Introduction
Asignificant proportion of individuals who experience a stroke in the left cerebral hemisphere—particularly in the inferior frontal gyrus, including Broca’s area and its surrounding regions—develop nonfluent aphasia. One of the most prominent manifestations of nonfluent aphasia is agrammatism [1]. Agrammatism is defined as a linguistic disorder characterized by nonfluent, effortful, and slowed speech. The speech of individuals with agrammatism is primarily composed of content words, while grammatical morphemes—both free and bound—are often omitted or substituted. Production deficits in agrammatism are relatively easy to identify; however, comprehension impairments are less apparent. In fact, comprehension difficulties were largely overlooked until the 1970s, when the studies by Caramazza and Zurif (1976) revealed that individuals with agrammatism, in addition to production impairments, also struggle with understanding complex and noncanonical sentence structures [2].
A noncanonical sentence is one that deviates from the typical word order or conventional syntactic structure of a language. Such deviations are generally employed for semantic or emphatic purposes [3]; examples include topicalization (to emphasize a particular element within the sentence), passivization, and cleft constructions. In Persian, the canonical word order follows a “subject–object–verb (SOV)” pattern (e.g. /Maryam, Ali rā did / “Maryam, saw Ali”). If this sentence is transformed into /in Maryam bud ke Ali rā did/ “It was Maryam who saw Ali”, the subject “Maryam” has been emphasized, and subject topicalization has occurred; as a result, the sentence is considered noncanonical in structure. Similarly, if the object is syntactically moved to the beginning of the sentence—contrary to its default position (e.g. /Ali rā, Maryam hol dād/ ) object topicalization has occurred. Other examples of noncanonical constructions include syntactically complex sentences such as passives, where the subject is omitted (e.g. /sib xorde šod/ “The apple was eaten”), and object-cleft sentences, which not only highlight the object but also include a relative clause (e.g. /in Maryam bud ke Ali holaš dād/ “It was Maryam who Ali pushed”).
Research has shown that certain types of noncanonical sentences—particularly passive constructions, object-topicalized sentences, and object-cleft structures—pose greater challenges for individuals with agrammatism [1, 4, 5]. In this regard, Bastiaanse and van Zonneveld (2005), in their investigation of the difficulties associated with the comprehension and production of sentences featuring noncanonical word order among Dutch-speaking individuals with agrammatism, demonstrated that canonical word order structures are relatively easier for these individuals to comprehend. In contrast, interpreting noncanonical structures, in which sentence elements are syntactically displaced and thus considered derived, proves to be more difficult [5]. Based on these findings, they proposed the derived order problem hypothesis (DOP-H). This hypothesis posits that each language has its own base word order, and any deviation from that order—through the syntactic movement of sentence constituents—increases processing complexity and disrupts the accurate use of syntactic knowledge required for sentence interpretation. The difficulty associated with derived structures as predicted by the DOP-H has been supported by various cross-linguistic studies, including those conducted in English [6], Dutch [7] Italian [8] Turkish [4] and Czech [9]. These findings reveals the significance of cross-linguistic research in this domain [10-12].
In light of these challenges, one of the central questions in this context concerns whether production and comprehension impairments in individuals with agrammatism always manifest simultaneously, or whether a distinction can be made between these two language functions. Research findings on this issue have been contradictory; some studies suggest that these impairments do not necessarily occur concurrently and that production and comprehension abilities can be independently compromised. Such differences have been reported both in cross-linguistic studies and within-language investigations. For example, some Italian speakers with agrammatism exhibited deficits in producing grammatical sentences, while their sentence comprehension skills remained intact [13, 14]. A report by Nespoulous et al. on a French-speaking patient also indicated that, despite severe impairments in producing ungrammatical sentences, the comprehension of complex sentences remained unaffected [15]. Similar findings have been documented by Italian and German researchers [6, 13, 14]. However, an investigation by Meyer et al. into the relationship between production measures—such as mean length of utterance (MLU)—and the comprehension of passive sentences in English-speaking individuals with agrammatism found that those with lower MLU scores experienced greater difficulties in sentence comprehension [16].
In Persian, relatively few studies have examined the comprehension skills of Persian-speaking individuals with agrammatism (PSWA) in comparison to those in other languages. Raghibdoost, in a study investigating the comprehension abilities of two PSWA compared to healthy controls, found that the participants demonstrated relatively adequate comprehension across various sentence types, with only mild deficits observed in processing object-cleft and object-relative structure [17]. In a separate study, Shiani et al. reported severe difficulty in comprehending noncanonical sentences involving syntactic movement—particularly object-relative clauses—among six agrammatic participants [18]. Azad also documented significant challenges among four individuals with agrammatism in understanding relative and object-cleft constructions [19-21]. Although both Shiani et al. and Azad [17, 18] found a statistically significant difference between the comprehension of complex versus simple sentences in agrammatic individuals, their findings stand in contrast to those reported by Raghibdoost.
The inconsistencies in findings across both domestic and international studies, along with cross-linguistic differences and the variability of language performance in individuals with agrammatism, reveals the need for further research in this area—particularly in the context of the Persian language. A review of the literature suggests that, in addition to individual differences among participants and the severity of their production and/or comprehension impairments, part of these discrepancies may stem from the use of diverse assessment tasks and protocols across studies, which can sometimes yield contradictory results [6, 22, 23]. Given the limited findings available in Persian, the present study aimed to investigate the performance language profile of a broader sample of PSWA in the domain of sentence comprehension and its relationship with production measures. The objectives of this study were as follows: 1) To examine sentence comprehension (canonical and noncanonical) in PSWA and compare their performance with that of non-brain-damaged (NBD) controls; 2) To compare the comprehension of canonical and noncanonical sentences within the PSWA group; 3) To analyze the MLU and mean length of sentence (MLS) in both PSWA and NBD groups; and 4) To investigate the relationship between MLU and MLS and the sentence comprehension score in PSWA.

