ‘maize’ was first introduced in West Timor, originating from the Austronesian languages of Timor, and slowly diffusing into the other TAP communities and languages over time (Holton and Robinson 2014). Possible cognates to the AP set can also be currently found in Austronesian languages in Timor such as Tetun batar ‘maize’, further solidifying this theory.

From the dataset here, with cognates sorted from west to east Alor, it is clear that the phonological innovations or sound laws are similar to that of ‘betel nut’, pAlor *b > b > f > p, in the initial consonant:

/b/ in adn ˈbate klz botiʔ kyo bat kvd batar;

/f/ in abz ˈfɑt, kpu (no data);

/p/ in sub patiː tiy pati kvw peˈter woi patei tpg pəˈte swt pata.

Additionally, almost exact to the betel nut data, pap baːtɪ and kira bati are exceptions with /b/ initial consonants despite not being located in the western region.

In conclusion, these findings could help researchers to study the migration and/or trading patterns of the communities. The Austronesian trade in the Alor–Pantar Archipelago could have introduced these plants to the Alor people. The ancient Austronesian people, being great sailors, would have had the ability to enter the Alor region from any direction. However, it would naturally have been more favorable and convenient to enter from the west coast (northwest and southwest) of the island.

Additionally, these findings could also be linked to settlement patterns of the Papuan people in Alor. As the ancient Papuan people were known to be poor sailors, they likely had to enter Alor from the southwest, which was the only place where the sea currents did not crash simple boats into the cliffs. Some archaeological findings in the region, such as pottery from the west, are determined to be more ancient than the ones found in the east.

The phonological exceptions in the data, i.e., the forms in pap- and kira-speaking communities, may suggest that pap- and kira-speaking villages/communities could be more ancient than the other surrounding communities.

However, more research will have to be conducted to confirm these hypotheses.

4.3. Mango

The pAP root for ‘Mango’ (scientific name: Mangifera indica) has not been reconstructed by scholars thus far.

Possible cognates: adn be; klz beʔ; kyo N/A; kpu N/A; kvd N/A; hmu N/A; abz ˈmɛɑ; sub lawe/lɑveː; pap mɛ; tiy laweʔ; kira muˈwei; kvw N/A; woi wae; tpg wagi; swt N/A.

Lexemes from (fieldworkers) used are adn (Robinson 2010), klz (Robinson 2010), abz (Kratochvíl 2017), pap (Delpada 2011), sub (Han 2014; Kratochvíl 2008), tiy (Delpada 2011), kira (Holton 2010), woi (Schapper 2011), and tpg (Williams 2010).

The first observation from this set of cognates is the prefix la- found in sub and tiy forms and mu- in kira. These prefixes are not omnipresent in the data and are likely to be later additions to the original form or morphological innovations.

Phonetic analysis shows an initial consonant of adn, klz /b/, abz, pap, kira /m/ and sub, tiy, kira, woi, and tpg /w/. As bilabial forms often signal an early phonetic stage, *b- could be the reconstructed initial consonant. Thus, the proto-form would be pAlor *be-.

To account for the sound changes in the other initial consonants, b > m could have occurred through the process of denasalization. Denasalization is a prevalent process of sound change in Austronesian languages (Lackey 2019). As Alor–Pantar is on the border with Flores, an island consisting of mostly Austronesian languages, Austronesian influence on Papuan languages is believed to be strong with possible phonetic and morphemic adaptations brought on from Austronesian.

The sub form lave gathered by Kratochvíl (2008) presents some interesting insights. Focusing on the assumed root -ve, the initial consonantal sound /v/ appears to be unique to this cognate. Initially, this may seem like a probable transcription error, as Holton et al. (2012) did not find *v to be an existing consonant in the present Western Pantar consonant inventory or in pAP. However, in many Asian languages, /v/ and /w/ are phonetically similar, such as in Hindi (Grover 2016), or phonetically different but semantically similar, such as in Mandarin Chinese (Shen [1959] 2006). If this is applicable to the case of Alor languages, /v/ and /w/ could both be existing phonemes, but it could be challenging for non-native researchers to recognize the phonetic distinctions. As the Papuan AP languages are still not comprehensively studied, more research needs to be conducted to confirm these phonetic observations.

