Towards a new typology of comparative constructions in East Asian languages

3.1 Definition

A double-comparatum (double-compm) comparative construction involves two compee-like NPs, namely the comparatum (compm) and the attribute comparatum (attr-compm). The compm refers to the NP that is compared to the st. In contrast, the attr-compm is the NP that possesses a specific attribute, often described by an adjective or more specifically, by p in the comparative construction. In many languages, the two concepts are expressed by the same NP, the compee. For example, in the English sentence The price of this book is higher than that of another, the compm is the price of this book, which is compared to the st that of another. Meanwhile, The price of this book is also the attr-compm since it is being described by the adjective high. In this case, the compee can be treated as the combination of the compm and the attr-compm. In MSM, however, the two concepts are distinguishable,^[7] as illustrated by (2) above. In essence, by considering the degree of relatedness, the compm, attr-compm, st, and p can be grouped into two pairs: compm-st and [attr-compm]-p.

It should be emphasized that the conceptual distinction between compm and attr-compm is not an original contribution of this paper. When discussing “比”字差比句 ‘bǐ-comparatives’ in MSM,^[8] Liu (2012) identifies two types of compee-like NPs, which he labels as 比较主体 bǐjiào zhǔtǐ ‘compared subject’ and 属性主体 shǔxǐng zhǔtǐ ‘attribute subject’. These two types of zhǔtǐ can co-occur in what he terms as 双主体差比句 shuāng zhǔtǐ chābǐjù ‘double-subject comparatives.’ However, the use of “subject” to translate both 主体 zhǔtǐ ‘main body’ and 主语 zhǔyǔ ‘(grammatical) subject’ can lead to ambiguity in English. While zhǔtǐ ‘main body’ refers to the central or principal entity, it may not necessarily correspond to the grammatical subject of a sentence. In order to avoid this confusion, this study adopts the terms “comparatum” and “attribute comparatum” to represent Liu’s ‘compared subject’ and ‘attribute subject’, respectively. Consequently, Liu’s ‘double-subject comparatives’ are denoted as “double-comparatum comparatives” in this research.

In double-compm comparatives, the position of compm and attr-compm can vary. As shown in (2), the compm can appear before the attr-compm. Alternatively, the attr-compm may precede the compm (4a) or immediately precede the p (4b). Example (4c) is an atypical case in which the compm and st have a mismatch: The former is a non-possessive NP, while the latter is a possessive NP. These “compm-st mismatched comparatives” are considered not only a subtype of double-compm comparative constructions, but also a subtype of the “compee-st mismatched comparatives”, which will be discussed in Section 4.

It should be noted that the constructions like (5) are not double-compm comparatives.

Example (5) is a typical MSM bǐ-comparative construction (with a headless possessive NP). In this sentence, 这家店的 zhè jiā diàn de ‘this store’s’ is a headless possessive NP, referring exactly to “(this store’s) price”. Accordingly, zhè jiā diàn de and jiàgé are not differentiated as two types of comparatums, and the sentence is thus not a double-compm comparative construction discussed in this paper.

compm and attr-compm typically exhibit a semantic possessive relationship in which the compm is the possessor and the attr-compm is the possessee. This can be illustrated via adding the possessive marker (e.g., de in MSM) between compm and attr-compm. However, as indicated in footnote 4, it is important to note that they do not form a syntactic possessive relationship nor do compm-[attr-compm] function as a single NP.

3.2 The topicality of double-compm comparatives

“Topic” is a pragmatic concept that pertains to the theme or subject matter of a sentence or discourse. Certain languages, referred to as “topic-prominent” languages (Li and Thompson 1976, 1981), have morpho-syntactically marked grammatical categories for sentence topics. For instance, MSM (as well as other Sinitic varieties,^[9] especially Wu) is commonly recognized as topic-prominent, where topic is considered a grammatical notion (see, among others, Shi 2000; Xu and Liu 2007); Similarly, Japanese and Korean employ dedicated topic markers that are distinct from subject markers. The definition of the topic as a descriptive category (Haspelmath 2010) varies across languages, with its status as a grammatical notion not being the focus of this paper. Instead, this study employs “topic” as a comparative concept (Haspelmath 2010) which is defined as: A topic of a sentence is what the sentence is about or “set up to add new information to” (Shi 2000: 386).

