Speech presentation in newspapers: an empirical reader response

This article contributes to the understanding of how readers experience speech presentation. Speech presentation is part of the discourse presentation model (Leech and Short, 1981), which outlines a series of strategies for how a speech event can be presented. The model outlines six categories of speech presentation and positions these categories on a scale of faithfulness. Each category of speech presentation has a prototypical faithfulness assumption which is influenced by textual features such as the reporting verb, amount of propositional content presented to the reader and other linguistic features. Speech presentation has received lots of attention by scholars; Semino et al (1999) and McIntyre et al (2004) demonstrate the model’s applicability in both written and spoken discourse. Despite the model being developed to account for the presentation of speech writing and thought (Short et al, 2002), few studies consider readers’ experiences of speech presentation. This study uses a reader response questionnaire to show how respondents experience speech presentation. It will also demonstrate the usefulness of qualitative and quantitative analysis; use statistical methods to identify significant differences in respondents’ experiences of the texts; and carry out stylistic analyses of the texts in order to account for the results observed. The results observed suggest that readers have high levels of confidence in speech that is presented using certain linguistic features. Therefore, the article examines these linguistic features more broadly and discusses the implication of their presence in speech presentation and how this can affect the experience of a reader. I also outline the practical implications of my research in areas such as law and journalism, and how the findings I report on can enable text producers, or those responsible for presenting others’ speech, to mitigate any confusion or unintended misrepresentation that could be perceived by the recipient.


Introduction
The present study reports on an investigation of respondents' experiences of speech presentation in newspapers. Speech presentation is part of the discourse presentation model (Leech and Short, 1981) which provides a framework for the analysis of presented discourse in a variety of communication forms, including spoken and written discourse (Walker and Karpenko-Seccombe, 2017, p.81). The speech presentation model was first used to consider the effects of presented speech in literary texts; however, developments of the model have resulted in scholars considering speech presentation in a range of non-literary texts (see McIntyre et al, 2004;Semino and Short, 2004). The speech presentation model states that there are different forms of presentation, which are characterised by linguistic features such as pronouns and tense, the use of speech marks and the amount of propositional content. The categories are positioned on a spectrum which reflects the perceived faithfulness of the presented speech to the original speech. The research I report on in this paper provides new insights into the understanding of respondent's experiences of speech presentation. Until recently, empirical testing of claims made by literary models was uncommon (see Whiteley and Canning, 2018) and no study has thus far explored how readers 2 experience speech presentation in non-literary texts such as newspapers.
The research aims to (1) understand the degree to which respondents feel the texts shown to them reflect what was originally said and (2) understand the effects which certain linguistic features have on how respondents experience speech presentation.
By using examples of speech presentation from newspapers, the findings I present are able to demonstrate a respondent's susceptibility to misrepresentation of speech, and the effects which particular stylistic features have on them. So, it is possible to observe how particular choices made by a text producer may result in a respondent reporting low levels of confidence that a given speech has been presented accurately.
In order to fulfil the aims of the research, I use a mixture of qualitative and quantitative methods. Firstly, I construct a reader response questionnaire which elicits information from respondents about the speech presentation examples shown to them. I then use the results from the reader response to carry out my own stylistic analyses of texts in order to account for the respondents' experiences. Table 1 Speech presentation categories Leech and Short (1981) introduced one of the most comprehensive models of discourse presentation. Their model presents three parallel clines which account for speech, writing and thought presentation in a text. Each parallel cline measures the faithfulness of a particular text according to how much speech is presented. The cline ranges from least faithful to most faithful, and where a stretch of presented speech is placed on the cline is determined by how the text has been constructed and the subsequent category of speech presentation it belongs to. Table 1 outlines the five categories of speech presentation in order of faithfulness to the original discourse: Narrator's Report of Speech (NRS), Narrator's Report of Speech Act (NRSA), Indirect Speech (IS), Free Indirect Speech (FIS), and Direct Speech (DS). In Table 1, we can observe examples of each speech presentation category.

