Echo answers

2.1 Introduction

When Sadock and Zwicky (1985) talk about short answers to yes/no questions, their classification concerns answering systems, not languages. No definition of “echo language” or “echo answer” is given. It seems that they do not confine echo answers to languages with an echo system as they consider the accompanying answer it is in (4) as an echo answer in English.

In the literature, if an answer includes any repeated element from the question, it is referred to as a “repetitional answer”^[5] or “repeat” (e.g., Enfield et al. 2019; Bolden et al. 2023b; Heritage and Raymond 2012; Sorjonen 2001b). Sometimes echo answers are differentiated from other types of repetitional answers by using terms such as “verb-echo answers” (Holmberg 2016), “echo responsives” (Jones 1999), “verb repeat” (Hakulinen and Sorjonen 2009), and “verbal answers” (Simpson 2014). As will be shown in Section 2.4, an echo system is not confined to using verbs as short answers. Defining echo answers as solely related to verbs may therefore not be entirely accurate. Of the 130 languages surveyed by Holmberg (2016) on answering systems, 62 languages use an echo system. To acknowledge this typological difference and to be in line with the term “echo” given by Sadock and Zwicky (1985), I differentiate echo answers from repetitional answers. Echo answers are a typologically distinct subtype of repetitional answers. Repetitional answers are defined in a broad sense: the term refers to any kind of repeated response; whereas echo answers are defined in a narrow sense, referring to repetitional answers to neutral polar questions that are present in languages such as Irish as in (1) above, yet not present in languages such as English as in (2).^[6] Here neutral polar questions refer to yes/no questions, in which the whole proposition is questioned and there is no narrowly focused element other than the verb, if present. The differences in illocution between repetitional answers and echo answers will be explored in detail in Section 2.2. The polarity-based nature of the echo system will be discussed in Section 2.3. Section 2.4 will give an overview of possible echo answering strategies attested in individual languages.

2.2 Repetitional answers, echo answers and illocution

In this section, I further discuss the differences between echo answers and repetitional answers, based on the illocutions in which they can occur.

As illustrated in (3) above, repetitional answers can occur in contexts where certain pragmatic effects are involved, e.g., focus, denial, agreement, etc. Additionally, beyond polar questions, repetitional answers can occur in response to other types of illocutions: statements (5a); declarative questions (5b); alternative questions (5c); tag questions (5d); and even imperatives (5e). These kinds of repetitional answers are either used to achieve certain pragmatic effects (contradiction (5a), confirmation (5b), (5d), reassurance (5e)), or are a conventionalized way to respond, given the nature of the question type, as illustrated in (5c). Most likely, languages, whether echo or non-echo, do not differ significantly in this type of repetitional strategy.

In contrast, echo answers are limited to responses to polar questions, for the reason that echoing an element from the polar question is not a commonly available strategy in many languages.^[7] I follow Sorjonen (2001a) in distinguishing between polar questions that seek information and those that seek confirmation. Information-seeking questions are a type of polar question where a questioner has no prior knowledge about the truth or falsity of the propositional content. The respondent is invited to affirm or disaffirm as in (6a). Confirmation-seeking questions are asked by a questioner who already holds certain knowledge or expectations about the answer and seeks confirmation, as in (6b).

Echo answers are restricted to these two types of polar questions, as illustrated by the following examples in Mandarin Chinese (henceforth Chinese), an echo language. The questions in (7a) and (7b) look identical, yet they inquire about different information in two different contexts. In (7a), when John asks Mary whether she has been to the restaurant before, he has no knowledge about the truth status of the state-of-affairs that he inquires about. Hence, his question is seeking a new piece of information. However, in (7b), John has an assumption about whether Mary has been to the restaurant before based on the contextual evidence that Mary orders food without looking at the menu; hence John is seeking confirmation for this assumption. In response to both questions, the echo answer is used to affirm (7a) and to confirm (7b).