From the definition, two notable characteristics of topic emerge that are relevant to the discussion in this paper. Firstly, a sentence can have more than one topic. Secondly, a topic does not necessarily correspond to the (grammatical) subject (Lambrecht 1994). These characteristics can be illustrated through example (6) from MSM.

In terms of the number of topics, (6a) has two interpretations. Reading (i) contains one topic, 大象 dàxiàng ‘elephant’, and the remainder of the sentence constitutes the comment that adds new information to the topic; In reading (ii), there are two topics: dàxiàng ‘elephant’ serves as the primary topic, and 鼻子 bízi ‘nose’ the secondary topic (Xu and Liu 2007). As noted above, a sentence can contain more than two topics; however, this aspect is beyond the scope of this paper. Example (6b) shows that a topic does not necessarily align with the grammatical subject of a sentence. In this example, the topic 那场火 nèi chǎng huǒ ‘that fire’ lacks a clear thematic role in relation to the predicate 来得快 lái de kuài ‘come quickly’, as their semantic relationship is distant. Rather, a subtle “pragmatic relation” (Lambrecht 1994) exists between the topic ‘that fire’ and the comment 幸亏消防队来得快 ‘fortunately the fire brigade came quickly’.

On the one hand, it is noteworthy that an NP can fulfill dual roles as both the topic and the subject. For instance, in the sentence John went to school, the NP John serves as the subject and can also assume the role of the topic when the sentence is about John. On the other hand, as a sentence typically has only one subject, in sentences with two subject-like NPs preceding the predicate, often only one of the NP functions as the subject, while the other assumes the role of a topic. Xu and Liu (2007) present an operational definition that delineates this distinction, positing that within sentences featuring two subject-like NPs, other things being equal, the NP that bears a direct semantic or argument relationship with the predicate tends to be deemed as the subject, whereas the NP lacking such a relationship tends to be analyzed as the topic. Accordingly, nèi chǎng huǒ ‘that fire’ in (6b) is analyzed as the topic of the sentence, while xiāofángduì ‘fire brigade’ is the subject of the predicate lái de kuài ‘come quickly’.^[10] Considering that there are additional factors influencing topic recognition, such as position (initial or non-initial) and definiteness (definite or indefinite), it is justifiable to argue that, within the scope of this study, in a sentence containing two or more subject-like NPs, the NP(s) lacking a direct semantic or argument relationship with the predicate demonstrate(s) a significant level of topicality.

Liu (2012) provides evidence that in double-compm bǐ-comparatives in MSM, the attr-compm has a direct semantic or argument relationship with the p, and the compm commonly has no such relationship. In (4a), for example, [attr-compm]-p is acceptable as in 价格贵 jiàgé guì ‘the price is expensive’, whereas the compm, 这家店 zhè jiā diàn ‘this store’, lacks a direct semantic or argument relationship with guì ‘expensive’. The compm thus exhibits a considerable degree of topicality. Furthermore, if the compm appears in the initial position of the sentence, as in the case of zhè jiā diàn ‘this store’ in (4b), its topicality is further enhanced.

Notably, Liu (2012) observes that in bǐ-comparatives, the st may not necessarily exhibit a direct semantic or argument relationship with the p, as seen in examples like 其他动物 qítā dòngwù ‘other animals’ and 长 cháng ‘long’ in (2), as well as 那家店 nà jiā diàn ‘that store’ and 贵 guì ‘expensive’ in (4). This is because the st is used to establish comparisons with the compm. When the compm lacks a direct semantic or argument relationship with the P, the st performs the same. The fact that the st can function independently of a semantic or argument relationship with the p reveals its topicality, despite occurring in a non-typical position for a topic. This is because the st only has a pragmatic relation to the core predicates in such cases.

3.3 Double-compm comparatives in Sinitic varieties

The Chinese linguistic community has long held the belief that while there are significant phonological differences between Sinitic varieties, there is only one universal Chinese grammar (Chao 1968). However, recent research has demonstrated that the grammatical differences between Sinitic varieties are greater than previously thought (Chappell et al. 2007; Dryer 2003; Hashimoto 1976; Szeto and Yurayong 2021; Szeto et al. 2018).

Comparative constructions, as a part of the Chinese grammatical system, exhibit a rich diversity. Some studies have been conducted on this topic. Dryer (2003) points out that the word order of comparative constructions in MSM blends the characteristics of both SOV and SVO languages. He provides examples of the bǐ-comparatives. See (7).