Background
Direct Speech (DS) is the most faithful form of speech presentation. DS is characterised by a reporting clause and a stretch of reported speech (Leech and Short, 1981, p.256). A reporting clause (e.g. John whispered) does not present the reader with any propositional content. Instead, reporting clauses introduce a reader to a stretch of presented speech. The reporting clause also informs the reader about how the speech event has been performed. This is achieved through the presence of a reporting verb (e.g. shouted, whispered). While DS purports to be faithful to the original text, a narrator can still embed their attitude over a proposition by using reporting verbs which have particular connotations (Thompson, 1996). Free Direct Speech (FDS) is recognised by later discourse presentation models (e.g. Semino and Short, 2004; Short 2012) as a stylistic variant of DS and is distinguished from DS by the absence of any reporting signals (e.g. 'I am ready for change'). FDS does not contain a reporting clause or a reporting verb. Indirect Speech (IS) contains a reporting clause which is proceeded by a subordinate clause (e.g. Richard said that he would become a candidate). The main difference between IS and DS is a change in tense (Jeffries, 2010). The change from present tense in DS to past tense in IS distances the reader from the original speech event. IS contains past tense verbs and third person pronouns. Free Indirect Speech (FIS), meanwhile, is a merge of DS and IS: FIS does not contain a reporting clause but retains the tense and pronoun selection which would exist in the equivalent form of IS (Leech and Short, 1981, p.261). FIS is criticised by Semino and Short (2004) among others, due to how it subsumes the presentation of speech and thought. The merge of two forms of discourse presentation makes FIS an ambiguous speech presentation category to interpret and identify (see Sotirova, 2006).
The final two forms of speech presentation on the model are Narrator's Report of Speech Act (NRSA) and Narrator's Report of Speech (NRS). NRSA presents the illocutionary force of the original utterance 1 in addition to an indication of the topic or context in which the speech event has occurred. The example of NRSA in Table 1 (The man shouted to the crowds) contains only a subject (The man) the speech act carried out (shouted) and a prepositional phrase (to the crowds) informing the reader who the subject was speaking to. No propositional content is presented; the anterior discourse therefore cannot be reconstructed (see Walker and Karpenko-Seccombe, 2017, p.82). NRS is distinguished from NRSA by the verbalisation process used by the text producer. NRS informs a reader only that a speech event has taken place. Despite the subtle difference in the verbalisation processes used, Leech and Short (1981) argue that NRSA is slightly more faithful to the original discourse than NRS, since the verb used in NRSA contains more information.

Methodological approach
I made a series of methodological choices to ensure objectivity throughout the design and analysis of the research. To address each in turn:

Data collection
The reader-response questionnaire involved showing 50 respondents different examples of speech presentation from newspapers. Respondents were recruited by using personal networks and no background information was collected concerning their demographics 2 . Respondents were shown four examples of each presentation category that I obtained from a corpus of newspaper texts which had been annotated for speech presentation 3 . Using a corpus as a source of data enables the data collection phase to remain as replicable and objective as possible.

Questionnaire design
The study's primary aim is to test the degree to which respondents felt speech was, in fact, presented accurately in each example. Readers were required to rate a total of 20 newspaper extracts during the questionnaire. Jeffries et al (2019) piloted a survey design which tests readers' tolerance of unfaithful reporting in non-literary texts and I adapted their survey design to include a five-point Likert scale in combination with five options which reflect the degree to which respondents felt speech was presented (see Figure  1). The five-point Likert scale is based on a continuum ranging from 'No speech is presented' (Likert scale 1) to 'Speech is presented and I know what was actually said' (Likert scale 5). quantitative analyses during reader response studies. So, including a comments section allowed respondents to comment on their experience of a text as well as other factors which influenced their decision. The comments made by respondents provided a useful basis for understanding the effects that different linguistic features within speech presentation had on respondents.

Analytical methods
I used a range of analytical methods in order to gain a picture of the degree to which respondents felt speech was presented. I analysed the results of the questionnaire by considering the raw frequencies of each response made by respondents. Where there was a difference in how respondents rated one example of a speech presentation category in comparison to another example of the same category, I used ANOVA-One Factor tests in order to verify the statistical significance of the differences I observed in the ordinal datasets (see Rasinger, 2013, p.237). An ANOVA-One Factor test can be used to test two means of a dataset and, where a difference is observed, the test will indicate whether it is significant. This is indicated by the p>value which is reported upon completion of the test (See Rasinger, 2013, p.238).