Materials and Methods
This descriptive-analytical study was conducted in Tehran during the years 2023–2024. The study population consisted of Persian-speaking adults with nonfluent aphasia who, in 2024, were referred to rehabilitation centers affiliated with the University of Rehabilitation and Social Health Sciences. Participants were selected through non-probability sampling from among these referrals, based on predefined inclusion and exclusion criteria.
The inclusion criteria were as follows: being an adult (18 years or older), right-handed, and having a language impairment resulting from a left-hemispheric brain lesion caused by a single stroke, with at least three months having passed since the incident. Participants were either monolingual Persian speakers or bilingual/multilingual individuals with sufficient proficiency in Persian. Observable signs of agrammatism were evident in their connected speech, including reduced speech rate (a score of 8 or lower on the fluency criterion of the discourse task from the Persian diagnostic aphasia battery (P-DAB-1), telegraphic speech or syntactic simplification, utterance length ranging from one to four words, and limited or omitted use of function words. Additionally, participants were required to have adequate comprehension and verbal communication abilities to perform aphasia assessments, normal or corrected-to-normal hearing and vision, and sufficient cognitive functioning—including alertness, orientation, attention, and cooperation—as confirmed by a certified speech-language pathologist.
Exclusion criteria included the presence of apraxia of speech or severe dysarthria, severe aphasia with minimal or absent verbal output, a history of substance abuse, acute neurological or psychiatric disorders, and unwillingness to continue participation. All criteria were determined through patient interviews, medical record reviews, and expert evaluation by the rehabilitation team to ensure the accuracy of participant selection and the integrity of research data.
Ultimately, 42 participants were included in the study, divided into two groups: the first group consisted of 21 PSWA (9 women and 12 men), and the second group comprised 21 NBD adults (10 women and 11 men). The control group was matched with the aphasic group in terms of gender, age, and educational background (although perfect matching was not fully achieved) and had no neurological or psychiatric conditions that could affect language abilities. These individuals also demonstrated normal vision, hearing, and cognitive functioning. A demographic questionnaire, completed by the participants themselves, confirmed their medical history and general health status. All participants signed a written informed consent form. 