MSM has SVO as its basic word order, for which Predicate-Standard NP is predicted according to word order universals (Dryer 2003), but it adopts the Standard-Predicate order typical of SOV languages.^[11]

Chappell and Peyraube (2015) note, however, that in addition to bǐ-comparatives, there is another type of comparative construction known as the “surpass” comparatives in Sinitic varieties, which, with the p-st order, perfectly matches the word order of VO languages. For example:

The marker of standard gwo in (8) has the meaning of “surpass,” thus this type of comparative construction is referred to as the “surpass” comparatives in contrast to the “compare” comparatives represented by bǐ-comparatives.

The core difference between the “compare” and “surpass” comparatives lies in their word order and the marking pattern of the m (dependent-marking vs. head-marking), and the marker of the standard is not limited to bǐ and gwo and their counterparts in various varieties. See (9).

Further insights into the usage, distribution, and diachronic origins of the markers of standard in Sinitic varieties are also found in Cao (2008) and Zheng (2012).

The presence of another comparative construction in Sinitic varieties, in which the ms are post-positioned case markers, has received less attention. This type of comparative construction, termed as “case-marking” comparatives for convenience,^[12] primarily exists in the Gansu-Qinghai linguistic area in northwest China,^[13] where long-term and intensive contact has occurred between Mandarin Chinese, Tibetan and Altaic languages. See (10).

The comparative marker xa in (10a) is a widely distributed dative-accusative marker in the Gansu-Qinghai linguistic area (Zhou 2020), which serves as a comparative marker in this context. It is likely derived from the Mandarin Chinese locative particle 下 ɕia ‘down’ (Zhou 2019a, 2019b). The m ɕiɛ in (10b) is an ablative marker meaning ‘from’. The use of case markers in the Gansu-Qinghai varieties^[14] to mark the st is a result of language contact with nearby Amdo Tibetan and Altaic languages. Under the influence of language contact, some Gansu-Qinghai varieties have also developed the rare dual comparative marking of dative-ablative (Zhou 2020).

Previous studies have explored the diversity of comparative constructions in Sinitic varieties, but a universal feature has been overlooked: the prevalence of the double-compm comparative construction. This construction is found in multiple groups within Chinese and used in different types of comparative constructions, including the “compare”, “surpass”, and “case-marking” comparatives.

The use of double-compm comparatives is explicitly mentioned in the work of Xia (2020), Sheng (2021), Wang (2022), and Zhou (2022, 2023a).

These data suggest that double-compm comparatives are not limited to any specific group or type of comparative construction (i.e., “compare,” “surpass,” and “case-marking” comparatives).

To provide further evidence for the prevalence of double-compm comparatives in Sinitic varieties, this study conducted a corpus analysis of 25 varieties from the Corpus of Chinese Dialect Grammar (CCDG) (Liu and Xia 2023).^[16] The CCDG consists of 711 test sentences from 22 grammatical categories, including comparative constructions, for each variety, and covers the ten major variety groups of Chinese as well as several unclassified varieties. The double-compm comparatives test sentence used in this study is (15), where the compm is 你 nǐ ‘you’ and 我 wǒ ‘I’ in the two clauses, respectively, and the attr-compm is 样式 yàngshì ‘style’ and 价钱 jiàqián ‘price’, respectively.^[17] The test sentences were recorded by CCDG contributors in the local variety to capture the most natural expressions of each variety.

The results are presented in Table 5.

If the test sentence (15) is documented in the CCDG as a double-compm comparative construction, it is categorized as “tested” in Table 5. Conversely, if (15) is recorded as a “non-double-compm” (i.e., “single-compee”) comparative construction, it is “not tested”. Of the 25 varieties examined in Table 5, the double-compm comparative construction was tested in 14, which accounts for 56 % of the sample. These varieties represent the ten major groups within Chinese, with seven of them (Mandarin, Yue, Wu, Gan, Xiang, Jin, and Pinghua) exhibiting evidence of double-compm comparatives.

The CCDG provides only a snapshot of the grammar of each variety, and the natural expressions recorded in the database may not exhaustively represent all grammatical structures of each variety. Therefore, the absence of double-compm comparative construction in the CCDG for a particular variety does not necessarily imply that it is absent in that variety. For instance, even though the CCDG data do not verify the use of double-compm comparatives in Qimen, the respondent has reported their occurrence in (16).