Predictions
Based on my pilot study, I felt that my full study would show the degree to which respondents felt speech had been presented in different speech presentation categories. I expected DS and FDS to be considered by respondents as presenting a considerable amount of speech in comparison to examples of NRSA and NRS. This is because DS and FDS have higher prototypical faithfulness assumption claims and thus present the words of the original speaker verbatim (Leech and Short, 1981). IS meanwhile captures enough propositional content in order for the reader to reconstruct what was originally said and have a good idea about the intended meaning (Walker and Karpenko-Seccombe, 2017, p.82).
FIS is a hybrid between DS and IS. However, Semino and Short (2004) argue that FIS is an ambiguous form of speech presentation. Keen (2006, p.220) suggests that FIS is most likely to produce empathy in a reader due to how it merges the presentation of speech with thought. Examples of NRSA and NRS shown to respondents did not contain the same amount of propositional content as other examples of speech presentation in the study; they may be indicative of a considerable amount of content (e.g. he talked for a long time), yet little propositional content is presented to the respondents. The only difference between NRSA and NRS lies in the semantic properties of the reporting verbs (e.g. shouted and spoke) which the texts use. I predicted that respondents will evaluate NRSA and NRS as only presenting enough speech to be able to guess what was said.

Results
We will now consider the results of the readerresponse questionnaire by first observing the overall results and then the results for each category of speech presentation. Figure 2 indicates that, overall, respondents feel that the degree to which speech is presented in categories of speech presentation which have low faithfulness assumptions (e.g. NRSA and NRS) is lower than categories of speech presentation with higher faithfulness assumptions (e.g. DS and IS). Figure 2 shows that DS has the highest mean score (4.61).

Figure 2
Mean Likert scale score for each category There is only a subtle difference between how respondents rate DS (4.61) and FDS (4.27). Given that FDS lacks a reporting clause and therefore additional contextual information, respondents still feel a considerable amount of speech is presented in FDS. Figure 2 also shows that the degree to which respondents believe speech is present in IS is lower than for FDS and DS, but higher than for NRSA and NRS. Figure 2 suggests that respondents felt NRSA presented a greater degree of speech than NRS, and thus matches my expectations. NRSA has a higher mean score (2.14) than NRS (1.71).

Figure 3
Mean Likert scale score for each category Figure 3 indicates that all examples of DS are rated similarly and in Table 2 we can observe the examples of DS shown to readers. As discussed in the previous section, respondents were confident speech was being presented in DS. Respondents most frequently assign DS to Likert scale 5 and feel that the examples enable them to know what was actually said. In Table 3a we can observe that all texts contain a reporting clause, a reporting verb and a stretch of reported speech. Although respondents feel confident that the examples of DS shown to them present a considerable amount of speech, Figure 3 indicates that respondents experience DS4 differently to the other examples of DS. Respondents' experiences of DS4 may have been influenced by the text producer's use of apposition within the reporting clause. The reporting clause 'John's mother, Pauline, said' in DS4 is constructed using the juxtaposition of two noun phrases with the same referent (Jeffries, 2010, p.53). The effect of this use of apposition may result in respondents being unsure as to whether John or Pauline has spoken, given that the reporting clause introduces the respondent to two potential speakers. Apposition is also used in the reporting clause of DS2 (Phillip, the café owner). The first noun phrase in DS2 (Phillip) indicates who has performed the speech event, while the second noun phrase (the café owner) provides additional information about the speaker. Unlike DS4, the reporting clause in DS2 does not introduce the reader to another person who may have performed the speech event. Thus, the presence of apposition in DS4 appears to affect respondents' confidence in whose speech is being presented and warrants further investigation. The outcome for how respondents experienced DS4 was not expected since respondents were being asked about whether speech was being presented, rather than who said it. Nevertheless, this result indicates how the characterisation of the reporting clause can affect respondents' experience of speech presentation and potentially mislead a respondent into whose speech is being presented.