Assuming /v/ exists in the sub form and is accurate, b > v could have occurred through the process of betacism. Betacism is found to occur very commonly in languages over time, with the voiced bilabial plosive [b] and voiced labiodental fricative [v] commonly being confused by speakers. Betacism has been recorded to have taken place in well-documented languages like Greek, Hebrew, and other Romance languages. The initial consonant sound change could thus be postulated to be pAlor *b > b > v > w.

Another phonetic observation is the presence of glottal stops in klz beʔ and tiy laweʔ. Glottal stops are believed to be innovations in the AP languages as there is not enough evidence to prove its existence in pAP (Holton et al. 2012). Additionally, the dataset itself does not provide sufficient evidence that the two forms could be ‘original’. Therefore, the final glottal stop is not considered to be part of the pAlor root-form.

The vowel /ɛ/ in pap mɛ and abz ˈmɛɑ also raises some intrigue. It is important to note that pap is closely related to abz, likely to be a subgroup of abz that deviated at some point in history (Schapper 2014; Perono Cacciafoco and Cavallaro 2018). These two forms could have originated from an earlier shared form, explaining the unique presence of /ɛ/. However, /ɛ/ has not been reconstructed as a possible pTAP vowel by scholars (Heston 2017), making the possibility of vowel lowering *e > ɛ in pap highly probable. Forms abz ˈmɛɑ, woi wae, and kira muˈwei are also noted to consist of diphthongs /ɛɑ/, /ae/, and /ei/, respectively, possibly indicating a process of diphthongization: *e > ɛɑ, *e > ae, and *e > ei.

In summary, the proto-form is postulated to be a C-V root, pAlor *be-. The lexemes for ‘mango’ in the AP languages studied appear to show a clear similarity, with explainable phonetic differences occurring in the initial consonant, vowels, as well as other peculiar features. Firstly, the word-final glottal stops found in klz beʔ and tiy laweʔ are believed by scholars to be innovations (Holton et al. 2012). Next, to explain the existence of multiple sounds /b/, /v/, and /w/ in the initial consonant position of the forms, a proposed phonetic chain passage would be *b > b > v > w. To account for differing vowels, *e > ɛ could have occurred in pap mɛ, through the process of vowel lowering, while *e > ɛɑ, *e > ae, *e > ei (abz ˈmɛɑ woi wae kira muˈwei) could indicate diphthongization.

4.4. Coconut

Holton et al. (2012) proposes *wat(a) as the reconstructed pAP form for ‘Coconut’ (scientific name: Cocos nucifera).

Possible cognates: adn ˈfa; klz waʔ; kyo ʔəˈtaː; kpu N/A; kvd bat; hmu N/A; abz wata; sub wate; pap wɔtɑ; tiy wate; kira bat; kvw waˈtaʔ; woi wate; tpg gwata; swt waˈta.

Lexemes from (fieldworkers) used are adn (Choi 2015), klz (Robinson 2010), kyo (Choi 2015), kvd (Holton 2010), abz (Schapper 2010), sub (Han 2014), pap (Delpada 2011), tiy (Delpada 2011), kira (Holton 2010), kvw (Holton 2010), woi (Schapper 2011), tpg (Williams 2010), and swt (Choi 2015).

Holton et al. (2012) reconstructed *w- in the initial consonant position as it is the widespread form. They also found a pattern of *w > f in adn, while other languages preserve *w-. However, an interesting observation is the form bat in the related languages of kira and kvd. This unique initial consonant /b/ could possibly be an earlier, more ‘ancient’ form. As suggested by Perono Cacciafoco and Cavallaro (2018, p. 5), bilabials such as /b/ in roots could, ‘sometimes, indicate a remote stadium of language’. *b- may, therefore, be a more ancient form that shifted to w-, *b > w, with w- becoming the regularized widespread form over time.