Example (16) provides evidence for the existence of double-compm comparatives in the Hui group. Further investigation reveals that the other two major groups, i.e., Hakka and Min, also use double-compm comparatives, see (17)–(18), respectively.

The current findings present compelling evidence for the extensive distribution of the double-compm comparatives in Sinitic varieties. The construction has been identified in all ten major groups within Chinese, along with several unclassified varieties, and is utilized in a variety of comparative types, namely the “compare,” “surpass,” and “case-marking” comparatives.

3.4 Double-compm comparatives in East Asian languages

The double-compm comparatives are also found in numerous East Asian languages.^[18] This includes Tibeto-Burman languages (19–24), Zhuang-Dong languages (25–28), Turkic languages (29), as well as Japanese (30) and Korean (31).

Note that (30) represents an exceptional case of double-compm comparatives where a mismatch occurs between the compm and the st. In this case, the compm is a non-possessive NP, while the st is a possessive NP. Consequently, similar to example (4c), (30) can also be classified as a compm-st mismatched comparative construction, which falls under the subtype of compee-st mismatched comparatives discussed in Section 4.

4.1 Definition

A comparee-standard (compee-st) mismatched comparative refers to a comparative construction that exhibits a “seemingly logical inconsistency” (Zhu 1982: 214) due to a mismatch between the compee and the st. This construction usually involves two NPs, one of which is a possessive NP while the other is a non-possessive NP. The key feature of this construction is the syntactic mismatch that arises when the non-possessive NP semantically refers to a possessive NP. The non-possessive NP lacks a direct semantic or argument relation with the P. An example of this construction is given in (32) from MSM:

In (32a), the possessive NP 大象的鼻子 dàxiàng de bízi ‘the elephant’s nose’ is syntactically mismatched with the non-possessive st 其他动物 qítā dongwu ‘other animals.’ Similarly, in (32b), the compee becomes a non-possessive NP 他 tā ‘he’, while the st 我的玩具 wǒ de wánjù ‘my toys’ is possessive. Although the non-possessive NPs qítā dòngwu ‘other animals’ and tā ‘he’ formally correspond to the possessors dàxiàng ‘elephant’ and wǒ ‘I’, respectively, they semantically refer to the nose of ‘other animals’ and the toys of ‘him’.

These non-possessive NPs lack a direct semantic or argument relation with p, as evidenced by the unacceptability of constructions like ^*其他动物长 qítā dòngwu cháng ‘other animals are long’^[19] and ^*他多 tā duō ‘he is more’. This feature highlights the topicality of the compee-st mismatched comparatives.

Owing to the scarcity of data, this paper refrains from delving into the intricacies of the compm-st mismatched comparatives referenced above, which are a subtype of compee-st mismatched comparatives where the compee is divided into a compm and an attr-compm.

4.2 compee-st mismatched comparatives in Sinitic varieties

Zhou (2023a) mentions the existence of compee-st mismatched comparatives in Zhoutun, a Sinitic variety which is greatly influenced by Amdo Tibetan in the Gansu-Qinghai linguistic area:

This construction is also found in several other varieties.

There has been little attention given to compee-st mismatched comparatives in grammatical descriptions.^[20] As a result, explicit descriptions of this construction in literature are scarce, with isolated instances found only sporadically. To systematically investigate this issue, the present study turns to the CCDG, which includes a test sentence (37) featuring an inferior comparative construction with a compee-st mismatched interpretation.

Note that whether or not (37) is a compee-st mismatched construction is determined by the presence of the possessive marker 的 de following 你 nǐ ‘you’: in cases where de is present, the sentence does not belong to the category of compee-st mismatched comparatives. This is due to the syntactic-semantic match between the st 你的 （头发） nǐ de (tóufa) ‘your (hair)’ and the compee 我的头发 wǒ de tóufa ‘my hair’. Conversely, when the sentence lacks the possessive marker de after nǐ, it is considered a compee-st mismatched comparative construction: the compee wǒ de tóufa ‘my hair’ is syntactically incongruent with the st nǐ ‘you’.

Table 6 shows that 12 out of 25 varieties in the CCDG have been confirmed to use compee-st mismatched comparatives, making up 48 % of the sample. This construction has been observed in seven of the ten major groups, namely Mandarin, Yue, Wu, Jin, Xiang, Hakka and Hui. While there is currently no confirmation of this construction in the remaining three major groups (Gan, Pinghua and Min), it should be noted that the lack of evidence in the CCDG does not necessarily indicate its absence. For instance, although the CCDG data does not confirm the use of compee-st mismatched comparatives in the Duchang variety, the contributor has reported its presence in (38).