DS1
We are from the Daily Mirror" said the reporter. DS2 Phillip Kelly, the cafe owner, said: "There were people lying dead or dying all over the cafeteria floor and in the grounds." DS3 "We have to have political cement" he said. DS4 Jonathan's mother, Pauline, said: "I want them to get married. But they're of a different generationthat sort of thing has never been important to them." Mean Likert scale score for FDS None of the FDS shown to respondents contain a reporting verb or a reporting clause. This resulted in a number of respondents commenting on how they experience difficulties interpreting whether speech was being presented: Reader 036: Overall easy to understand, but a lack of quotation marks is confusing. I had to reread it a few times.
Reader 024: Have to infer that this is the character's speech, not the narrator's.
Despite respondents experiencing difficulty in identifying who is responsible for the speech event, the majority of respondents assign FDS to Likert scale 5. Figure 4  Reader 003: Not in quotes but appears to be a phrase therefore reasonable to assume it's accurate.
Reader 032: It's missing the quotation marks but seems to be a direct report of what she's saying.  clear that respondents do believe speech is presented in the text. Figure 5 shows that respondents most frequently assign IS to Likert scale 4 and appear to experience the texts in a similar manner. Table 5a shows the examples of IS shown to respondents and it can be noted that all texts contain the same basic stylistic features: a reporting clause, reporting verb and stretch of verbiage.

Figure 5
Mean Likert scale score for IS Reader 007: How was it "made clear"?
Reader 004: Probably speech but could have been a written statement.
Reader 039: Could have been via email or other writing. Table 5a shows that IS1 and IS4 contain the verb 'say', while IS3 contains the verb 'said'. These reporting verbs are more indicative of a speech event than the verb 'made clear'. As one respondent comments, it is possible to attribute 'made clear' to another form of discourse, e.g. writing.
Respondents' comments suggest that the reporting verb chosen by the text producer of IS2 affects how they interpret the text. The reporting verbs used in other examples of IS (i.e. say and said) can only be used to describe an action which involves language (Halliday and Matthiessen, 2004). Respondents may feel more comfortable attributing 'say' and 'said' with the act of speaking in comparison to 'made clear'.

Free Indirect speech
Overall, respondents report low confidence of speech being presented in FIS and this is reflected in an overall mean Likert scale score of 1.92 as Figure 6 shows.

Figure 6
Mean Likert scale score for FIS FIS2 is rated by most respondents (33) as not presenting any speech at all. Table 6a shows that the only indication respondents are given about speech being presented in FIS2 is the presence of the verbalization processes 'regret' and 'respected'. Ken Coates agreed. They were nice young lads and had had more sunshine than they were used to. FIS4 He commenced briskly and uncompromisingly. He would not release me from my obligation to report directly to him.
'Regret' and 'respected' likely account for why respondents believe that it is someone's feelings being presented, rather than someone's speech: Reader 030: It suggests his feelings about the situation rather than his speech.
In addition to being verbalization processes, 'regretted' and 'respected' can also be mental cognition verbs, used to represent someone's thoughts (see Short et al, 2003). The high frequency of respondents who do not believe FIS2 presents any speech may, in part, be a result of the verbalization processes used by the text producer. Sotirova (2006) notes that verbs which can be used to describe a person's thoughts can make it difficult for respondents to work out whether a text presents the thoughts of the narrator or the speaker. From the raw frequencies, it appears that the majority of respondents believe FIS1 does not present any speech either: a total of 31 respondents assign it to Likert scale 1. Respondents comment on how the structure of FIS1 makes it difficult for them to judge whether speech is being presented:

Reader 27: No mention of if he actually said what he feels.
Reader 031: Only says pessimistic.
Reader 011: This could report thoughts rather than speech.
FIS1 is constructed using a predicative complement that connects the subject of a sentence to further information. They do not include any information about the action which has taken place. FIS1 contains 'he' in the subject position, which is followed by the predicate (and head verb) 'was'. Subsequently, the predicative complement 'pessimistic about prospects' only provides the respondent with further information relating to the subject 'he', as opposed to informing the respondents about the action which has taken place, i.e. a speech event. The only information which FIS1 gives the respondent is that the subject was 'pessimistic'. This adjective does not explicitly inform a respondent that speech has taken place. It is possible that the original speech event was something along the lines of 'I am pessimistic about Likert scale 9 these prospects'. However, respondents may interpret 'pessimistic' as being used as a description of someone's feelings. The combination of the predicative complement and the use of the adjective 'pessimistic' results in respondents not being explicitly informed that an action involving language has taken place. Figure 6 indicates that respondents have high levels of confidence that speech is being presented in FIS3. FIS3 has the highest mean score (2.54), with a total of 34 respondents confident that speech is presented. A number of these comment on the presence of the verb 'agreed': Reader 041: Does Ken think that or say it (agreed)?