Consistent with Holton et al.’s (2012) findings, the final /t/ is proposed by them to be lost in adn but preserved in other languages. It is, thus, probable that the proto-form may be pAlor *bat-, retained in kira and kvd. The vocalic epenthesis final -(a) added could be an innovation, as it is not found in the kira, kvd, or adn forms.

Adang: *bat > *fat > faʔ/fa;

*bat > *wat > wat(a).

Another possible theory could be that the proto-form contains the consonant cluster *gw- (i.e., tpg gwata). This could explain why *b- becomes *w- instead of the expected *f- (> *p-), because the initial component *g- could have blocked the fricativization and favored the glide becoming a stop *g.w- > *b- ? In other words, the relative chronology and sound laws would be *gw- > *w- in languages that drop *g-, and *gw- > *b- in languages that de-cluster the initial stop and subsequently assimilate the second glide component *w- to become a stop *b-.

A theory brought up by Perono Cacciafoco in his preliminary notes on possible reconstructions (Kratochvíl 2017), was that tpg gwata could be sort of an original ‘relic’ form, a phonetically unchanged descendent of hypothesized proto-form *gwata. He postulates that initial dropping of */g/ could have occurred in the other languages.

Lastly, one cognate that appears peculiar is kyo ʔəˈtaː., the adn form gathered by Choi (2015) in the Otvai village. The ʔə- prefix appears to be an innovation (as glottal stops are usually believed to be the case). However, it could also be a stylistic speech pattern (such as a pause or fillers) used by the specific speaker interviewed. This can be evidenced by the prevalence of the speech in a large portion of lexemes in the wordlist across multiple categories, ranging from plants to animals to count numbers to metaphorical concepts and adjectives, with instances of reduplication as well. Additionally, Baird’s (2003) wordlist of kyo in the nearby village of Bring did not contain this prefix in any of the lemmas. As such, this prefix can be safely ignored in the analysis. This will also be the case for the next section on ‘Jackfruit’.

In summary, Although Holton et al.’s (2012) reconstruction pAP *wat(a) may appear sound, the initial consonant /b/ can be argued to be more ancient than /w/. Additionally, the vocal epenthesis -(a)- does not occur in many of the languages, indicating that it could have been an innovation. Therefore, an alternative proto-form proposed through analysis of this data is pAlor *bat-. Alternatively, the proto-form could be reconstructed as the consonant cluster *gw- as as retained in tpg gwata. A theory proposed by Perono Cacciafoco (Kratochvíl 2017) is that proto-form *gwata is retained in the tpg gwata, with all other languages dropping the initial consonant *g-.

4.5. Jackfruit

The pAP root for ‘Jackfruit’ (scientific name: Artocarpus heterophyllus) has not been reconstructed by scholars thus far.

Possible cognates: adn toŋ/iˈfaŋˈtɔŋ; klz toŋ; kyo ʔəˈtɔːn; kpu N/A; kvd N/A; hmu N/A; abz soːŋ/ˈsɔŋ; sub ton/tɔn; pap sʊɔŋ; tiy ton; kira soni; kvw N/A; woi ton; tpg N/A; swt N/A.

Lexemes from (fieldworkers) used are adn (Robinson 2010; Choi 2015), klz (Robinson 2010), kyo (Choi 2015), abz (Saad 2015; Kratochvíl 2017), sub (Kratochvíl 2008; Kratochvíl 2017), pap (Delpada 2011), tiy (Delpada 2011), kira (Holton 2010), and woi (Schapper 2011).

The first issue present in these data is the prefix iˈfaŋ- in the second adn form gathered by Choi (2015) from the Otvai village. There appears to be compounding with -ˈtɔŋ. According to the specific wordlist for Otvai, one of the very few words with the affix -faŋ- include daˈrɛfaŋ, ‘sweet’. It is possible that the affix -faŋ- could be a morphological innovation describing the sweetness (characteristic) of the fruit.