Assessment and procedure
The P-DAB-1 test, which was developed based on the structure and diagnostic criteria of the Western Aphasia Battery, was standardized in Persian by Nilipour and colleagues (2014) for screening and assessing the severity of acquired language disorders resulting from brain injury [24]. The test comprises six subtests: Content quality of speech, speech fluency, auditory comprehension, comprehension of continuous commands, repetition ability, and naming ability. Each subtest is scored out of 10 points. To calculate the aphasia quotient (AQ), the sum of all subtest scores is divided by 6 and then multiplied by 10. A score between 92 and 100 indicates no aphasia, while a score below 92 indicates aphasia of varying severity. Validity and reliability assessments of this test have demonstrated good internal consistency (α=0.71) and acceptable test-retest reliability (r=0.65, P<0.05).
To assess sentence comprehension, the Bilingual Aphasia Test (BAT), developed by Paradis et al. [25] in 1987, was employed. The Persian version of the BAT was translated by Nilipour and Tahereh Paribakht. In the current study, the grammatical comprehension subtest of the BAT was used, which follows a sentence-to-picture matching format. In this subtest, a sentence is read aloud to the participant, who must then select the image that best matches the meaning of the sentence. The grammatical subtest includes a total of 70 items (items 66 to 136), covering two types of sentence structures: Standard (canonical) sentences such as “The boy is chasing the girl,” and nonstandard (noncanonical) sentences, including subject clefts (e.g. “It is the boy who is chasing the girl”), object topicalization (e.g. “The dog, the cat bites”), subject clefts (e.g. “It is the cat that bites the dog”), and object clefts (e.g. “It is the cat that the dog bites”). Additional examples of these structures are available in the explanatory guide of the screening section of the test [25] . Scoring was binary: 1 point for correct responses and 0 for incorrect or missing responses. Participant responses were recorded and scored in real-time with precision.
To extract the MLU and MLS, participants’ discourse samples (defined as a unit of language longer than a single sentence [26] were analyzed. In this study, three types of discourse elicitation methods were employed: Descriptive speech (one task), narrative speech (six tasks), and story retelling (one task). A total of eight tasks were administered, as described below:
Descriptive speech: In this task, the serial picture story “The Bird’s Nest” was used. Following the instructions provided in the P-DAB-1, participants were told: “These pictures show a story; please tell me the whole story.” They were encouraged to describe all the details of the images. If participants produced only single-word responses, they were prompted with: “Try to tell me the story in full sentences.”
Narrative speech: In this section, three topics—“birthday celebration,” “Nowruz,” and “leisure time”—were presented to participants, and they were asked to speak about these topics in both past and future tenses (six tasks in total). First, participants were asked to recount each topic in the past tense (“Tell me a memory of a birthday celebration in the past,” “Tell me a memory of Nowruz in the past,” and “Talk about your leisure time in the past”). The same topics were then addressed in the future tense (“Talk about a birthday celebration in the future,” “Talk about a Nowruz celebration in the future,” and “Talk about your future leisure time”). If participants asked whether they should speak about themselves or someone else, they were given the freedom to choose. Similarly, if they asked whether to recount a distant or recent past event, the choice was left to them.
Story retelling: In this task, participants were shown six sequential black-and-white images while listening to a pre-recorded story. During playback, the researcher pointed to each image to help guide the participant’s attention. After the story was fully played, the audio was turned off, the images were placed in front of the participant, and the examiner pointed to each image as the participant was asked to retell the story [27].
Discourse analysis was conducted in accordance with the best practice guidelines for reporting discourse in individuals with aphasia and neurogenic communication disorders [28]. Based on these guidelines, all speech data from each participant were audio-recorded and orthographically transcribed to ensure that no important information was lost [29]. Subsequently, each participant’s utterances and sentences were segmented. An utterance is defined as “a unit of language that is meaningful within context and is separated by pauses or changes in the speaker’s intonation” [10, 30]. A sentence is defined as “a unit of language composed of one or more clauses and containing at least one verb” [31].
To calculate MLU, the total number of content and function words in each discourse sample was counted and divided by the number of utterances. For MLS, the total number of content and function words was similarly counted and divided by the number of sentences. All intelligible words (e.g. nouns, verbs, prepositions), including those containing grammatical errors (e.g. verbs with inflectional errors), were included in the count as long as the errors did not render the word unrecognizable. This process was guided by the definition of a word in Persian, where a word consists of one or more morphemes and may have independent meaning (e.g. a verb) or be functionally dependent (e.g. conjunctions). Compound nouns and verbs (e.g., dast zadan [to clap]) and prefixed verbs with auxiliary components (e.g. dāram dast mizanam [I am clapping]) were each considered as single words. Words were counted based on their components (e.g. mano, a reduced form of man rā [me], was counted as two words). In addition, words with phonological substitutions, deletions, or insertions, as well as informal expressions (e.g. ey vāy, nuch nuch), were included in the analysis [32]. 
Unintelligible speech, false starts, non-words (e.g., em, um), simple binary responses (e.g. “yes,” “no”), fillers (e.g. well), frequently repeated clichés (utterances repeated more than five times), immediate word repetitions, and personal remarks (e.g. “I don’t know what else to say”) were excluded from analysis. 
To address the study objectives, the following variables were calculated: mean overall comprehension score (based on the total number of test items), mean scores for canonical sentences (based on the number of canonical sentence items), noncanonical sentences (based on noncanonical items), subject-cleft sentences (based on subject-cleft items), and object-cleft sentences (based on object-cleft items). Each individual’s score was converted into a percentage. Group-level means were then calculated. Additionally, within the PSWA group, comparisons were made between mean percentages of canonical vs. noncanonical sentences, subject clefts vs. object clefts, subject-relative vs. object-relative clauses, and the relationship between MLU and MLS and scores on the BAT grammatical comprehension subtest.

Data analysis
For between-group comparisons, the independent t-test was used for normally distributed data, and the Mann–Whitney U test was applied for non-normal data. For within-group analyses, the Wilcoxon signed-rank test was employed for non-normal data, and the paired t-test was used for normally distributed data. To account for individual variability within the aphasia group, a linear mixed-effects modeling approach was applied, with participants treated as random variables and time post-onse (MOP) and AQ treated as fixed variables. This approach was used to control for the potential confounding effects of these variables on language outcomes. In addition, Pearson correlation was used to examine the relationship between production measures and the score on the grammatical comprehension subtest.

