# An Investigation of Relative Clauses in Hla'alua
## Part 1: Literature Review and Research Background
* **Presenter:** LIN CHIH-HAN
* **Date:** 2025-05-27
---
### Introduction - Importance of Relative Clauses
* **Relative Clauses (RCs): Universality and Complexity**
<details>
<summary>Definition of RCs</summary>
* An RC is a subordinate clause that <mark>delimits the reference of a noun phrase (NP) by specifying the role</mark> of its referent in the situation described by the RC (Andrews 2007: 206).
</details>
<details>
<summary>Universality and Diversity of RCs</summary>
* RCs are universal across languages, yet display high variability in their structural types, common argument (CA) identification and gap strategies, and verbal morphology.
</details>
<details>
<summary>Goal of this Research</summary>
* This research aims to delve into the grammar of RCs in Lha'alua, engaging in dialogue with existing typological and Formosan language literature.
</details>
---
### Research Background - Voice Systems in Austronesian Languages
* **Austronesian Languages: A Paradigm for Voice Systems**
<details>
<summary>Overview of Austronesian Voice Systems</summary>
* **Austronesian languages** (including Formosan languages in Taiwan's context) are renowned for their distinct **Voice Systems**, where verbal morphology is intricately linked to <mark>the semantic role of the grammatical subject</mark> (Schachter 1976; 1987; Paul Jen-kuei Li 1997a).
* **RC-Voice System Interaction**: In many Austronesian languages, the verb morphology within RCs (especially their voice markers) often **aligns with** the semantic role of the Common Argument (CA) in the RC.
</details>
<details>
<summary>Typical Voice Types</summary>
* Typical **Voice** types include: **Actor Voice (AV)**, **Undergoer Voice (PV、LV、CV)**, which determine which semantic role is promoted to grammatical subject.
</details>
---
### Typological Perspectives on Relative Clauses - Andrews (2007) Distinction
* **Andrews (2007) Typology of Relative Clauses**
Andrews (2007) classifies embedded relative clauses into three main types based on the structural relationship between the relative clause(RC) and the modified noun(CA) phrase (Andrews 2007: 208):
<details>
<summary>External RCs</summary>
* **<mark>External RCs</mark>**: RC appears outside the CA. Can be **Post-nominal** or **Pre-nominal**.
<details>
<summary>Examples</summary>
* English Post-nominal: "The book [I bought yesterday]"
* Japanese Pre-nominal: "[Yamada-san ga kat-te i-ru] saru" - ヤマダさんが飼っている猿
</details>
</details>
<details>
<summary>Internal RCs</summary>
* **<mark>Internal RCs</mark>**: RC contains the CA within itself (Andrews 2007: 212). For instance, in Navajo, the CA is embedded within the RC, and the entire RC functions as an argument of the main clause.
<details>
<summary>Examples</summary>
* Navajo internal:
```lha'alua
[(Tl’´e´ed˛a˛´a) ashkii al´-h˛a˛´a’-˛ ´ a˛´a] y´adool-tih
last.night boy 3sg.imperf.snore-rel.past fut.3sg.speak
‘The boy who was snoring last night will speak’
```
</details>
</details>
<details>
<summary>Free RCs</summary>
* **<mark>Free RCs</mark>**: RC itself constitutes an NP, possibly lacking an overt domain nominal (Andrews 2007: 213). (e.g., English "what the cat left in its bowl").
</details>
---
### Typological Features of Austronesian RCs - Himmelmann (2005) Insights
* **Himmelmann (2005) Overview of Austronesian RCs**
<details>
<summary>Subject Properties</summary>
* **Subject Properties**: The main argument in RCs (i.e., the CA within the RC) typically exhibits subject-like properties, especially prominent in Philippine-type focus systems.
</details>
<details>
<summary>Linkers</summary>
* **Linkers**: Many Austronesian languages use **<mark>linkers</mark>** (e.g., Tagalog's `na` or `ng`) to connect RCs with their head nouns. These linkers may also appear between **adjectival** phrases and head nouns, suggesting an 'adjectival' function for RCs (Himmelmann 2005: 229, ex. 60).
</details>
<details>
<summary>CA Gapping</summary>
* **CA Gapping**: The CA's position within the RC is commonly realized as a **<mark>gap</mark>**, rather than an overt relative pronoun.