The main observation from this dataset is the initial consonant. There are two prominent initial consonantal sounds, /t/ and /s/, among the following forms:

/t/ in adn toŋ/iˈfaŋˈtɔŋ klz toŋ kyo ʔəˈtɔːn sub ton/tɔn tiy ton woi ton;

/s/ in abz soːŋ/ˈsɔŋ pap sʊɔŋ kira soni.

We cannot account for this phonological change through Holton et al.’s (2012) s > t sound change theory, as they claim this change to be specific to east Alor languages (see Figure 5). The initial /t/ is, however, present in multiple languages scattered in different parts of the island.

With a lack of substantiated comparative data on the initial consonant, the regular practice for AP language researchers is to focus on Klon (kyo), which is believed to be the most linguistically ‘conservative’ and archaic AP Papuan language. kyo cognates are regularly referenced for the consonantal reconstruction of pAP proto-form, such as medial -g- in the reconstruction of AP root ‘banana’ *mogol, even though -g- is not omnipresent in the cognate datasets (Holton et al. 2012). In this set, ‘Jackfruit’ in kyo is ʔəˈtɔːn, making /t/ the most probable initial root consonant. For the vowel, there are some variations as well, mainly /o/, /ɔ/, and /ʊɔ/.

As /ɔ/ is not reconstructed to be a vowel by scholars in pTAP (Heston 2017), it is more plausible that /o/ is the original vowel. Similarly, for the final consonant, Holton et al. (2012) did not find /ŋ/ to exist in pAP based on their extensive analyses and reconstructions. A likely proto-form is thus pAlor *ton.

In summary, the pAlor form for ‘Jackfruit’ is postulated to be *ton. As kyo is believed by scholars to be the AP language to retain the most archaic features, the form ʔəˈtɔːn is referenced for the reconstruction of the initial consonant /t/. The vowel and consonant in the nucleus and coda position are based on the pAP consonant and vowel inventories reconstructed by Holton et al. (2012) and Heston (2017).

4.6. Bamboo

Holton et al. (2012) proposes *mari as the reconstructed pAP form for ‘Bamboo’ (scientific name: Bambusoideae).

Languages with no possible cognates: kpu N/A; hmu N/A; tiy N/A; kvw N/A; tpg N/A; swt N/A.

Possible cognates (Set 1): adn ˈpituŋ; klz peteŋ; kyo peːt; kvd pitan.

Possible cognates (Set 2): abz mai/ˈmɑːj; sub maj/mɑi; pap mɑɪ; kira mar; woi maːi.

Lexemes from (fieldworkers) used are adn (Choi 2015), klz (Robinson 2010), kyo (Choi 2015), kvd (Holton 2010), abz (Schapper 2010; Kratochvíl 2017), sub (Han 2014; Kratochvíl 2008), pap (Delpada 2011), kira (Holton 2010), and woi (Schapper 2011).

These are two sets of cognates that are clearly phonetically different from each other.

For set 1, adn ˈpituŋ, klz peteŋ, kyo peːt, and kvd pitan, these forms are found in the languages of western Alor. There are obvious sound correspondences among these cognates, such as initial consonant /p/, [e,i]/p_t, and medial consonant /t/. The suffixes adn -uŋ, klz -eŋ, kyo ∅, and kvd -an suggest that besides kyo, the other forms contain a suffix -V-C, with the final consonant being velar or alveolar nasal /ŋ/ or /n/. Delving deeper into the origins of this set of cognates, it is found that the word for ‘Bamboo’ in the Austronesian language Alorese is patuŋ (Sulistyono 2018), which appears to be a cognate to the set based on phonetic correspondences. It can therefore be postulated that the Papuan AP forms from Set 1 could have been borrowed from the Austronesian form.