Research reliability
Inter-rater reliability for the extraction of production measures was assessed at two levels: transcription and data analysis. Transcription and segmentation of discourse into utterances and sentences were initially carried out by a certified speech-language pathologist (the first author). To ensure point-by-point agreement, two trained speech therapy students each reviewed half of the transcriptions by listening to the audio recordings, comparing them to the first author’s transcriptions, and independently re-segmenting the discourse into utterances and sentences. Any disagreements or ambiguities were resolved, prior to data analysis, under the supervision of an expert linguist (the third author).
For data analysis, after the initial analysis was performed by the first author, ten discourse samples from each participant group were randomly selected and independently re-analyzed by two additional trained students. Intra-class correlation coefficients (ICCs) were used to assess inter-rater reliability. The results showed ICC>0.95, indicating a high level of agreement between raters.
Intra-rater reliability for the analysis of production data was also assessed by the first author on a separate set of ten additional samples. Correlation coefficients ranging from 81% to 100% (mean: 90.5%) were obtained. Additionally, intra-rater reliability for scoring correct and incorrect responses and assigning values for variables related to the grammatical comprehension test was evaluated by the first author on ten randomly selected samples from each group. Correlation coefficients ranged from 95% to 100% (mean: 97.5%), confirming the high reliability of the measured variables.

Results
An independent t-test indicated no significant differences between the two groups in terms of age (t=0.34  P=0.85) or level of education (t=–0.96 , P=.40). Additionally, a chi-square test revealed no significant difference in gender distribution between the groups (χ²=0.096, P=0.75). These results suggest demographic homogeneity between the PSWA and NBD groups, supporting the validity of the subsequent between-group comparisons. Demographic details are presented in Table 1.



Between-group comparisons revealed that the NBD group performed flawlessly across all sentence comprehension variables, while the PSWA group performed significantly lower across all measures. Statistically significant differences were observed between the two groups for both comprehension and production indices, including overall comprehension scores, comprehension of canonical and noncanonical sentences, as well as MLU and MLS. Details of these differences are presented in Table 2.



Within-group analyses showed that comprehension of canonical sentences was significantly better than that of noncanonical sentences. Additionally, significant differences were observed among subtypes of noncanonical structures—specifically, between subject-cleft and object-cleft sentences, and between subject-topicalized and object-topicalized constructions. These results are presented in Table 3.



Mixed-effects linear modeling revealed that time post-onset had no significant effect on any of the variables studied. However, AQ significantly predicted comprehension of object-cleft sentences(β=–3.98, P<0.05), as well as MLU (β=0.068, P<.05) and MLS (β=0.58, P<0.05).
Finally, Pearson correlation analysis revealed a significant relationship between production measures and grammatical comprehension scores, details of which are presented in Table 4.