</details>
<details>
<summary>Word Order</summary>
* **Word Order**: Austronesian RCs display diverse word orders, including both pre-nominal and post-nominal types. <mark>Philippine-type languages often prefer RCs to follow the head noun.</mark>
</details>
---
### Lha'alua RC Literature Review - Pan (2012) Stance
* **Pan (2012) Description of Relative Clauses in Lha'alua**
<details>
<summary>Main Classification and Unspecified Types</summary>
* Pan's (2012) earlier PhD thesis provides a comprehensive grammatical description of Lha'alua. While it refers to Andrews' (2007) typology, its primary analysis of Lha'alua RCs focuses on **<mark>External RCs</mark>**, particularly **Post-nominal External RCs**, as the dominant type. The thesis did not explicitly detail the widespread presence of Internal RCs or Free RCs.
</details>
<details>
<summary>Core Findings: Semantic Role's Impact on Morphology</summary>
* **Core Findings**: Pan (2012) emphasizes that the **<mark>semantic role</mark>** of the CA in the RC (Actor vs. Non-Actor) impacts the **<mark>morphology</mark>** of the RC verb:
* **Actor CA**: The RC verb **does not** take the `-isa` marker.
* **Non-Actor CA** (Patient): The RC verb **takes** the **<mark>-isa marker</mark>**.
</details>
<details>
<summary>Linkers and Gapping Strategy</summary>
* **Linkers**: `a` or `ka` are used to connect the head noun and the RC.
* **Gapping**: The CA typically appears as a gap within the RC.
</details>
---
### What this Research want to do?
This study aims to contribute to the typology of Austronesian relative clauses by enriching and refining existing theories with empirical data from Lha'alua.
*
To validate many findings from Pan (2012) regarding External Post-nominal RCs and the `-isa` marker's distribution.
* **Examine the order of CA and RCs (Post or Pre-nominal)**
* **the `-isa` Marker Rule and Semantic Subtypes**
* **RC Type Distribution and Semantic Role Correlation**: This research will further investigate the conditions (especially based on CA's semantic role) under which Lha'alua chooses between External Post-nominal RC or Internal RC, and explore potential interchangeability.
---
## Part 2: Empirical Analysis and Findings
---
### Methodology
* **Data-Based Analysis**
<details>
<summary>Data Collection</summary>
* Primary data collected through 4 days fieldwork in Lha'alua-speaking villages.
* Data consists of transcribed and glossed texts, spontaneous speech, and elicited sentences.
* Total data size: about 25 sentences.
* Key language consultants: 游仁貴(Amalanamahlʉ )、宋玉清(Mai)、游枝潔(Vanau).
</details>
<details>
<summary>Analytical Approach</summary>
* **RC identification** based on formal syntactic criteria (modifies a noun, contains a gap/bound argument, specific linkers).
* **Classification of RCs** into types (External Post-nominal, Internal, etc.).
* Detailed morphological analysis of **RC verb forms** (e.g., presence/absence of `-isa`, voice markers, aspect).
* Semantic role assignment for CAs within RCs.
* Quantitative analysis of distribution patterns and exceptions.
</details>
---
### External Post-nominal RCs: The Dominant Type
* **Structure:** CA + RC (RC follows the head noun).
* **Key Observation:** This is the **<mark>most common RC type</mark>** observed in the data.
<details>
<summary>Pattern 1: Actor CA in External Post-nominal RCs</summary>
* **Rule:** RC verb **does NOT** take the `-isa` marker when CA is Actor.
* **Lha'alua Example (Day 3, S5):**
```lha'alua
m-aruka=cu a cucu’u=na [hli=arivakʉsʉ tasau]RC
AV-got.lost=PERF CORE person=DEF [EXP=beat(AV) dog]RC
```
* **Translation:** 'The person who beat the dog is gone.'
* **Analysis:** CA (`cucu’u=na`, Actor) is external. RC verb (`arivakʉsʉ`) is Actor Voice and **lacks `-isa`**.
* **Lha'alua Example (Day 1, S18):**
```lha'alua
‘Avi [ku=aiipʉ]RC k<um>a-ki-kita ‘ahlingu
‘Avi [NEG=take.course(AV)]RC <IRR>-RED-see TV
```
* **Translation:** 'Avi [who doesn't study] is watching TV.'
* **Analysis:** CA (`‘Avi`, Actor) is external. RC verb (`ku=aiipʉ`) is Actor Voice and **lacks `-isa`**.
* **Consistency:** This pattern aligns with Pan (2012)'s findings and general Austronesian patterns.