Set 2 consists of some of the cognates analyzed by Holton et al. (2012), leading to the reconstruction of proto-root pAP *mari. The most relevant cognate they used for comparison is the form *mari from the Papuan Blagar language found in Pantar, a language that is not included in the dataset here, as it is beyond this study’s geographical scope. From the cognate data, the prefix root ma- appears to be consistent among the forms, likely retained from the original proto-form. Holton et al.’s (2012) reconstruction proposes a consonant reduction where /r/ is dropped in most of the Alor languages, except kira mar, where -r would be the supposed ‘relic’ of the original form. The kira form is not the only one in the current dataset that contains the /r/ phoneme proposed by Holton et al. (2012) to be part of the original root; the kira and kvd forms belong to the two different cognate sets. As mentioned, kira and kvd are languages that are classed to be similar and closely related. This indicates that the kvd lexeme could be borrowed from Austronesian, while the kira lexeme is presumably linked to an ‘original’ and ancient root in the AP Papuan languages.

Although pAP *mari is a plausible root-form, an alternative proto-form could be pAlor *mai. The root mai- seems to be relatively stable, as the form appears in multiple of the languages studied: abz mai, sub mɑi, pap mɑɪ, and woi maːi. As a medial /r/ consonant occurs in very few of the forms, it could have been a phonetic addition to the root at some point in history.

On the issue where kira mar appears to not contain the final vowel /i/, the lack of research on specific vowel patterns in most AP languages makes it difficult to postulate the possibility of vowel dropping. However, a study on Abui (abz) phonology found that words in the language with /l/ and /r/ phonemes ‘are articulated without friction and capable of being prolonged like a vowel’ (Delpada 2016, p. 55). As such, we can hypothesize that other AP languages could have these phonetic properties as well. kira, in particular, is located relatively near to the Abui populations, around the central part of the island. kira could have adapted the form from abz through contact, explaining why it deviates from the kvd form. Assuming this to be the case, the vowel quality of /r/ can explain a sound change *i > r:

kira mar: *i > r could have occurred in the coda position pAlor *mai > mar.

In summary, there are two proposed sets of cognates for ‘bamboo’. Cognates in set 1, adn ˈpituŋ, klz peteŋ, kyo peːt, and kvd pitan, are believed to have derived from an Austronesian loanword, evidenced by the existence of similar Austronesian forms representing ‘bamboo’, such as the Alorese patuŋ. The second set, abz mai/ˈmɑːj, sub maj/mɑi, pap mɑɪ, kira mar, and woi maːi, could have originated from the ‘original’ pAlor form, reconstructed as pAP *mari by Holton et al. (2012). An alternative reconstruction proposed in this study is pAlor *mai, assuming the occurrence of phonetic shift *i > r in the coda position of kira form mar.

4.7. Sweet Potato

The pAP root for ‘Sweet Potato’ (scientific name: Ipomoea batatas) has not been reconstructed by scholars thus far.

Possible cognates: adn falaj/faˈlac; klz N/A; kpu N/A; kvd ai_lak; hmu N/A; abz bɑˈleː; sub bilel/bɪlɛl; pap bɪlɪɛl; tiy bilel; kira ei_lak; kvw ai_lek; woi bileːl; tpg balune; swt N/A.

Lexemes from (fieldworkers) used are adn (Robinson 2010; Choi 2015), kvd (Holton 2010), abz (Kratochvíl 2017), sub (Han 2014; Kratochvíl et al. 2014), pap (Delpada 2011), tiy (Delpada 2011), kira (Holton 2010), kvw (Holton 2010), woi (Schapper 2011), and tpg (Williams 2010).

According to Perono Cacciafoco’s (Kratochvíl 2017) notes, Baleè is a generic form for ‘potato’, and not necessarily specific to ‘sweet potato’. Wordlists for most languages appear to solely contain the lemma ‘sweet potato’, as ‘sweet potato’ is often misunderstood or extensively associated with ‘potato’ in the Alor communities. In abz, the word bilèl means ‘young seed’, the ‘current/new moment of the sowing’. He further proposes that the abz form bɑˈleː could be linked to this concept of ‘young seed’ and that sub bilel/bɪlɛl, pap bɪlɪɛl, tiy bilel, and woi bileːl could be cognates as well.