Reader 010: Agreed with what?
'Agreed' is used in the contextualising sentence of FIS3 (Ken Coates agreed). Its presence implicitly informs respondents that a process involving language is taking place. Although 'agreed' can be indicative of other actions (e.g. nodding) and does not explicitly indicate speech, some respondents interpret it as a verbalization process and this may account for the high frequency of respondents who are confident that FIS3 presents speech.

Narrator's Report of Speech Act
According to Leech and Short (1981), NRSA is only able to offer respondents a summary of a speech event. A number of respondents who are confident NRSA presents speech comment on how the texts only provide them with a summary of what has been said: Reader 008: Know the overall focus of the conversation.
Reader 012: It's an overview of the main point of speech.
Walker and Karpenko-Seccombe (2017, p.82) argue that the absence of propositional content in NRSA is likely to result in respondents being unable to reconstruct the anterior discourse of NRSA. This is likely to be why respondents interpret NRSA as a summary. Despite this, Figure 7 indicates a number of differences in how respondents rate the texts:

Figure 7
FIS shown to readers This was how one witness, Phillip Milburn, described on ABC radio the Tasmanian killer's coldblooded method. NRSA4 Many are said to be in critical condition.
NRSA2 has the highest mean score (2.64), and is most frequently assigned to Likert scale 3. This represents a significant difference in how respondents rate NRSA2 in comparison to other examples of NRSA (p>0.001). Table 7a shows that NRSA2 is the longest example of NRSA.
While NRSA2 does not contain any propositional content, it indicates the context in which the speech event occurred through an extended summary (Short, 2012). This summary occurs prior to the verbalization process and within the prepositional phrase (boldened). Respondents comment on how the information provided in NRSA2 enables them to guess what has been said. In addition to the verbalization process 'demanded', NRSA2 also contains the verb 'negotiated'. Both 'demanded' and 'negotiated' belong to the semantic domain POLICE NEGOTIATION 5 . The additional information may enable respondents to confidently guess what was said. Other contextual information (e.g. He had been shooting at a cordon of 200 police with heavy calibre military-type rifles) clearly defines the context in which the speech event took place. These features are likely to activate a respondent's schema for police negotiations (Cook, 1994), and respondents may guess the nature of the propositions made in the speech event. Figure 7 indicates that NRSA1 is viewed as presenting the lowest degree of speech. 39 respondents do not believe NRSA1 presents any speech. NRSA1 is the shortest example: it does not contain a prepositional phrase or provide further contextual information about the subject ('Mr Clarke'). This may explain why most respondents do not believe the text presents speech. However, the verbalization process 'admitted' may also play a role: Reader 021: The word admitted implies speech, but the context is unknown.
Reader 001: Says that he admitted but that doesn't have to be something he said.
An admission can be made through other modes of communication. Thus 'admitted' does not explicitly inform a respondent that a process which involves spoken language occurred. These features contribute to why most respondents believe NRSA1 does not present speech.

Narrator's Report of Speech
The final category of speech presentation is NRS. Figure 8 shows that respondents rate examples of NRS in a consistent manner and in Table 7 we can observe the texts shown to readers:

Figure 8
Mean Likert scale score for NRS Some respondents comment on how they believe NRS indicates that only speech has taken place:

Reader 044: just talks
Reader 032: Don't actually know what she talked about.
As with NRSA, no verbiage or propositional content is supplied which is potentially why respondents believe NRS only informs them that a speech event has taken place. In Table 8a we can observe the examples of NRS shown to respondents in the survey: Reader 012: Is a statement not speech.
It is likely that the absence of any verbiage results in respondents thinking NRS is a statement. However, it is also possible that the reporting verb 'spoke' is not interpreted by respondents as a presentation of speech: it does not present an illocutionary force, and respondents may interpret it as being a noun. So respondents may believe that 'spoke' functions to name the action which has been carried out, rather than to present speech.