</details>
<details>
<summary>Pattern 2: Non-Actor CA in External Post-nominal RCs</summary>
* **Rule:** RC verb **typically takes** the **<mark>-isa marker</mark>** when CA is Patient or Locative.
* **Lha'alua Example (Day 2, S1 - Patient CA):**
```lha'alua
hli=um-u=aku vutukuhlu [hli=pu’a(UV)=isa ‘Avi=na]RC
EXP=AV-eat=1SG.NOM fish [EXP=buy(UV)=3SG.GEN Avi=DEF]RC
```
* **Translation:** 'I ate the fish that ‘Avi bought'
* **Analysis:** CA (`vutukuhlu`, Patient) is external. RC verb (`hli=pu’a=isa`) **has `-isa`**.
* **Lha'alua Example (Day 2, S4 - Patient CA+LV):**
```lha'alua
maruka pahlivinga [hli=ala-ana=aku tavava-ana]RC
got.lost thing [EXP=take-LOC=1SG.NOM kitchen-LOC]RC
```
* **Translation:** 'The thing I brought from the kitchen is missing.'
* **Analysis:** CA (`pahlivinga`, Patient) is external. RC verb (`ala-ana`) **has `LV` and `aku`**.
* **Consistency:** This pattern largely supports Pan (2012)'s core finding regarding `-isa` marking for Non-Actor CAs.
</details>
---
### External Pre-nominal RCs:
* **Structure:** RC + CA (RC precedes the head noun).
* **Key Observation:** This order is often interchangeable with the post-nominal one, except in <mark>complicated sentence structures, where the speaker tends to prefer the pre-nominal form.</mark>
<details>
<summary>Pattern 1: Actor CA in External Pre-nominal RCs</summary>
* **Rule:** RC verb **does NOT** take the `-isa` marker when CA is Actor.
* **Lha'alua Example (Day 1, S14):**
```lha'alua
[hli=um-a-ruca-rucakʉ ihlaku]RC Caʉpu=na m-usala=cu
[EXP=AV-STAT-RED-angry 1SG.INDP]RC Caʉpu=DEF AV-go=PERF
parana mai-savu-savu-anʉ
palce AV-hand.action-RED-medicine.grass
```
* **Translation:** 'The Caʉpʉ [who scolded me] went to hospital.'
* **Analysis:** CA (`Caʉpu=na`, Actor) is external. RC verb (`m-usala=cu`) is Actor Voice and **lacks `-isa`**.
* **Lha'alua Example (Day 1, S18):**
```lha'alua
ausi ku=asasangarʉ a [hli=p-ausahlʉ]RC Vanau=na
seem NEG=happy CORE [EXP=UV-get.wet] Vanau=DEF
```
* **Translation:** Vanau, who got caught in the rain, seems unhappy.'
* **Analysis:** CA (`Vanau`, Actor) is external. RC verb and **lacks `-isa`**.
* **Consistency:** This pattern aligns with Pan (2012)'s findings and general Austronesian patterns.
</details>
<details>
<summary>Pattern 2: Non-Actor CA in External Pre-nominal RCs</summary>
* **Rule:** RC verb **typically takes** the **<mark>-isa marker</mark>** when CA is Patient or Locative.
* **Lha'alua Example (Day 1, S19 - Patient CA):**
```lha'alua
[hli=pu’a=isa ‘Avi=na]RC vutukuhlu apa-u
[EXP=buy(UV)=3SG.GEN ‘Avi=DEF]RC fish CAUS-eat
```
* **Translation:** 'The fish [which Avi bought] was eaten.'
* **Analysis:** CA (`vutukuhlu`, Patient) is external. First RC verb (`hli=pu’a=isa`) **has `-isa`**. (Note: Second RC `apa-u` is an exception discussed later).
* **Lha'alua Example (Day 2, S7 - Locative CA):**
```lha'alua
alua ihlaku a [hli=avava=isa ‘Avina]RC camai
alu-a ihlaku a [hli=avav-a=isa ‘Avi=na]RC camai
```
* **Translation:** ' I took the food that ‘Avi cooked'
* **Analysis:** CA (`camai`, Patient) is external. RC verb (`avav-a`) **has `-isa`**.
* **Consistency:** This pattern largely supports Pan (2012)'s core finding regarding `-isa` marking for Non-Actor CAs.
</details>
---
### Unveiling Internal RCs in Lha'alua
* **Challenging the Mainstream View:** Empirical Evidence for Internal RCs in Lha'alua.