It is impossible to know if the links between the concepts and the forms are applicable, and not simply phonetically similar but unrelated (non-cognates). The semantic link between ‘young seed’ and ‘sweet potato’ cannot be convincingly postulated with the sparse information available. A possibility could be that ‘sweet potato’ was considered a ‘young seed’, as it was introduced to the area likely only from the 16th century, originally being a Central American crop. Conversely, it could just be an abstract reference. These theories are impossible to confirm and still pose an enigma at this point in time.

Complicating this situation, ‘banana’ in abz is bɑˈlɛːj (Kratochvíl 2017), which differs from other AP cognates for ‘banana’, and the proto-form *mogul proposed by Holton et al. (2012) for ‘banana’ is unlikely to be phonetically related to the abz form. This may imply that the form for abz ‘young seed’; abz, sub, pap, tiy, woi ‘sweet potato’; and abz ‘banana’ could represent a metaphorical concept that cannot be properly determined.

Another observation from the data is that the peculiar forms kvd ai_lak, kira ei_lak, and kvw ai_lek appear to suggest the occurrence of metathesis between kvw and kira forms, as the /e/ and /a/ are transposed in the onset position of the first syllable and the nucleus position (vowel) of the second syllable. These three forms may be related to each other but not with the other lexemes in the set. Implausible phonetic changes, b > e and l > k, would have to have occurred to link the kvd, kira, and kvw with the rest of the forms. adn falaj/faˈlac could also be a non-cognate in the set, as among the languages studied, the established sound change *b > f only occurs in abz (Holton et al. 2012; see Figure 4) and is thus not applicable. As such, it is difficult to account for the sound changes in the adn form.

At a purely phonetic level, excluding the kira, kvw, and adn forms, the actual cognates should be as follows:

abz bɑˈleː; sub bilel/bɪlɛl; pap bɪlɪɛl; tiy bilel; woi bileːl; tpg balune.

A proto-form can be postulated at a phonetic level, possibly pAlor *bile(l); *b is a plausible ancient root consonant and is also present in all the cognates. Vowels *i and *e requires more of an assumption but studies on recurrent vowel sound correspondences by Heston (2017) found *i and *e to have likely existed in pTAP. The medial /l/ is also supported by Holton et al.’s (2012) pAP reconstruction. The final /l/ is also supported by Holton et al. (2012). However, it is not present in all the cognates within the dataset, such as abz bɑˈleː and tpg balune, which do not contain a final /l/. Thus, we cannot be certain of the vocalic addition of -(l) in the reconstructed form.

In summary, abz bɑˈleː, sub bilel/bɪlɛl, pap bɪlɪɛl, tiy bilel, and woi bileːl are likely related cognates. However, similar forms are found in words representing other concepts in languages such as abz bilèl ‘young seed’ and abz bɑˈlɛːj ‘banana’. Therefore, it is difficult to postulate the meaning of this form, as it could represent some specific characteristic of the fruit or a metaphorical comparison, instead of the concept of the fruit itself. A proto-form can still be reconstructed, however, in pAlor *bile(l) based prevalence of the phonetic sounds and cross-checking with reconstructed consonants in pAP (Holton et al. 2012) and vowels in TAP (Heston 2017).

4.8. Gum Tree/Eucalyptus

The pAP root for ‘Gum Tree/Eucalyptus’ (scientific name: Eucalyptus globulus) has not been reconstructed by scholars thus far.

Possible cognates: adn pu; klz puʔ; kyo N/A; kpu N/A; kvd pokoil; hmu N/A; abz puokdai/puokal; sub foː bɔŋ; pap pʊkɑɪ_bɑtaː; tiy N/A; kira pokul; kvw N/A; woi foo; tpg N/A; swt N/A.