Chinese has mixed answering systems: the agree/disagree system and the echo system.^[8] The agree/disagree system uses the particles shi ‘yes’ and bu (shi) ‘no’. In response to questions, these particle answers, along with interjection-type answers such as en, wei, xing, hao, slightly outnumber repetitional answers in naturally occurring conversations (Wang 2021). However, there is a division of labor between them. As seen from (7a), particle answers are not acceptable in response to information-seeking polar questions, while they are acceptable in response to confirmation-seeking polar questions. In other words, in Chinese, information-seeking polar questions prefer echo answers. This preference is also reported to be dominant in other echo languages, such as Finnish (Sorjonen 2001b), Polish (Weidner 2023), and Brazilian Portuguese (Harjunpää and Ostermann 2023; Rosemeyer and Schwenter 2019). Enfield et al. (2019: 292) investigate polar answers in 14 languages and find that repetition-type answers show their “highest relative frequency in response to requests for new information”. Therefore, it seems that echo answers prototypically occur in response to information-seeking questions.

Nevertheless, as seen in (7b), echo answers can be available as a response to confirmation-seeking questions. In this kind of echo answers, there is a gradience of markedness in pragmatic meanings across echo languages. In some languages, such as Chinese, where both particle and echo answers are available, echo answers are more epistemically marked while in some other languages, such as Tzeltal, where echo answers are the default answering strategy (Brown 2010), they are not as epistemically marked. Response particles are anaphors (Krifka 2013) in the sense that they carry no inherent propositional content, which instead has to be recovered from the antecedent question, and they also accept the stance or attitude framed in that propositional content (Enfield and Sidnell 2015); in contrast, repetitional answers assert the propositional content in the way as if it were independent of the antecedent question, thereby claiming more epistemic authority over the information at issue (Heritage and Raymond 2012). For instance, in (7b), the particle answer shi de simply confirms or agrees with John’s assumption whereas the echo answer lai guo is more assertive, taking greater ownership over what is being said in the response, as lai guo can function as an independent sentence with its own propositional content, which is asserted by the speaker. Therefore, compared with particles answers, echo answers are more pragmatically marked in languages where both types of answers are common. This distinction is illustrated more clearly in the following Russian example (8). In response to Vara’s confirmation-seeking question ‘A thick one?’, Grandma confirms with an echo answer while simultaneously, Grandpa confirms with a particle answer. Grandma is the one who gives the instruction to draw a ‘big stick’, so she confirms that ‘thick’ is in fact what she meant by ‘big’, thus asserting her epistemic authority; on the other hand, Grandpa simply affirms the accuracy of Vara’s understanding (Bolden 2023).

Enfield and Sidnell (2015), and Enfield et al. (2019) hypothesize that there is a “strong typological preference in the world’s languages for using interjection-type answers to polar questions” (2019: 301) and the repetition-type answers are “pragmatically marked and reserved for specific communicative purposes” (2019: 300). This is true in non-echo languages such as English, in which the use of repetition to confirm a polar question is not common in natural conversation (Enfield and Sidnell 2015: 138; Schegloff 1996). It is partly true for echo languages that have both answering options available for use in response to a confirmation-seeking question, such as Chinese, Russian (Bolden 2023), Finnish (Sorjonen 2001b), Polish (Weidner 2023), Estonian (Keevallik 2010), Brazilian Portuguese (Harjunpää and Ostermann 2023; Rosemeyer and Schwenter 2019), and Greek (Alvanoudi 2022). In these languages, echo answers occur much more frequently than in English, and thus they are generally less pragmatically marked than in languages like English. However, it is not true for Yurakaré (an isolate spoken in Bolivia; Gipper 2024) and for Mesoamerican languages such as Tzeltal, in which echo answers are the most frequent and default form for responses to polar questions, as shown by analyses of data from naturally occurring conversations (cf. Brown 2010: 2640; Brown et al. 2021) and repetitional answers do not impose epistemic authority or primacy as in languages such as English, Chinese, and Russian (Brown et al. 2021). The level of conventionalization of echo answers is even higher in languages such as Latin and Scottish Gaelic, in which there are no equivalents of yes/no words. Consequently, the frequency of echo answers in Latin and Scottish Gaelic is presumably higher than that in Tzeltal. In the languages that have echo answers as a default strategy, echo answers are least pragmatically marked. Hence, among the world’s languages, there is a spectrum of pragmatic markedness, with languages such as English on one end, languages such as Scottish Gaelic on the other, and others somewhere in between, as shown in Figure 1. There is a decrease of pragmatic markedness of repetitional or echo answers from non-echo to echo languages. Within each type of languages there is variation.