<details>
<summary>Definition and Identification</summary>
* Internal RCs embed the head noun (CA) **within** the relative clause itself, with the entire RC structure functioning as an argument in the main clause (Andrews 2007).
* <mark>This contrasts with Pan (2012)'s primary focus on External RCs in Lha'alua.</mark>
</details>
<details>
<summary>Lha'alua Empirical Evidence </summary>
* **Lha'alua Example: (Day 2, S1, by Vanau)**
```lha'alua
hli=um-u=aku [hli=pu’a(UV)=isa vutukuhlu ‘Avi=na]RC
EXP=AV-eat=1SG.NOM [EXP=buy(UV)=3SG.GEN fish ‘Avi=DEF]RC
```
* **Translation:** 'I ate the fish [which Avi bought].'
* **Analysis:**
* The CA `vutukuhlu` (fish) is clearly *inside* the bracketed RC `[hli=pu’a(UV)=isa vutukuhlu ‘Avi=na]`.
* The entire RC functions as the direct object of the main clause verb `hli=um-u=aku` (ate).
* CA (`vutukuhlu`) is a **Patient** in the RC event (`buy`).
* The RC verb `hli=pu’a(UV)=isa` **carries the `-isa` marker**, consistent with Non-Actor CAs.
* **Other examples:**
* iniahlukua [hliumaracʉ **tasauna** ihlaku]RC (051516)
* sanʉcuaku [hlipu’aisa **vutukuhlu** ‘Avina]RC (051704)
* marukacu a [hliarivakʉsʉ **cucu’una** tasau]RC (0517 05)
* **Significance:** This provides **<mark> empirical evidence</mark>** for the presence of Internal RCs in Lha'alua, enriching its typological profile.
</details>
### Exploring RC Type Distribution and Interchangeability
* **Factors Influencing RC Type Choice:** This section presents observed patterns regarding RC type choice in Lha'alua. Given the limited data size, these are presented as **<mark>preliminary observations and hypotheses for future research</mark>**.
<details>
<summary>Semantic Role of CA and Observed RC Type Distribution</summary>
* There is **no clear** distribution between **Actor CA**、**Patient CA**、**Locative CA** with respect to **Pre/Post nominal order**
Quantitative Summary of RC Types:
| RC Type | Number of Instances | Example IDs |
| :-------------------- | :-------------------- | :---------------------------------- |
| Pure Pre-nominal | 3 | 051514, 051813, 051815 |
| Pure Post-nominal | 4 | 051603, 051604, 051617, 051807 |
| Pre/Post Interchangeable | 3 | 051518, 051519, 051802 |
| Pure Internal | 1 | 051704 |
| Post/Internal Interchangeable | 2 | 051601, 051705 |
| Pre/Post/Internal Interchangeable | 1 | 051516 |
| **Total Pre-nominal Occurrences** | 7 | |
| **Total Post-nominal Occurrences** | 10 | |
| **Total Internal Occurrences** | 4 | |
Detailed Analysis of RC Examples by Type, Actor Status, and Voice:
| Example ID | RC Type | Actor Status of CA | Voice |
| :--------- | :-------------------------- | :----------------- | :-------------------- |
| 1-1 | Pre-nominal | Actor = Subject | AV |
| 2-1 | Pre/Post/Internal | Actor = Subject | AV |
| 3-1 | Pre/Post | Actor = Subject | AV |
| 4 | Pre/Post | Actor = Subject | PV |
| 4-1 | Internal/Post | Actor != Subject | PV |
| 4-2 | Internal | Actor != Subject | PV |
| 5-1 | Post-nominal | Actor != Subject | PV |
| 5-2 | Post-nominal | Actor != Subject | PV, LV |
| 7-2 | Post-nominal | Actor != Subject | PV |
| 8 | Post/Internal | Actor = Subject | AV |
| 9-1 | Pre/Post | Actor != Subject | PV |
| 11 | Pre-nominal | Actor != Subject | PV |
| 12 | Pre-nominal | Actor = Subject | AV |
</details>
<details>
<summary>Future Directions</summary>
* **Future Research**:
* Investigate discourse prominence, **new vs. old information status**, or other pragmatic factors that might drive the choice between these RC types.
* Collect a **<mark>larger and more diverse corpus</mark>** to establish statistically significant distribution patterns.
* Conduct controlled elicitation tasks to test the precise conditions for different RC types.