Lexemes from (fieldworkers) used are adn (Robinson 2010), klz (Robinson 2010), kvd (Holton 2010), abz (Schapper 2010; Saad 2016), sub (Kratochvíl 2017), pap (Delpada 2011), kira (Holton 2010), and woi (Kratochvíl 2017).

The initial consonant likely went through a sound change, *p > f, in sub foː bɔŋ and woi foo. Besides the significantly greater prevalence of /p/ in the cognates compared to /f/, this theory is supported by Holton et al.’s (2012) findings that pAP *p remains unchanged in initial positions for most of the languages they studied, except for woi, in which they found the tendency for /p/ to weaken to /f/. Although there is an exception in sub fo, this theory remains the most plausible.

In the root pu- found in adn pu, klz puʔ, abz puokdai/puokal, pap pʊkɑɪ_bɑtaː, and kira pokul, the vowel /u/ could have gone through vowel lowering u > ʊ in pap pʊkɑɪ_bɑtaː and u > o, accounting for the forms kvd pokoil and kira pokul. The proto-form could thus be pAlor *puk- with the /k/ dropped in adn pu and klz puʔ.

Regarding the additional suffixal morphemic parts of the lexemes, as the word for ‘tree’ in pap is bɑtaː (Delpada 2011), the suffixal root -bata in pap pʊkɑɪ_bɑtaː is likely a morphological innovation representing ‘gum’ and ‘tree’. Similarly, for the sub form foː bɔŋ, -bɔŋ is ‘tree’ in sub (Kratochvíl 2017). However, with the lack of data for comparison, meanings for the other morphological innovations, kvd -oil, abz -dai/-al, and kira -ul, are still a mystery and cannot be accurately derived.

In summary, the pAlor form reconstructed in this study is *puk-. The initial consonant *p > f in sub foː bɔŋ and woi foo is more probable from a phonetic standpoint and based on the prevalence of /p/ amongst the forms. The final consonant /k/ is postulated to have dropped in adn pu and klz puʔ. Some morphological innovations are found in pap, sub, kvd, abz, and kira. The pap and sub roots are known innovations meaning ‘tree’ in the respective languages. However, the kvd, abz, and kira innovations remain a mystery. Further research can be conducted to explain the purpose, linguistic patterns, and meaning of these compounding innovations.

4.9. Uncooked Rice/Rice Plant

The pAP root for ‘Uncooked Rice/Rice Plant’ (scientific name: Oryza sativa) has not been reconstructed by scholars thus far.

Possible cognates: adn ˈala; klz haːlaʔ; kyo ərɑk/ʔaˈrak_ʔəˈniŋ; kpu N/A; kvd arak; hmu ala; abz ajak/aʋek; sub aː; pap ɛk; tiy aː; kira arak; kvw aˈrak/araˈkɛs; woi aa beita; tpg aˈjak_ˈgəs; swt araka_ˈtana/ˌʔɑrɑˈkeːs.

Lexemes from (fieldworkers) used are adn (Choi 2015), klz (Robinson 2010), kyo (Baird 2003; Choi 2015), kvd (Holton 2010), hmu (Baird 2003), abz (Schapper 2010; Saad 2016), sub (Han 2014), pap (Delpada 2011), tiy (Delpada 2011), kira (Holton 2010), kvw (Holton 2010, Choi 2015), woi (Kratochvíl 2017), tpg (Choi 2015), and swt (Choi 2015; Kratochvíl 2017).

‘Rice’ is one of the lexemes that appear the most in wordlist besides ‘banana’, providing great potential for analysis. Cooked versus uncooked rice may often be represented by different words and forms. Here, specific words for ‘uncooked rice’ or ‘rice plant’ are studied.