Figure 1:

Gradience of pragmatic markedness of repetitional/echo answers in the world’s languages (this figure is just to illustrate roughly the spectrum of different degrees of pragmatic markedness. Only a few languages are included based on the findings in the literature, and there is no attempt to locate accurately every mentioned language on the axis. No doubt more quantitative analyses are needed to give a more fine-grained picture).

In this section, I have discussed the differences between echo answers and repetitional answers regarding illocution. Repetitional answers are not restricted to echo languages, and they have restriction on the illocutions that they can co-occur with. Echo answers have a strong tendency to occur as a response to an information-seeking polar question. When languages have both particle answers and echo answers available for use, they tend to carry different functions in response to confirmation-seeking questions. The more conventionalized an echo strategy is, the less pragmatically marked it is.

2.3 The polarity-based nature of echo answers

Apart from the echo system, the other two systems classified by Sadock and Zwicky (1985) are the yes/no system and the agree/disagree system. The former is also known as the ‘polarity-based’ system, as the particle yes indicates a positive polarity, while no indicates a negative polarity. Unlike the echo system, both the yes/no and agree/disagree systems use response particles to answer. The yes/no system uses yes/no particles, and the agree/disagree system uses agree/disagree particles. The difference between these response particles is particularly evident in how they answer a negative question with a negative bias (expecting a negative answer). If the polarity of particles like yes/no in English aligns with the polarity of the accompanying sentence answer, whether implicitly or explicitly expressed, or that of the antecedent question, then the language follows a polarity-based system. For example, in (9), the presence of ‘still’ explicitly indicates that the questioner expects a negative answer, based on evidence, e.g., that the answerer is still sitting in the kitchen. Confirmation is provided by using the negative particle no, which means ‘I haven’t’, thus aligning with the negative polarity of the question. Disconfirmation (contradiction), on the other hand, is conveyed by using the affirmative particle yes, meaning ‘I have’, thus dis-aligning with the negative polarity of the question. Hence, a polarity-based answering system functions by aligning the polarity of the answer with that of the preceding question.

However, in terms of polarity alignment, the opposite is true for an agree/disagree system, where the ‘no’ particle is used for disconfirmation while the ‘yes’ particle is used for confirmation. As in Japanese, in (10), the particle hai is used where English would use no, and the particle iie is used where English would use yes. This means that the polarity is misaligned between the particle and its accompanying answer, if expressed, and between the particle and the question. Therefore, the agree/disagree system is not based on polarity alignment. It is termed the “agree/disagree” system because the response particles are agree and disagree particles. As in Japanese, hai is always used to agree and iie is always to disagree, regardless of the polarity of the antecedent. For example, if the question in (10) is changed to an affirmative question, the agree particle hai, which indicates ‘I didn’t go’ in (10-A1), can still be employed to indicate a positive answer ‘I went’.

Semantically, a yes/no (polar) question offers a selection from two alternatives which consist in the polar values of True/Not True (Dik 1997: 263). For instance, if the questioner’s expectation for a negative answer is disregarded, the yes/no question of (10) is semantically like the alternative question (11a). The answerer is expected to choose between two alternatives with opposing polar values. An affirmative answer is similar to (11b), while a negative answer is similar to (11c).

In other words, the English no is semantically similar to (11c), while yes is semantically similar to (11b). However, in an agree/disagree system, the alternative question runs like (12a). The Japanese hai is semantically similar to (12b), while iie is semantically similar to (12c). Regardless of the polarity of the antecedent, hai always indicates ‘it is true’ while iie always indicates ‘it is not true’.

The discussion above shows that the distinction between the yes/no system and the agree/disagree system lies in the different levels of polar values they address.

For a question like (10), a language with an echo system would confirm by repeating both the predicate and its negation, while it would disconfirm by repeating only the predicate, as illustrated by (13) in Chinese. Mei qu in (13-A1), consisting of the verb qu and the negator mei, is similar to the English response ‘No, (I didn’t.)’, while qu in (13-A2) is similar to ‘Yes, (I did).’

Hence, echo answers also function by aligning/misaligning with the polarity of the antecedent question in the same way as yes/no particles in the yes/no system. In other words, echo answers, like yes/no particles, address the same level of polar values for the antecedent question. Therefore, the echo answering system is also polarity-based. The difference between the yes/no system and the echo system lies in how their answers are formed: one through the means of particles while the other through repetition.