Discussion
This study was conducted to investigate the comprehension of canonical and noncanonical sentences, as well as its relationship with production measures—specifically, MLU and MLS—in PSWA. In line with this aim, four research objectives were defined, and the findings are discussed accordingly.
The first objective addressed overall sentence comprehension and the comprehension of canonical and noncanonical structures in the PSWA group, with comparisons to normative performance in NBD individuals. The findings indicated that NBD participants correctly identified all sentences in the BAT grammatical comprehension test. This performance is consistent with numerous studies emphasizing intact comprehension abilities in individuals without brain injury [2, 4, 33, 34]. In contrast, data from the PSWA group showed that their overall sentence comprehension scores were significantly lower than those of the NBD participants. This significant difference aligns with cross-linguistic research in German [7],Turkish [4], Hebrew [35], and Italian [8], as well as with previous studies by Shiyani et al. [18] and Azad [19, 20] on PSWA.
Moreover, the findings revealed that comprehension of both canonical and noncanonical structures was markedly weaker in the PSWA group compared to the NBD group. Between-group comparisons of these sentence types showed that, although agrammatic individuals performed better on canonical structures than on noncanonical ones, their performance still fell short of that of their healthy counterparts. This result is in line with studies reporting that canonical word order sentences are generally easier for individuals with agrammatism to comprehend, although some degree of impairment remains in comparison to unimpaired speakers [5, 9].
On the other hand, the observed pattern of impaired comprehension even for canonical sentences is consistent with studies highlighting variability among individuals with agrammatism in comprehending different sentence types. For example, in a meta-analysis of 15 studies, Berndt et al. identified three general performance patterns: 1) Relatively preserved comprehension of both simple and complex sentences; 2) Impaired comprehension across both sentence types; and 3) Better performance on simple compared to complex sentences [36].
The second objective of this study was to conduct within-group comparisons in the PSWA group, examining comprehension of canonical versus noncanonical sentences, as well as different types of noncanonical constructions. In the first step, it was found that comprehension of noncanonical sentences was significantly weaker than that of canonical sentences. This finding aligns with the study by Caramazza and Zurif in English [3] , and with other cross-linguistic studies conducted in English [6], Italian [8], Czech [9], and Turkish [4] all of which reported that individuals with agrammatism face greater difficulties comprehending noncanonical sentences due to their increased syntactic complexity. A’meri et al. similarly reported that Persian-speaking individuals with Broca’s aphasia showed significant difficulties in comprehending syntactic structures with noncanonical word order compared to canonical ones [37]. These findings are further supported by the results of Azad [19, 20], who also reported significant differences in the comprehension of these sentence types. However, they contrast with the study by Raqibdoost, which found only mild deficits in the comprehension of various noncanonical structures among two individuals with agrammatism [17].
In the current study, comprehension of subject-topicalized sentences was found to be easier than comprehension of object-topicalized sentences. A similar pattern emerged when comparing subject-cleft to object-cleft structures. These results are in line with observations by Shiani et al. and Azad [18, 19] and also support findings from other studies indicating that sentence structures involving syntactic movement pose greater challenges for individuals with agrammatism. For example, Friedmann and Shapiro reported that Hebrew-speaking individuals with agrammatism had little difficulty comprehending sentences with canonical word order, but showed significant impairments in understanding structures that involved the movement of noun phrases from their canonical positions [35]. Similarly, in Turkish—a language with flexible word order that allows for six possible configurations—object-topicalized sentences were more difficult to comprehend than subject-topicalized ones [4, 34, 38]. Nonetheless, not all studies have reported consistent results. For instance, Jap et al. (2016) found no significant difference in the comprehension of active versus passive canonical sentences in Indonesian individuals with agrammatism [33]. Aziz et al. (2020) reported similar findings among Malay-speaking participants with agrammatism, observing no notable performance differences between active and passive constructions [39]. Despite these inconsistencies across languages, the present findings suggest that word order significantly affects sentence comprehension in PSWA. As previously noted, although Persian allows some flexibility in word order and permits object fronting, the canonical structure is SOV. Object-cleft and object-topicalized sentences, in which the object is moved from its base position to the sentence-initial position, are considered noncanonical structures. According to the DOP-H, these are derived constructions involving syntactic movement.
Furthermore, the analysis revealed that AQ had a significant effect on the comprehension of object-cleft sentences. This finding suggests that aphasia severity specifically affects the processing of structures involving syntactic movement and altered word order; individuals with lower AQ scores may experience greater processing demands when interpreting such constructions. Accordingly, the observed pattern in the PSWA group aligns with the DOP-H hypothesis: Participants exhibited the most difficulty with derived structures that involve movement, specifically object-cleft and object-topicalized sentences. These findings lend further support to the DOP-H in languages such as Dutch [40], Czech [41], Hebrew [36], and Turkish [34]. Additionally, the current results underscore the importance of syntactic complexity and the role of syntactic movement in sentence processing, with significant implications for designing targeted language rehabilitation programs in this domain.
The third objective of the study was to examine MLU and MLS in the PSWA and NBD groups. The comprehensive analysis of eight discourse tasks across both groups, based on a broader sample than previous Persian-language studies, revealed specific production-related markers consistent with earlier findings. The data from this study, in line with multiple prior investigations, indicate that in the domain of syntactic production, the language output of individuals with agrammatism is characterized by reduced sentence length and telegraphic output. Specifically, these individuals produce fewer words per utterance, which is typically reflected in a lower MLU [40, 42] Additionally, examination of their sentence structures has shown a clear preference for significantly shorter sentences [42-44]. Earlier Persian-language studies have also reported syntactic limitations and structural simplifications, manifested as reduced MLU [43-45] and MLS [41], in individuals with agrammatism when compared with speakers who possess intact language abilities [45]. Moreover, the present findings demonstrated that the severity of aphasia affects both production measures: increased aphasia severity was associated with a decrease in both MLU and MLS. This significant correlation highlights the variability in syntactic output among individuals with different levels of impairment and underscores the importance of accounting for such differences in clinical rehabilitation planning.
The fourth objective of this study was to examine the relationship between syntactic production measures—MLU and MLS and the grammatical comprehension scores in the PSWA group. The findings indicated a significant correlation between MLU and MLS and syntactic comprehension. These results are consistent with studies that have confirmed a link between production and comprehension performance in individuals with agrammatism [16]. However, they contrast with other studies that have contradicted a direct connection between language comprehension and production. For example, the study by Raghibdoost on two PSWA, the case report by Nespoulous and colleagues on a French-speaking patient [15] and research involving Greek-speaking individuals with agrammatism [14] all reported different outcomesMoreover, although the present study found a significant relationship between sentence comprehension and production indices, the significant effect of aphasia severity on these variables suggests variability in participant performance. This highlights the need for caution in interpreting the results. One important aspect of this discussion is the clinical application of the relationship between sentence comprehension and production, which has also been addressed in some previous studies. In fact, examining this relationship in individuals with agrammatism can provide valuable information for clinical decision-making. Specifically, research investigating uni-modal treatments—i.e. therapies focused solely on comprehension or production—and cross-modal generalization following comprehension- or production-based therapy has shown a one-way relationship from comprehension to production [46-48]. That is, sentence comprehension deficits do not appear to improve following production-based interventions, as there is no evidence of skill generalization from production to comprehension. However, when sentence production deficits are present, it is recommended to first assess comprehension. If comprehension is also impaired, therapy should begin with or include comprehension training, as this may increase the likelihood of generalization from comprehension to production.
The question of whether shared syntactic representations underlie agrammatic deficits in both comprehension and production remains an open area for future research. Overall, the scarcity of empirical studies and scientific resources on agrammatism in Persian limits the ability to use previous studies as a reliable baseline for comparison. Future research employing structured protocols for assessing and analyzing syntactic comprehension and production can strengthen the scientific foundation in this area and contribute to a deeper understanding of agrammatism in Persian-speaking individuals. It is essential that upcoming studies focus on developing standardized discourse assessment tools to enhance cross-study comparability and improve clinical applicability in Persian-speaking populations.