The first observation from the cognates is that most of the forms have initial consonant /a/, except klz haːlaʔ and kyo ʔaˈrak_ʔəˈniŋ with their /h/ and /ʔ/ initial consonants, respectively. Bearing in mind that kyo retains rather ancient forms, and as the data from the two languages are retrieved from villages relatively close to each other in west Alor, the proto-form may likely contain an initial consonant before the vowel. /h/ is the most plausible, with the glottal stop assumed as a phonetic innovation in the AP languages as mentioned previously (Holton et al. 2012). Thus, the initial consonant sound change would be *h > ʔ in kyo and other languages would have dropped the /h/: *(h) > ∅.

A pattern seen is the prevalence of medial consonants /l/ and /r/ in the following languages:

/l/: adn ˈala; klz haːlaʔ; hmu ala.

/r/: kyo ərɑk/ʔaˈrak_ʔəˈniŋ; kvd arak; kira arak; kvw aˈrak/araˈkɛs; swt araka_ˈtana/ˌʔɑrɑˈkeːs.

Although a medial */r/ or */l/ is plausible (Holton and Robinson 2014), *l is likely to be more ancient. This process of rhotacism in *l > r is hypothesized by Perono Cacciafoco and Cavallaro (2018) to have occurred in abz. An example is the abz verb aral (‘burn off’), which was analyzed by researchers to have derived from*alal (Kratochvíl and Delpada 2014). As such, the proposed pAlor root would be *(h)ala-, with forms like the kyo ərɑk/ʔaˈrak_ʔəˈniŋ, kvd arak, abz ajak, kira arak, kvw aˈrak, and tpg aˈjak_ˈgəs undergoing paragoge and gaining a final consonant /k/.

As seen from the possible cognates listed, there are multiple words with clear morphological innovations, such as kyo -ʔəˈniŋ, kvw -ˈkɛs, woi -beita, tpg -ˈgəs, and swt -ˈtana/-ˈtana. The meaning and purpose of these morphological innovations remain unknown.

In summary, the proposed reconstruction of ‘rice’ is *(h)ala-, as the initial consonant could have been dropped in most languages but retained in klz, and gone through the sound change *h > ʔ in kyo. The process of the rhotacism *l > r is postulated to have occurred in the medial consonant, and the kyo, kvd, abz, kira, kvw, and tpg forms could have gone thorough paragoge and gained final consonant /k/. Some obvious morphological innovations in the kyo, kvw, woi, tpg, and swt forms remain unexplained but provide potential for future research on the morphological uses of these suffixes.

5. Conclusions and Future Research

To summarize, an impressive body of lexical work on the Alor–Pantar languages has been produced thus far by researchers through rigorous data collection fieldwork. However, historical relations amongst the AP Papuan languages are still inadequately established. A particular field that is in urgent need of greater emphasis and focus by researchers is the study of word origins and etymologies, such as the reconstruction of ancient root-forms and the study of phonetic correspondences among the languages. Early research centered on AP words that described basic universal concepts of time and environment features like ‘sky’ or ‘moon’, which was covered by Perono Cacciafoco and Cavallaro (2018) in their work ‘Roots from the Archipelago: Proto-Alor Etymologies in the Context of Alor Languages’. The current study focuses on expanding the range of study, focusing on another stable and important aspect of the environment and livelihood of the local Alor people—the names of plants (botanical names). Many plants are believed to be sacred by the locals, linked widely to local legends, and used in traditional herbal remedies, which attests to the ancient nature of these names.

This study utilizes the diachronic Comparative Method to compare sampled forms from fifteen languages across Alor Island and provide hypothesized proto-root reconstructions based on historical phonetic (sound change) analysis, and a critical analysis of existing reconstructions from past scholars. Sociocultural theories are also proposed from the data insights, such as prehistoric population contact (Papuan and Austronesian communities) and migration/settlement dynamics.

This research attempts to be as accurate and comprehensive as permitted by the available data. However, as the AP languages are still not well documented, many linguistic forms and properties remain unexplained or speculative. Future research should focus on conducting further comparative etymological work on a wider range of words and concepts, based on the current body of lexemes collated from the AP languages (or additional data from future fieldwork), to provide further comparison and to confirm the reconstructed forms proposed in this study.