However, it is not correct to claim that only the yes/no system and the echo system are polarity-based. The agree/disagree system also addresses polar values by directly concerning the truth status of the proposition expressed in the question, thus making it polarity-based as well. This is unsurprising, as all answering systems are framed by the nature of polar questions, which inquire about the polar values of the propositional content. Therefore, the term “polarity-based” is not useful for referring to just one type of the three answering systems.

2.4 Types of echo answers

According to Sadock and Zwicky, echo answers are made up of repetitions of the verb from the question, “with or without additional material” (1985: 191). It is unclear what they mean by “verb”. A verb may refer to a lexical verb, an auxiliary verb, or a copula. When a lexical verb functions as the predicate in a sentence, it may be morphologically marked for grammatical information such as tense or aspect, or it may appear bare, without any inflectional marking, as is the case in isolating languages. In this paper, the abbreviation “V” refers to a lexical verb, whether bare or inflected. “Auxiliary” refers to both auxiliary verbs and copulas, as copulas are considered a subtype of auxiliaries. This classification is because copulas only provide support for non-verbal predicates without contributing independent semantic content to the sentence, as evidenced by their non-obligatoriness (Dik 1987/2011, 1997; Hengeveld 1992). Predicates can be either verbal or non-verbal. Verbal predicates may consist of a single lexical verb, or a verbal complex that includes other elements besides the lexical verb, while non-verbal predicates are of various types, such as nominal, adjectival, adpositional, or possessive (Dik 1997). Unless otherwise specified, the term “predicate” in this paper refers to verbal predicates.

In this section, I will provide an overview of the types of possible echo answering strategies attested in individual languages, focusing on what kind of verbal material and/or additional material can serve as echo answers.^[11] The summary is presented in Figure 2, and each type listed will be followed by detailed illustrations. It is important to note that the following discussion primarily addresses regular affirmative short answers (i.e., no special pragmatic effects are needed to validate the accessibility of a certain type of answer). As mentioned in Section 2.1, the polar questions involved are neutral affirmative ones.

Figure 2:

An overview of possible echo answering strategies attested in individual languages.

A typical echo answer involves repeating the lexical verb, including its finiteness if applicable, from the antecedent to affirm, or repeating it with a negative word to disaffirm, as shown in (14). This strategy applies regardless of the transitivity status of the antecedent question.

The finite marking on the verb is usually adapted in the answer. As seen in (14), the person marking is adapted into the first person if the question involves the second person. In rare cases, it may be directly copied, as in Brazilian Portuguese in (15).

However, when the question involves a third person, the person marking is always directly copied, as in (16).

Regarding tense marking, it is also directly copied, as in (14) and (15), yet it can also be adjusted as in (17a) in European Portuguese. Aspect marking is usually copied and repeated in the answer as in (17b).

The verbal predicate in the antecedent question can be complex, consisting of a lexical verb as the main verb along with other elements that indicate tense, aspect, and modality, such as auxiliaries. Languages vary as to which part of the complex predicate can be repeated in an echo answer: the auxiliary alone (18a), the main verb without an auxiliary (18b), or both the auxiliary and the main verb (18c).

When the complex predicate involves a serial verb construction (concatenation of verbs without overt marking of subordination or coordination), the availability of echo answers is like that of the “auxiliary + verb” construction: the first verb (19a-A1), the whole verb series (19a-A2), or the second verb (19b).

When the predicate in the question involves arguments, in some languages, the subject or the object can be repeated along with the verb as in (20).

In Slovenian, the clitic object pronoun can even occur systematically alone to mean ‘yes’ as in (21).

When phasal polarity expressions such as ‘already’ occur in the answer, be it an adverb or an aspect marker, they either occur alone as in (22a) and (22b), or along with the verb, even when they are not present in the antecedent question as in (22c).

Sometimes, even a completive aspect particle can occur alone as in (23). This particle la in Tzeltal, in all other contexts a bound morpheme that cannot occur alone, functions alone as a yes answer.

It has also been found that certain additional material that reinforces the meaning of the echo answer such as ‘indeed’ and ‘of course’ can be added as in (24).