Conclusion
The findings of this study demonstrated that sentence comprehension skills in individuals with agrammatism were significantly lower than those of the NBD control group. The results revealed marked difficulty in processing complex and noncanonical sentences, particularly those involving altered word order—namely, object-cleft and object-topicalized structures—in PSWA. Accordingly, these findings support the DOP-H, which posits that sentences involving syntactic movement of noun phrases and derived word orders require more complex processing—an issue clearly reflected in the performance of the PSWA group in this study.
Additionally, the results showed a significant relationship between production measures and comprehension performance in individuals with agrammatism. Sentence comprehension, particularly for object-cleft structures, as well as the production indices of MLU and MLS, varied according to aphasia severity. These variations have direct implications for the design of rehabilitation programs. Therefore, it is recommended that thorough assessments of syntactic comprehension deficits be conducted with consideration of aphasia severity in order to enhance rehabilitation services for individuals with agrammatism. These findings may offer valuable guidance for clinicians and researchers seeking to refine assessment strategies and implement individualized, evidence-based interventions.

Ethical Considerations

Compliance with ethical guidelines
This study was approved by the Ethics Committee of the University of Rehabilitation and Social Health Sciences (Code: IR.USWR.REC.1402.059).

Funding
This article is derived from a research project supported by the Clinical Research Development Center of Rofeideh Rehabilitation Hospital, Tehran, Iran (Project No.: 4121). 

Authors' contributions
Conceptualization: Fariba Yadegari, Methodology and data analysis: All authors; Drafting, editing, and finalizing the manuscript: Fariba Yadegari and Samaneh Sazegarnezhad; Supervision: Fariba Yadegari, Rabab Teymoori, Enayatollah Bakhshi.

Conflict of interest
The authors declare that there is no conflict of interest.

Acknowledgments
The authors would like to thank the Clinical Research Development Center of Rofeideh Rehabilitation Hospital, Tehran, Iran for their valuable support and collaboration. They also extend their gratitude to the participants and student assistants for their meaningful contributions. 



 