</details>
---
### Other finding during this research:
* **Locative Voice:**
<details>
<summary>Definition and Identification</summary>
* In Lha’alua, the locative voice encodes a nominal argument with the semantic role of location and **designates it as the grammatical subject**(cf. Blust 2009). It is morphologically <mark>marked on the verb by -a(na), -i, or -ani.</mark> Compared to the actor and patient voices, instances of the locative voice are relatively infrequent in textual data.
</details>
<details>
<summary>The rare examples in my data:</summary>
* **Lha'alua Example: (Day 2, S14, by Mai)**
```lha'alua
maruka pahlivinga hli=ala-**ana**=aku tavava-ana
got.lost thing EXP=take-**LOC**=1SG.NOM kitchen-LOC
```
* **Translation:** 'The thing I brought from the kitchen is missing.'
* **Analysis:**
* The verb ala is marked with the locative voice suffix -ana, introducing a new locative argument into the sentence.
* **Potential hypothesis**:
* In Lha’alua, the -ana marker has been reduced to the suffix -a.
* **Lha'alua Example: (Day 4, S12, by Amalanamahlʉ)**
```lha'alua
a-vur-**i** amahlʉ=na tʉpʉhlana suhlatʉ
IRR–**LV** amalhʉ=DEF measure.word book
```
* **Translation:** 'I want you to give it to him.'
* **Analysis:**
* This verb is marked with the suffix -i, as it introduces one additional participant.
* The locative voice marker predominantly appears in nouns, such as *tavavana* (kitchen), *ia’aana* (shop), *anana* (snack bar), and *tanana* (restaurant).
</details>
---
### Conclusion and Future Work
* **Key Findings Summarized:**
<details>
<summary>Summary of Main Findings</summary>
* Lha'alua predominantly uses External Post-nominal RCs.
* The `-isa` marker robustly distinguishes Non-Actor CAs in many RCs.
* Crucially, **<mark>Internal RCs exist</mark>**, enriching Lha'alua's typological profile.
</details>
<details>
<summary>Future Research Directions</summary>
* Conduct a comprehensive, quantitative analysis of all RC types in a larger corpus to establish precise distribution patterns.
* Explore the pragmatic and discourse factors influencing the choice between different RC types (e.g., Internal vs. External Post-nominal).
* Delve deeper into the nature of `-isa` exceptions and analyze their syntactic/semantic properties to determine if they constitute distinct clause types.
* Investigate the diachronic development of RC strategies in Lha'alua and other Austronesian languages.
</details>
---
## References
- Andrews, Avery D. 2007. *Relative clauses*. In *Language Typology and Syntactic Description*, Volume II: *Complex Constructions*, 2nd edition, ed. by T. Shopen, pp. 206–236. Cambridge: Cambridge University Press.
- Himmelmann, Nikolaus P. 2005. Chapter 5: *The Austronesian languages of Asia and Madagascar: typological characteristics*. In *The Austronesian Languages of Asia and Madagascar*. New York: Routledge. (Section 3.3–Section 6)
- Pan, Chia-jung. 2012. *A Grammar of Lha’alua, an Austronesian Language of Taiwan*. Ph.D thesis, James Cook University, Australia.
- Ross, Malcolm. 2009. *Proto Austronesian verbal morphology: A reappraisal*. In Adelaar, Alexander and Andrew Pawley (eds.), *Austronesian Historical Linguistics and Culture History: A Festschrift for Robert Blust*, pp. 295–326. Pacific Linguistics 601. Canberra: The Australian National University.
- 宋麗梅、鄭奕揚。2021。〈第 7 章台灣南島語〉，收於黃宣範主編《語言學：結構、認知與文化的探索》，頁 173–204。台北市：國立臺大出版中心。
- 潘家榮。2018。《拉阿魯哇語語法概論》。語法概論系列，原住民族委員會出版。
[https://alilin.cip.gov.tw/Book/569](https://alilin.cip.gov.tw/Book/569)
<style>
/* Styling for better readability */
h1, h2, h3 {
color: #2E86C1;
font-family: Arial, sans-serif;
border-bottom: 2px solid #ddd;
padding-bottom: 5px;
}
h1 {
font-size: 28px;
}
h2 {
color: green;
font-size: 24px;
}
h3 {
font-size: 20px;
}
p {
font-size: 16px;
line-height: 1.6;
}
code {
background-color: #f4f4f4;
padding: 3px 5px;
border-radius: 4px;
}
blockquote {
font-style: italic;
color: #7B7D7D;
border-left: 4px solid #3498DB;
padding-left: 10px;
}
</style>
Sign in with Wallet
Connect another wallet