The compee-st mismatched comparatives have also been observed in two other groups, namely Pinghua (39) and Min (40).

4.3 compee-st comparatives in East Asian languages

Although the concept of compee-st mismatched comparatives is not yet widely recognized, and no systematic discussion of such constructions can be found in the literature, examples of this phenomenon have nevertheless been identified in a number of East Asian languages or linguistic groups, including Tibeto-Burman (41–44), Zhuang-Dong (45–47), Korean (48), and Japanese (49).

Note that in (43) the compee is a headless possessive NP.

Example (47) is a compm-st mismatched comparative construction, a subtype of compee-st mismatched comparatives; it is also a subtype of double-compm comparatives, since there is compm and attr-compm.

Note that in Japanese, a typical comparative construction (compee-st-m-p) does not seem to allow for a mismatch between compee and st. Hiroyuki Suzuki (pers. comm.) explains that the sentence zou-no hana-ga hoka-no doobutsu-yori naga-i (elephant-poss nose-nom other-poss animal-m long-aux) (lit. “elephant’s nose [is] other animal(s)’ than long”) can only be interpreted as a comparison between “the length of the elephant’s nose” and “the length of other animals”. That is, it can only mean ‘the length of the elephant’s nose is longer than the length of other animals’. The mismatched reading, ‘the length of the elephant’s nose is longer than the length of other animals’ nose’, is not allowed. The compee-st mismatched construction presented in (49) (st-m, compee-p), is atypical. A reviewer highlighted that (49) cannot be readily analyzed as a case of compee-st mismatched comparative construction, as the role of hō ga is merely to emphasize the comparee. Notably, it is not possible to substitute Amerika-yori in (49) with ^*Amerika-no hō-yori (‘than America’s side’). From a formal standpoint, given that the compee is a non-possessive NP while the st is a possessive NP, I tentatively classify (49) as an instance of an atypical compee-st mismatched comparative construction. However, the definitive categorization of (49) within this framework remains an open question.

5.1 Definition

Standard-topic (st-topic) comparatives are structured as “topic-comment,” where the st serves as the topic in the sentence-initial position, providing a reference point for the sentence. The comment portion then compares the st to the compee. For example, in Shaoxing Wu (Yimin Sheng, pers. comm.), one might say:

Here, ‘Little Zhang’ serves as the st in the topic position (it can be followed by a topic marker and a pause), with the comment portion stating that ‘Little Wang is tall’, indicating that ‘Little Wang’ is taller than ‘Little Zhang’.

st-topic comparatives are cross-linguistically uncommon due to the prevailing placement of compee in the subject position (Dixon 2012): It is the compee that is more inclined to take on a topical role, while the st is typically positioned after the compee and displays relatively lower topicality. Heine (1997) proposed a “topic schema” within his framework (see Section 2), which follows the structure “compee and st, compee is p” (referred to as ‘X and Z, X is Y’). According to this schema, compee possesses at least an equal level of topicality as st. However, the st-topic schema introduced in this paper is “st, compee p”, where st assumes a significant topical role.

Note that expressions of the following kind, which resemble st-topic comparatives, do not actually qualify as st-topic comparatives.

In this sentence, ‘price’ serves as the topic but not the st, since it cannot be placed in the st position of a regular comparative construction (^*This book is more expensive than the price). Thus, such structures do not qualify as st-topic comparatives.

Depending on how the st is marked, st-topic comparatives in East Asian languages can be classified into two major types: “topic comparatives” and “‘look’ comparatives.” Topic comparatives typically indicate the st through the use of a pause and/or a topic marker, or by placing the st at the beginning of the sentence. Topic comparatives are primarily observed in Wu and Hui Chinese. On the other hand, ‘look’ comparatives introduce st with a m that conveys the meaning of ‘look’ (‘look’-m), and are primarily found in Northwest Chinese and neighboring languages.

5.2 Topic comparatives

In a topic comparative construction, the st is placed in the topic position and can be followed by a pause and/or a topic marker, as shown in (50). According to Li (2003) and Chappell (2015), this type of comparative construction is a result of conflating two conjoined clauses: “Comparing compee with st : it is still the case that compee is p. ’ > ‘As for st, compee is p .’ The st in this construction exhibits strong topicality.