1. Cappa SF. Neurological accounts of agrammatism. In: Bastiaanse R, Thompson CK, editors. Perspectives on agrammatism. Hove: Psychology Press; 2012. [Link]
2. Caramazza A, Zurif EB. Dissociation of algorithmic and heuristic processes in language comprehension: Evidence from aphasia. Brain and Language. 1976; 3(4):572-82. [DOI:10.1016/0093-934X(76)90048-1] [PMID]
3. Thompson G. Introducing functional grammar: 3rd ed. London: Routledge; 2013. [DOI:10.4324/9780203431474]
4. Duman TY, Altınok N, Özgirgin N, Bastiaanse R. Sentence comprehension in Turkish Broca’s aphasia: An integration problem. Aphasiology. 2011; 25(8):908-26. [DOI:10.1080/02687038.2010.550629]
5. Bastiaanse R, van Zonneveld R. Sentence production with verbs of alternating transitivity in agrammatic Broca’s aphasia. J Neurolinguistics. 2005; 18(1):57-66. [DOI:10.1016/j.jneuroling.2004.11.006]
6. Bastiaanse R, Edwards S. Word order and finiteness in Dutch and English Broca’s and Wernicke’s aphasia. Brain and Language. 2004; 89(1):91-107. [DOI:10.1016/S0093-934X(03)00306-7] [PMID]
7. Burchert F, De Bleser R. Passives in agrammatic sentence comprehension: A German study. Aphasiology. 2004; 18(1):29-45. [DOI:10.1080/02687030344000409]
8. Luzzatti C, Toraldo A, Guasti MT, Ghirardi G, Lorenzi L, Guarnaschelli C. Comprehension of reversible active and passive sentences in agrammatism. Aphasiology. 2001; 15(5):419-41. [DOI:10.1080/02687040143000005]
9. Hudousková A, Flanderková E, Mertins B, Tomšu K. [Hypotéza vymazávání stop a její platnost v češtině u pacientů s Brocovou afázií (Czech)]. Slovo a slovesnost. 2014; 75(3):163-80. [Link]
10. Miceli G, O’Connor, Menn L, Obler LK. Agrammatic aphasia: A cross-language narrative sourcebook. Amsterdam: John Benjamins Publishing Company; 1990. [Link]
11. Benedet MJ, Christiansen JA, Goodglass H. A cross-linguistic study of grammatical morphology in Spanish-and English-speaking agrammatic patients. Cortex. 1998; 34(3):309-36. [DOI:10.1016/S0010-9452(08)70758-5] [PMID]
12. Rausch P, Burchert F, De Bleser R. Comprehension of canonical and non-canonical structures within and across the verbal and nominal syntax domains in agrammatism. Brain and Language. 2007; 103(1-2):82-3. [DOI:10.1016/j.bandl.2007.07.055]
13. Miceli G, Mazzucchi A, Menn L, Goodglass H. Contrasting cases of Italian agrammatic aphasia without comprehension disorder. Brain and Language. 1983; 19(1):65-97. [DOI:10.1016/0093-934X(83)90056-1] [PMID]
14. Fyndanis V, Varlokosta S, Tsapkini K. (Morpho) syntactic comprehension in agrammatic aphasia: Evidence from Greek. Aphasiology. 2013; 27(4):398-419. [DOI:10.1080/02687038.2013.770817]
15. Nespoulous JL, Dordain M, Perron C, Ska B, Bub D, Caplan D, et al. Agrammatism in sentence production without comprehension deficits: reduced availability of syntactic structures and/or of grammatical morphemes? A case study. Brain and Language. 1988; 33(2):273-95. [PMID]
16. Meyer AM, Mack JE, Thompson CK. Tracking passive sentence comprehension in agrammatic aphasia.  Journal of Neurolinguistics. 2012; 25(1):31-43. [DOI:10.1016/j.jneuroling.2011.08.001] [PMID] 
17. Raghibdoust S. Comprehension and grammaticality judgement in Persian-speaking agrammatics [doctoral dissertation]. Ottawan: University of Ottawa; 1999. [Link]
18. Shiani A, Joghataei MT, Ashayeri H, Kamali M, Razavi MR, Yadegari F. Comprehension of complex sentences in the Persian-speaking patients with aphasia. Basic and Clinical Neuroscience. 2019; 10(3):199. [DOI:10.32598/bcn.9.10.185] [PMID] 
19. Azad O. [Syntactic comprehension in Persian-speaking agrammatics: further proof of trace deletion hypothesis (Persian)]. Journal of Language Horizons. 2021; 5(1):185-203. [DOI:10.22051/lghor.2020.32410.1346]
20. Azad O. Canonicity Effect on Sentence Processing of Persian-speaking Broca’s Patients. Basic and Clinical Neuroscience. 2022; 13(6):865. [DOI:10.32598/bcn.2021.2777.1] [PMID] 
21. Azad O. Comprehension of complex structures by persian-speaking aphasics: The role of cognitive load. Journal of Modern Rehabilitation. 2021; 15(4):227-38. [DOI:10.18502/jmr.v15i4.7743]
22. Cornell TL, Fromkin VA, Mauner G. A linguistic approach to language processing in Broca’s aphasia: A paradox resolved. Current Directions in Psychological Science. 1993; 2(2):47-52. [DOI:10.1111/1467-8721.ep10770692]
23. Fridriksson J, Morrow L. Cortical activation and language task difficulty in aphasia. Aphasiology. 2005; 19(3-5):239-50. [DOI:10.1080/02687030444000714] [PMID] 
24. Nilipour R, Pourshahbaz A, Ghoreyshi ZS. Reliability and validity of bedside version of Persian WAB (P-WAB-1). Basic and Clinical Neuroscience. 2014; 5(4):253-8. [PMID]
25. Paradis M, Libben G. The assessment of bilingual aphasia. London: Psychology Press; 2014. [DOI:10.4324/9781315802138]