Discussion
We have observed how respondents experience different forms of speech presentation taken from newspapers.  (Simpson, 1993). We have seen throughout the paper that respondents frequently comment on their uncertainty concerning whether a text is presenting speech due to the verbalization process used. Verbalization processes describe any action which involves language (Jeffries, 2010, p. 42) although some verbalization processes such as 'claimed' and 'assured' can also represent material action processes. Verbs such as 'claimed' and 'assured' can be used in contexts other than speech presentation. It was noted that respondents experienced difficulties in deciding whether NRSA1 (Mr Clarke admitted yesterday) presented speech. The reporting verb 'admitted' cannot be exclusively attributed to the presentation of speech and thus may have resulted in the significant difference we have observed. Further investigation could be carried out in order to understand any given respondent's experience of speech presentation which contain reporting verbs that have multiple transitivity patterns.
Another factor which has affected respondents' experience of speech presentation is the role played by reporting clauses in drawing respondents' attention to a stretch of presented speech. Reporting clauses introduce verbiage, attribute a speech event to a speaker and present information about the speech act performed (Vandelanotte, 2009). The examples of FDS and FIS in the survey do not contain a reporting clause, and respondents comment on how they are unable to identify who has spoken. The omission of the reporting clause means respondents are not given sufficient contextual information about the speech event, despite being confident that speech is being presented. My findings also indicate that reporting clauses which contain multiple referents -such as 'John's mother, Pauline, said' (DS1) -result in respondents being unsure whose speech is presented. Interestingly, respondents were not asked to attribute the speech event to a person, but nevertheless reported on this experience.

Conclusion
This study has reported on an investigation of respondents' experiences of speech presentation. We have observed that respondents reported the greatest level of confidence that speech was being presented in examples of DS and FDS over other forms. My stylistic analyses of the texts indicate that features of DS such as speech marks and the use of the reporting verb 'said' along with a reporting clause account for respondents' experience of DS.
Meanwhile the presence of speech marks in FDS appears to alert respondents to the fact that speech is being presented and enables them to have a high level of confidence that speech is being presented. In contrast, the same individuals experienced difficulties identifying speech presentation in examples of IS. They also commented on how they felt distanced from the original speech event.
My findings also strengthen Semino and Short (2004) and Sotirova's (2006) contention that FIS is an ambiguous form of presentation. Respondents did not have high levels of confidence that speech was being presented in examples of FIS. They were also unsure whether FIS presented speech or thought. A characteristic of FIS is that the verbalization process used can give rise to multiple transitivity patterns; the processes in the study can also be attributed to the presentation of thought. FIS also does not contain any explicit reporting signals such as reporting clauses.
Further empirical research should be carried out on a number of observations made. The semantic properties of certain reporting verbs (e.g. negotiated) appeared to influence respondents' confidence about the propositions made within a speech event. NRSA2 contains the processes 'demanded' and 'negotiated' which belong to the semantic domain POLICE NEGOTIATION and may activate a respondent's schema 5 . Consideration of schema theory (Cook, 1994) and its role in enabling a respondent to reconstruct a speech event would expand our understanding of how respondents experience reporting verbs in speech presentation. The omission of a reporting clause is a key characteristic of FDS, and although respondents believe speech was presented, they were aware that no contextual information was provided.
The research I have reported on in this paper has numerous practical implications for people who are responsible for the presentation of others' speech, such as journalists presenting others' speech in an article or lawyers who may be presenting the account of a spoken interaction. Understanding the effects which certain speech presentation categories have on how readers interpret a stretch of presented speech can influence readers' perception of the authenticity of a text. It was noted that a range of respondents expressed concerns with the degree to which they believed speech was being presented in certain texts and my analyses have highlighted the textual features which resulted in this happening, such as the absence of a reporting clause or reporting verbs which can also be attributed to other modes of communication.
Although respondents were asked to identify whether speech was being presented in the texts, it was interesting to note that some respondents commented on difficulties faced when identifying who was responsible for particular speech events. This was most common in one example of DS and all examples of FDS. Apposition was used in the reporting clause of an example of DS shown to respondents who reported that they were unable to identify whether John or Pauline had spoken.
Meanwhile, all examples of FDS resulted in a number of respondents commenting on difficulties faced when identifying who had spoken. So, by paying attention to readers' experiences of speech presentation in newspapers, and the textual choices which result in these experiences (e.g. construction of the reporting verb, structure of the reporting verb), those with an interest in presenting others' speech can mitigate any confusion or misrepresentation that could be perceived by the recipient of the presented speech. The research presented in this paper could also be used to inform people how to be critical readers and understand the effects which subtle stylistic choices in speech presentation can have on an audience.
Further research should be conducted in order to replicate these findings across a larger sample size in order to verify the patterns I have observed. Consideration of how respondents experience speech presentation in other forms of discourse, such as spoken interaction, would also provide further insights into speech presentation.