Topic comparatives are mainly found in a small number of Wu and Hui varieties (Chappell 2015; Li 2003). The data in (52) provide several examples with subtle differences.

Based on the occurrence of ‘still’ and ‘be’, st-topic comparatives in Wu and Hui Chinese can be categorized into four subtypes: (i) st (top) ‘still’ ‘be’ compee p; (ii) st ‘be’ compee p; (iii) st ‘still’ compee p; and (iv) st compee p, exemplified in (52a–d) respectively. The synchronic distribution of these four subtypes across varieties provides valuable insights into the developmental trace of topic comparatives in Wu and Hui varieties. While lacking a marker of standard,^[21] st-topic comparatives are grammatical comparative constructions due to their fixed nature, conveying the exclusive meaning of “compee is more p than st” but not the “st is more p than compee”.

Wang et al. (2018) report that the use of topic comparatives is also attested in other Wu varieties, including Hangzhou, Wenzhou, Tiantai, Jinhua, Suichang, Xuancheng, Shanghai, Danyang, and Ningbo. Topic comparatives can also be found in some Hui varieties such as Jixi, Shexian, Tunxi, Xiuning, Qimen, and Yixian (Li 2003).

Wang and Dong (2022: 46) assert that Luobozhai Qiang has topic comparatives, as in (53).

According to their analysis, the strong topicality of the st is indicated by the mandatory use of the topic marker when it is placed at the beginning of the sentence.

5.3 ‘Look’ comparatives

The ‘look’ comparative construction utilizes a marker of standard that conveys the meaning of ‘look’ (‘look’-m). The structure of this construction is ‘Looking at st, compee is p’ > ‘As for st, compee is p’, where the st exbibits topicality.

‘Look’ comparatives are commonly found in the northwest region of China. Such structures can be observed in Amdo Tibetan (54).

The comparative meaning of (54) comes from the literal reading ‘If looking at Lhasa, Xining is big’ > ‘As for Lhasa, Xining is big’. Shao (2012) demonstrates the occurrence of ‘look’ comparatives in classic Tibetan, as exemplified by (55).

Shao (2012) argues that the prevalent use of ‘look’ comparatives in modern Amdo Tibetan is a remnant of classic Tibetan. The early context in which the ‘look’-m is formed highlights the topicality of the st, which appears at the clause-initial position (and can be marked by a topic marker), setting the scene for the following comment.

Zhoutun also exhibits ‘look’ comparatives, as in (56).

‘Look’ comparatives are also seen in the Linxia variety (Sinitic; Zhang 2016: 5) and Kyrgyz (Turkic; Zheng 2012) in northwest China.

Sandman and Simon (2016: 112) have pointed out that both Salar and Wutun, which are spoken in the Gansu-Qinghai linguistic area, have ‘look’-m comparatives similar to those found in Amdo Tibetan. However, the examples provided by the authors (57) do not show the st in the topic position. Due to a lack of data, it is currently unknown whether st can be placed in the topic position in the synchronic or diachronic data of these languages. Based solely on example (57), these constructions cannot be classified as ‘look’ comparatives termed in this paper.

Likewise, the comparatives in Korean with pota (po- ‘look’) as the m do not qualify as ‘look’ comparatives, as the st does not occupy the topic position (which is taken by the compee). See (58).

Contrary to Salar and Wutun, there is ample diachronic evidence that the pota-comparatives in Korean have consistently had a compee-st-pota word order during their grammaticalization process (Rhee 2022). Therefore, it can be determined that pota-comparatives do not qualify as ‘look’ comparatives.^[22]

Sections 3 and 4 of this paper introduced double-compm comparatives, compee-st mismatched comparatives, and st-topic comparatives. All of these types can be classified as topic-prominent comparatives due to their display of certain topic-related properties. In double-compm comparatives and compee-st mismatched comparatives, there is no direct semantic or argument relation between compee (compm) and/or st and p, but rather a pragmatic connection. In st-topic comparatives, the st serves as the topic, forming a “topic-comment” type of topic structure. These characteristics contrast with “subject-prominent comparatives”, as represented by English comparative constructions, where both compee and st have direct semantic relations with p, compee does not distinguish between compm and attr-compm, and st does not serve as a topic element.