The above are concerned with the polar questions in which everything is explicitly expressed. When it comes to a fragment polar question, i.e. certain elements, especially the predicate, are not overtly expressed, yet implied in the context, the unexpressed elements can still be recovered to form an echo answer as in (25). In (25a), the copula wyt ‘are’ and the verb cymryd ‘take’ are implied in the context, yet the copula auxiliary is recovered to indicate ‘yes’, much as it would be if every element of the question were overtly expressed.

To summarize, as shown in Figure 2, a variety of echoing strategies is available in echo languages. These strategies include: the lexical verb alone, with its finiteness if applicable; the auxiliary, with or without its main verb; the main verb without the auxiliary; the pronominal subject or pronominal object with the predicate; the personal pronoun; phasal polarity expressions such as ‘already’; reinforcing words alongside the usual echo answer; or recovered elements that are implied but not explicitly expressed in the context.

4.1 Introduction

Section 2.4 provides an overview of echo strategies attested in individual languages. Based on these attested strategies, a survey was conducted across 26 languages to determine whether these strategies are available in each language and to investigate potential cross-linguistic constraints on the echo-ability of these strategies. It was found that languages share some strategies (Section 4.2) and vary in others (Section 4.3). There are two Echo-ability Hierarchies that impose constraints on which element(s) from the question can be repeated in an echo answer cross-linguistically (Section 4.3). Note that the elicited results are based on informants’ intuitions and may be subject to different judgments by different native speakers. On the other hand, intuition-based judgments are most probably concerned with the prototypical use of echo answers in a particular language.

4.2 Commonalities

In all the languages investigated, except for Latin,^[15] echo answers share four commonalities.

First, the lexical verb in a simple predicate can be repeated alone to indicate ‘yes’. No other element is found to be cross-linguistically as equally pervasive. When Sadock and Zwicky (1985) and König and Siemund (2007) define the echo answering strategy as repeating the verb, their hypothesis holds true if they are specifically referring to the lexical verb in a simple predicate, whether finite or bare. If there is morphological marking on the verb, person marking is usually adapted as in (26). No other language in the sample is found to copy the exact person marking as Brazilian Portuguese (cf. example 15). In contrast, markings such as tense are usually copied as in (26a), and most languages do not adapt the tense marking, except for European Portuguese, as in (17a) (Section 2.4), and Latvian, as in (26b).

Of course, if there is no morphological marking on the verb, bare verbs are repeated as in (27).

In many languages of the world, an adjective can function syntactically as a verb. They are called “adjectival verbs”. When they are available in a language, they can also stand alone as in (28).

The second commonality is that elements that either reinforce or mitigate the force of the affirmative answer, or elements that express epistemic stance of the answerer, can be added. Note that this addition is still within the realm of affirmation, which is different from the expanded answer that gives some extra information apart from affirmation. In (29a), the affirmative answer is enhanced by the addition of the word mesmo ‘indeed’; in (29b), the affirmative answer to the question sounds more assertive with the addition of the sentence-final particle lèè3. Regardless of the effects the added words may introduce – reinforcement, mitigation, or uncertainty – they remain part of the short affirmative answer.

The third commonality is that the verbs can be recovered even when they are not overtly expressed, as seen in (30). However, it does not mean that whenever the verb from the question is not overtly expressed, it can always be recovered and used as an echo answer. There appear to be certain language-specific restrictions on the linguistic context that permits the recoverability of the verb from the question. In all the languages investigated, in the context of (30), copying the unexpressed verb is acceptable; however, this is not the case in other contexts such as (25a), repeated here as (31a). In this context, the echo answer is acceptable in some languages such as Welsh as in (25a), but it is unacceptable in others such as Persian as in (31b), where the ‘yes’ particle has to be used for the answer to be acceptable. This is an interesting topic for future investigation.

The fourth commonality is that a subject or object pronoun cannot stand alone as a ‘yes’ answer in all 26 languages inspected. That the object clitic can function as a short answer is idiosyncratic of Slovenian (cf. (21)), in which the object clitic has undergone an unusual development that leads to the acquisition of predicative characteristics (Dvořák 2007: 209). This explains why it can stand alone as a short affirmative answer.

A summary of the commonalities is provided in Figure 3.

Figure 3:

Commonalities of echo answering strategies across the sampled languages.

4.3 Echo-ability hierarchies