26. Dipper L, Marshall J, Boyle M, Hersh D, Botting N, Cruice M. Creating a theoretical framework to underpin discourse assessment and intervention in aphasia. Brain Sciences. 2021; 11(2):183. [DOI:10.3390/brainsci11020183] [PMID] 
27. Ghayoumi Anaraki Z, Marini A, Yadegari F, Mahmoodi Bakhtiari B, Fakharian E, Rahgozar M, et al. Narrative discourse impairments in persian-speaking persons with traumatic brain injury: A pilot study. Folia Phoniatrica et Logopaedica. 2015; 66(6):273-9. [DOI:10.1159/000371443] [PMID]
28. Stark BC, Bryant L, Themistocleous C, Den Ouden D-B, Roberts AC. Best practice guidelines for reporting spoken discourse in aphasia and neurogenic communication disorders. Aphasiology. 2023; 37(5):761-84. [DOI:10.1080/02687038.2022.2039372]
29. Bastiaanse R. Production of verbs in base position by Dutch agrammatic speakers: Inflection versus finiteness. Journal of Neurolinguistics. 2008; 21(2):104-19. [DOI:10.1016/j.jneuroling.2006.10.006]
30. Menn L, Obler LK, Miceli G. Agrammatic aphasia: A cross-language narrative sourcebook. Amsterdam: John Benjamins Publishing Company; 2012. [Link]
31. Bateni M. [Describing the syntactic structure of Persian (Persian)]. Tehran: Amirkabir Publications Institute; 2002. [Link]
32. Saffran EM, Berndt RS, Schwartz MF. The quantitative analysis of agrammatic production: Procedure and data. Brain and Language. 1989; 37(3):440-79. [DOI:10.1016/0093-934X(89)90030-8] [PMID]
33. Jap BA, Martinez-Ferreiro S, Bastiaanse R. The effect of syntactic frequency on sentence comprehension in standard Indonesian Broca’s aphasia. Aphasiology. 2016; 30(11):1325-40. [DOI:10.1080/02687038.2016.1148902]
34. Maviş İ, Arslan S, Aydin Ö. Comprehension of word order in Turkish aphasia. Aphasiology. 2020; 34(8):999-1015. [DOI:10.1080/02687038.2019.1622646]
35. Friedmann N, Shapiro LP. Agrammatic comprehension of simple active sentences with moved constituents: Hebrew OSV and OVS structures. Journal of Speech, Language, and Hearing Research. 2003; 46(2):288-97. [DOI:10.1044/1092-4388(2003/023)] [PMID]
36. Berndt RS, Mitchum CC, Haendiges AN. Comprehension of reversible sentences in “agrammatism”: A meta-analysis. Cognition. 1996; 58(3):289-308. [DOI:10.1016/0010-0277(95)00682-6] [PMID]
37. A’meri H, Golfam A, Kolk H, A’shayeri H, Hagh-Shenas AM, Kord-Za’feranlou E. Inreraction of syntactic comprehension and ability of cognitive sequence processing in Persian speeking broca aphasic patients. Archives of Rehabilitation. 2008; 9(2):16-0. [Link]
38. Yağız O, Aydın B. Processing active/passive sentences in Turkish agrammatic aphasics. Procedia-Social and Behavioral Sciences. 2014; 158:254-9. [DOI:10.1016/j.sbspro.2014.12.084]
39. Aziz MAA, Hassan M, Razak RA, Garraffa M. Syntactic abilities in Malay adult speakers with aphasia: A study on passive sentences and argument structures. Aphasiology. 2020; 34(7):886-904. [DOI:10.1080/02687038.2020.1742283]
40. Vermeulen J, Bastiaanse R, Van Wageningen B. Spontaneous speech in aphasia: A correlational study. Brain and Language. 1989; 36(2):252-74. [DOI:10.1016/0093-934X(89)90064-3] [PMID]
41. Pavlína H. Sentence processing in agrammatic aphasia [MA thesis]. Moravia: Olomouc: Palacký University; 2016.
42. Rossi E, Bastiaanse R. Spontaneous speech in Italian agrammatic aphasia: A focus on verb production. Aphasiology. 2008; 22(4):347-62. [DOI:10.1080/02687030701407093]
43. Jonkers R, Bastiaanse R. How selective are selective word class deficits? Two case studies of action and object naming. Aphasiology. 1998; 12(3):245-56. [DOI:10.1080/02687039808249453]
44. Thompson CK, Lange KL, Schneider SL, Shapiro LP. Agrammatic and non-brain-damaged subjects’ verb and verb argument structure production. Aphasiology. 1997; 11(4-5):473-90. [DOI:10.1080/02687039708248485]
45. Sazegar Nejad S, Yadegari F, Teymouri R. Persian verb and sentence production protocol (PVSP) and its efficacy on Persian-speaking agrammatic non-fluent persons with aphasia. Aphasiology. 2021; 35(1):73-99. [DOI:10.1080/02687038.2019.1673305]
46. Schröder A, Burchert F, Stadie N. Training-induced improvement of noncanonical sentence production does not generalize to comprehension: Evidence for modality-specific processes. Cognitive Neuropsychology. 2015; 32(3-4):195-220. [DOI:10.1080/02643294.2014.968535] [PMID]
47. Adelt A, Hanne S, Stadie N. Treatment of sentence comprehension and production in aphasia: Is there cross-modal generalisation? Neuropsychological Rehabilitation. 2018; 28(6):937-65. [DOI:10.1080/09602011.2016.1213176] [PMID]
48. Matchin W, den Ouden DB, Basilakos A, Stark BC, Fridriksson J, Hickok G. Grammatical parallelism in aphasia: A lesion-symptom mapping study. Neurobiology of Language. 2023; 4(4):550-74. [DOI:10.1162/nol_a_00117] [PMID]