International large-scale assessments such as the Trends in International Mathematics and Science Survey (TIMSS) and the Programme of International Student Achievement (PISA) are increasingly used to affect educational policy in different countries. For example, in Germany, after the first cycle of the PISA in 2000, there was a "Pisa shock". Results revealed that the proportion of German students with low reading, mathematics, and science achievements was extraordinarily high and the enormous level of social inequality. After receiving the results, Germany shifted responsibility for education standards from states to the national level and expanded opportunities for support of disadvantaged students, including those with an immigrant background.

Little is known about the relationship between these two assessments. On the country level, the correlation between PISA and TIMSS results in 2003 was relatively high - 0.84. Carnoy et al. using the unique dataset from a national sample of Russian students that took the TIMSS test in the eighth grade and then the PISA test one year later, demonstrated that the standardized TIMSS results in mathematics are related with PISA scores at about 0.5 standard deviations when controlling for other characteristics. 

One of the possible explanations for these differences is that TIMSS aimed to measure academic content knowledge, while PISA requires applying practical knowledge to the everyday context. It was shown that PISA reading achievements could explain a considerable proportion of the variance in mathematics performance. Wu (2009) tried to explain this relationship by comparing the types of PISA and TIMSS mathematics tasks following their specifications given by their developers. Although it is shown that the complexity of the texts of practical problems affects the ability to solve them, we have not found any literature considering TIMSS and PISA tasks using standardized text analysis measures.

The model of practical tasks could be applied to analyze the texts of tasks. Practical means that a task with a lot of nonmathematical text requires activating the main executive processes of working memory to suppress irrelevant information and extract mathematics aspects to solve real-life problems. So, the use of reading literacy skills and the complexity of the text affects the ability to solve text problems as much or more as computational skills, and text understanding and reasoning skills play an essential role.

Russia is one of the countries that conduct both TIMSS and PISA. Russian students, as in many other countries, have low PISA and high TIMSS scores, and this result has remained the same for more than 20 years. However, PISA results are included as the indicators of education quality in Russia. To achieve that, the government has announced changes in national tests to achieve this quality, but it is not known whether this happened.

This study aimed to reveal differences between tasks' texts in TIMSS, PISA and Russian national high-stakes exam after the primary general (9 years) school using the frame of the practical task.

For the analysis we have collected tasks from the open courses: PISA – 79 tasks from 2009 to 2021, TIMSS – 107 tasks from 2018 to 2021, GIA-9 (high-stakes exam after 9 grade) – 270 tasks from demo variants from 2009 to 2021. To identify tasks we used three dimensions of practical tasks: everyday semantics, novelty and structure rigidity. The everyday semantic was defined using the proportion of mathematical terms, the top 20 words and the most significant collocations. The language novelty was measured by the previous three indicators, the number of words per task, the proportion of imperative mood and adjectives, and text complexity formulas: the Flash readability formula, the Flash-Kincaid index, and the Gunning Index. The language novelty was measured by the number of words per task, the proportion of imperative mood and adjectives.

In all three dimensions, PISA turned out to be a much more practice-oriented text. At the same time, TIMSS in most parameters did not significantly differ from the texts of national tests. There are much fewer mathematical concepts used in PISA tasks (15%) than in the TIMSS (40%) and national tests (35-45%). At the same time, PISA tasks contain more descriptions of everyday situations and concepts such as "speed", time categories and monetary units. At the same time, in the national tests, the terms "number", "triangle", "point", "graph", etc. are used to formulate the problem. In PISA, only 2% of words are imperative verbs, while in other tests more than 12%. Apparently, Russian students, accustomed to strict formulations and direct instructions, are lost when faced with the lack of familiar mathematical vocabulary and a large text of the task in the PISA test. According to the results of the PISA reading test, we know that the reading literacy of Russian students is quite low, and therefore they may simply not understand what is required of them, in the absence of the usual formulations "find" or "depict". Our hypothesis that the PISA math test is more practice-oriented than TIMSS, and, more importantly, the school curriculum tests, has been confirmed. Two main factors determine lower results in PISA among Russian students: the lack of familiar vocabulary and a different approach to the construction of the problem. Since Russia try to achieve high results in PISA, it is necessary to acquaint Russian students with the format of this test and prepare them to solve similar tasks. It is known that teaching mathematics at school is largely focused on "training" to solve the high-stakes tests, so it is critically important to bring them closer to practice orientation.

de Koning, B.B.; Boonen, A.J.H.; van der Schoot, M. The Consistency Effect in Word Problem Solving Is Effectively Reduced through Verbal Instruction. Contemp. Educ. Psychol. 2017, 49, 121–129. Carnoy, M., Khavenson, T., Loyalka, P., Schmidt, W. H., & Zakharov, A. (2016). Revisiting the relationship between international assessment outcomes and educational production: Evidence from a longitudinal PISA-TIMSS sample. American Educational Research Journal, 53(4), 1054-1085 Ding, H., & Homer, M. (2020). Interpreting mathematics performance in PISA: Taking account of reading performance. International Journal of Educational Research, 102 Duque de Blas G., Gómez-Veiga I., García-Madruga J. A. Arithmetic Word Problems Revisited: Cognitive Processes and Academic Performance in Secondary School //Education Sciences. – 2021. – Т. 11. – №. 4. – С. 155. Jiang, R.; Li, X.; Xu, P.; Lei, Y. Do Teachers Need to Inhibit Heuristic Bias in Mathematics Problem-Solving? Evidence from a Negative-Priming Study. Curr. Psychol. 2020. Kintsch, W.; Greeno, J.G. Understanding and Solving Word Arithmetic Problems. Psychol. Rev. 1985, 92, 109–129. Klyachko T. L. National project "Education": advantages and risks //Economic development of Russia. - 2019. - Vol. 26. - No. 6. Larina G. S., Analysis of practical problems in mathematics: a theoretical model and experience of application in the classroom // Questions of education. 2016. №3. Niemann, D., Martens, K., & Teltemann, J. (2017). PISA and its consequences: Shaping education policies through international comparisons. European Journal of Education, 52(2), 175-183. Waldow, F. (2009). What PISA did and did not do: Germany after the ‘PISA-shock’. European Educational Research Journal, 8(3), 476-483. Wu, M. (2009). A comparison of PISA and TIMSS 2003 achievement results in mathematics. Prospects, 39(1), 33.